Null hypothesis discussion thread

by Judith Curry

On the  politics of climate expertise thread, we discussed Kevin Trenberth’s statement:

Given that global warming is “unequivocal”, to quote the 2007 IPCC report, the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence. 

About two months ago, both Kevin Trenberth and I received the following invitation from Mike Hulme, editor of WIRES Climate Change:

Using Kevin Trenberth’s speech to the 2011 AMS Joint Presidential Session as the framing for these proposed essays, the journal WIREs Climate Change would like to invite you to develop a reasoned and lightly referenced argument for and against the proposition highlighted in the extract below – “the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence”.  . .  Both essays would be written ‘blind’ – i.e., without seeing the others’ essay – but the two essays would be peer-reviewed by the same three experts.  

Both Trenberth and I accepted the invitation.  Trenberth has apparently already submitted his essay, which was given a suggested submission deadline of 30 June 2011.  I probably won’t be able to start working on my essay until May 28.  So I am throwing this topic open for discussion.  I have an idea of how I want to frame my essay, but really haven’t had time yet to work on it.  I would be most interested in your thoughts and ideas on this topic.

Selected previous comments on the null hypothesis

Here are some “editors choice” comments on the topic of the null hypothesis from previous threads:

David L. Hagen | October 25, 2010 at 11:28 pm

Whatever happened to the “null hypothesis”?
Assuming nature will continue on as before unless proven otherwise?
Then again, 
it is important to recognize that an inherent difficulty of testing null hypotheses is that one cannot confirm (statistically) the hypothesis of no effect.While robustness checks (reported in the appendix), as well as p values that never approach standard levels of statistical significance, provide some confidence that the results do not depend on model specification or overly strict requirements for statistical significance, one cannot entirely dismiss the possibility of a Type II error.

Jana von Stein (2008)

TomFP | October 26, 2010 at 8:16 am |  
Finally, another null hypothesis nut! I was beginning to think it was just me. Judith the misgivings I have about your Italian flag, and the whole calculus of uncertainty, include the question David asks. It seems that to view every hypothesis solely through the prism of uncertainty allows the null hypothesis to fade from view like the Cheshire cat. 
I’ve seen plenty of evidence that climate scientists may not even properly understand (or at least share my understanding of) the term, in that they confuse it with the consensus. One see talk by climate scientists of x “becoming the new null hypothesis”, where what is meant is that x has attained the status of “the consensus view” and thus deserves to thought of as the default position. 
Elsewhere one sees the null hypothesis (on the rare occasions when it makes its unwelcome appearance) adduced as a sort of box to be ticked only at certain milestone junctures in the experimental process, rather than as an ever-present, if often irksome, companion the scientist, to be consulted at every step.
So in climate science it seems to me that the null hypothesis is poorly understood, and honoured more in the breach than in the observance. 
This underlies my earlier plea for a post on The Scientific Method.

Experimental science basically advances by demonstrating that “on the odds” theories don’t fit the evidence. Hypotheses get falsified, and by doing so their alternates gain credence.

My point is that scientists are absolutely free to choose what they are going to try and disprove. A scientist setting out can choose to try and falsify “AGW false” or falsify “AGW true”. In either case if they are successful they add weight to the alternate hypothesis and thereby have added to the sum total of our knowledge.

But this means there are as many null hypotheses as there are things scientists want to speculate about.

I should add that if a scientist is unsuccessful in knocking the null off its perch they haven’t proved anything. There are lots of reasons unrelated to what is really going on that might make it impossible to disprove something (e.g. all the things that happen in climate science – poor data, lots of confounding influences etc).

So Dr T is free to choose his null hypothesis as he sees fit, but that will determine what he has shown if he disproves it.

One of the supreme ironies of Dr T saying we should adopt “AGW true” as the null means that climate science should then be focused on disproving this. This is of course the skeptics’ position and it’s the failure of the believers to do this that leads to much of the angst.




435 responses to “Null hypothesis discussion thread

  1. Judy – It strikes me that the significance of which null hypothesis we choose resides primarily in the realm of economics, politics, and psychology than within the arena of science. Given uncertainties and inevitable arguments about the validity of measurements, I see it unlikely that either of the alternative nulls will be falsified to the satisfaction of those involved in the debates. We would therefore be faced with the following alternative conclusions, depending on our choice:

    a) A significant human influence on climate has not been proved (the standard null unfalsified).
    b) A significant human influence on climate has not been disproved (the Trenberth null unfalsified).

    It seems reasonably predictable who will favor which outcome.

    • To add to my above comment, it does seem that if the issue involves a distinction between “no human influence” and “some influence”, Trenberth has the easier task of providing convincing evidence against a complete lack of human influence. The more important issue is quantitative, and not easily subject to “either/or” arguments. For this reason, I see the forthcoming argument as somewhat artificial, and less likely to be informative (scientifically) than the contents of your previous thread regarding probabilistic assessments of climate sensitivity, where zero sensitivity is highly improbable.

      • Fred,

        I agree with most of what you said but would like to point out that the point of choosing a null hypothesis is a political one and not a scientific one. Scientists will still debate the question for many years no matter which hypothesis you choose to call “null”, but politically we must either act or not act- both choices have pricely consequences if we were to choose wrong.

        I call myself a skeptic not because I believe that humans are not significantly affecting the climate, but because I don’t believe there’s enough evidence either way. I also advocate for doing nothing to restrict the use of GHG. If I don’t believe either hypothesis than why do I advocate for acting as though the (current) null hypothesis is right? Well, largely because it is the null hypothesis, or rather, because humans have a tendency for messing up things they don’t understand by acting too quickly (we have tons of economic and political examples for this), and as a general rule it is usually safer to only make small changes and only when we’re pretty sure we’re right. This is equivalent to saying it is usually safer to act as the way we’ve always been acting (in the case of AGW I think this is especially true).

    • “a) A significant human influence on climate has not been proved (the standard null unfalsified).
      b) A significant human influence on climate has not been disproved”
      These are not hypotheses about climate. IF they were about climate they would read something like:
      a1) Humans have significantly influenced climate;
      b1) Humans have not significantly influenced climate.
      As stated a) and b) are hypotheses about climate science. They entirely beg the question whether humans significantly affect climate by diverting the argument to whether humans have significantly affected scientific knowledge. That question is in the realm of epistemology while the important question lies in the realm of ontology. When the competing hypotheses are stated with reference to climate, rather than to proof, something becomes obvious: to reject b1) it is only necessary to demonstrate the existence of one causal process which begins with humans and ends in a “significant effect” on climate; while to reject a1) it would be necessary to demonstrate the non-existence of an infinity of possible conjectured causal processes by which humans MIGHT significantly affect climate. In short, trying to prove a negative.

    • Fred,
      Trenberth’s ‘null hypothesis’ is the equivalent of the Monty Python skit determining the difference between a witch and a duck.
      IOW, it is not one.
      That you would defend it is beyond disappointing.
      Trenberth’s is ‘easier’ because it is the equivalent of ‘prove there is no God’.
      That he found himself in the position of having to choose this as his hypothesis only shows he could not actually make a meaningful point.

      • hunter 5/16/11 at 11:33 pm, Null hypothesis

        Your analogy is brilliant!

        The null hypothesis has been discussed at length on several Climate Etc. threads. I especially like the discussion in “On the consilience of evidence argument” where one finds the null hypothesis to be “natural climate variations will continue to occur”, “the indications [that the feedbacks are strongly negative] are too weak to justify serious consideration”, “climate sensitivity is low, less than 2.1ºC”, “H0: AGW does not exist”. Then we have a plurality of variable null hypotheses: “a wide null hypotheses may be modified”, “split the null hypothesis into a large set of alternative hypotheses”, “using one fixed null hypothesis for every step gives in this case totally wrong results”. All this gaggle on a thread with a garbled, redundant title. Consilience is always of evidence, making consilience of evidence redundant. What might have been intended was consilience of climate, or consilience of AGW: a bringing together evidence of AGW.

        Now Trenberth stirs the pot some more saying the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence . Perhaps he would want to modify his remarks to say that the burden of proof should now be reversed.

        HAS, above said,

        One of the supreme ironies of Dr T saying we should adopt “AGW true” as the null means that climate science should then be focused on disproving this. This is of course the skeptics’ position and it’s the failure of the believers to do this that leads to much of the angst.

        This implies that evidence in a binary hypothesis test is focused on H1, not H0, so to put some effort into the null hypothesis, the test must be reversed! So, does HAS contend that skeptics literally want to reverse the test, or would he be satisfied to say simply that true scientists would put emphasis on H0? Skepticism in the truest scientific sense of the word is good and is indeed essential to science. Michael Mann(!), interviewed on 11/26/09.

        Consider Roy Spencer, not only a scientist/skeptic, but a fallen angel. He said,

        The fact is that the ‘null hypothesis’ of global warming has never been rejected: That natural climate variability can explain everything we see in the climate system.

        In Popper’s bizarre world where definitions do not matter, no signal can develop within the noise. What these dialogs need is a Climate Etc. Glossary, and maybe gold stars for posts showing discipline. Then let’s get on with consilience of the null hypothesis.

      • Jeff,
        Thanks. Frankly I think historians of this social mania are going to wonder why Trenberth was not simply dismissed out of hand by all serious people after what he did.

      • Yeah, I see the resemblance.

        Looks like proxy work, if you ask me.

  2. Judith,

    This is one I will have to think about more before giving a full answer. But I’ll say this – debates about the null hypothesis have the potential to end up being more complex than the actual concept under discussion.

    wrt Climate Science, I am not even sure that the idea of a Null hypotehsis applies. Null hypotheses, as classically understood, apply to experiments in a laboratory in which all factors except the one of interest are controlled.

    However, we are not dealing with such a situation. We are dealing with many uncontrolled variables. So perhaps a better framework is to deal with the idea of multiple competing hypotheses . This is argued very well for ecology in Hilborn and Mangel’s “Ecological Detective”.

    • Andy, I haven’t read H&M, but what you say doesn’t seem to quite make sense. There may, in the study of any phenomenon, be one, or a multitude of competing (by which I take it you mean logically mutually exclusive) hypotheses, but each will, and must, be attended by its null. Otherwise it isn’t a properly framed hypothesis.

    • Lord Frijoles

      Andy Park said:

      “Null hypotheses, as classically understood, apply to experiments in a laboratory in which all factors except the one of interest are controlled”.

      Are you about what wrote above? In its purest definition and form, a null hypothesis applies to almost all aspects of life in general and of science in particular. The most famous example is that of people who are tried for committing a crime: “innocent until proven guilty” is the null hypothesis in such cases.

      Put simply: a null hypothesis refers to what is common knowledge and/or previously known to be “true”, so to speak.

      Or maybe I just didn’t understand your point.

  3. It would seem that “who gets to” establish the null hypothesis in a debate is the one that already has siezed suitable battlefield terrain. In my mind, I would think that with the preponderance of evidence in all areas of climate science research, it becomes clear that the proof/disproof ends up never having been whether AGW exists or does not, or if it has an effect or no effect.

    I think it’s a danger to accept that postulate framing going into it, but that is also what you might have done by agreeing to the essay. I think its universally held that AGW has a non-zero effect or contribution to the climate forcing equation…but the true issue is the magnitude of its dominance over other factors. Trenberth seems to have succeeded in shifting things a little bit, such that the ‘reverse’ of the original null is that AGW has ‘no effect’, whereas the real battle line is more of significance.

    It would seem that there is enough evidence to change the null hypothesis, but the evidence itself has also pointed to the null itself being on a different plane. I’d recommend changing the plane to one of proving relative significance/insignificance (among other null ideas) as part of the essay. This way, it is still quite palatable to accept climate science research that demonstrates the non-zero existence of various aspects of AGW while retaining a skepticism of the areas that were always of real interest.

  4. I do like this.: Whatever happened to the “null hypothesis”?
Assuming nature will continue on as before unless proven otherwise?
    Temperature of the earth has been extremely stable in a narrow range for ten thousand years. The past 130 years of run away warming, has been well inside of the range of temperature during the past ten thousand years. Nature will continue on as before. There is nothing in the data that indicates it will be otherwise.
    Ground rule: Unstable Climate Model (curve fit) projections are not data.

    • Nature has established the null hypothesis. The null hypothesis is that temperature has been stable for ten thousand years and nothing in the data indicates than anything has changed.

  5. I am viewing my assignment as refuting Trenberth’s statement, not as making an argument for a different null hypothesis

    • If I were given the assignment in the vague terms you quote, I might wonder how much latitude I had to reach the conclusion I considered most rational rather than an assigned conclusion. That might well differ from Trenberth’s position, but not necessarily without acknowledging that a case could be made in either direction. If a simple refutation is not what you would otherwise have advanced, perhaps you should check with the editors about how much room you have to maneuver. The danger of a simple pro vs con debate is that it will come to resemble a legal proceeding rather than a scientific one.

    • So you’re taking on the task of saying that there is no identifiable or measurable anthroprogenic contributor to the as-yet not fully understood ‘climate equation’ as it pertains to Global Warming? Even to a 0.0001% contribution?

      Or, are you simply going to be saying that it is still not time yet for the null to be reversed, challenging the ‘unequivocal’ wording. Upon further re-reads, the statement about Global Warming being ‘unequivocal’ doesn’t necessarily also say that anthorprogenic drivers is ‘unequivocal’– but it certainly is implied. Unfortunately, it still seems like his framing permits any non-zero anthroprogenic contribution as validating, regardless of significance.

      • Salamano,
        If Dr. Curry goes the route you suggest, she is in effect ratifying Trenberth’s inversion of the null hypothesis.
        Arguing on terms he set- especially when he set false terms for the discussion, would be a major mistake.

    • steven mosher

      The issue is that Trenberth hasnt said anything that can readily be turned into a testable null. The same of course is true for the other side which like to think that ‘natural variation’ is a null.

      Another way to go at it of course is to offer alternative versions of “trenberth’s null” What is the null? “AGW is true?” Why isnt “C02 caused more than 75% of the warming since 1850?” a suitable null? Basically, you can argue with trenberth by agreeing with him.. AGW is true is a suitable null, however we need an operational version to test.

      I guess the heart of the matter is what IS AGW exactly as a hypothesis?
      The projections of Ar4 are really testable hypothesis. Is attribution the null we test? what does it mean to say AGW is true. That humans influence the climate? how much?

      • “The same of course is true for the other side which like to think that ‘natural variation’ is a null.” Supposing myself to be of “the other side”, this is not what I think, whether I’d like to or not. I happen to suspect that ‘natural variation’ is overwhelmingly responsible for the climate’s trajectory, but that is not the same as thinking of it as a null. The true null – that there is no evidence of human influence warming the planet dangerously – may be compatible with “natural variation drives the climate”, and indeed incompatible with “human activity is dangerously warming the planet” but they are not logically identical.

      • steven mosher

        ” there is no evidence of human influence warming the planet dangerously – ”

        This statement isnt a testable null, and has nothing to do with the science of AGW.

      • Rob Starkey

        Steve writes:
        ” there is no evidence of human influence warming the planet dangerously – ”

        This statement isnt a testable null, and has nothing to do with the science of AGW.

        What you wrote is clearly, absolutely incorrect.

        You may believe that it has nothing to do with the study of the impact of additional CO2 on the environment, but for many other scientists who are studying the issue and advocating that policies be implemented to reduce CO2 emissions; it is certainly true.

      • steven mosher

        You and other miss the point entirely.

        “there is no evidence of human influence warming the planet dangerously”

        To have a null that is TESTABLE, you have to have a quantifiable statement. you know, use numbers.

        And look at the tense of that sentence. Its all wrong.

        On one view of AGW science ( which is not necessarily what the scientists say it is ) We could say the science is this.

        1. GHGs cause the planet to be warmer than it would be without them.

        This says nothing about danger. The claims about “danger” only come into play when we do the following.

        A. make assumptions about future emissions.
        B. put faith in all sorts of models, climate models, emissions models, damage models.. etc.

        I guess my point is this. Nobody has thought very hard about what exactly constitutes the actual science in AGW.

      • Steven – of course my null is not testable as stated. It was preceded by much discussion of the multiplicity of sub-hypotheses that attend the overarching hypothesis, so I didn’t think it necessary to elaborate.

        For a start, to be properly framed, it needs its alternative. Since I’m not one of those who believes testing it is worth lifting a finger for, I’m not really the one to provide it, but I note that the world is not short of those who think otherwise, and who can be relied upon to supply all manner of metrication with their catastrophic alternative hypotheses – polar bears melting, ice-caps dying out, models that project soaring temperatures – you know the sort of thing.

        So don’t worry – my null is perfectly sound in principle – it just needs some swivel-eyed warmist zealot to come along and furnish it with a properly framed alternative, and it will have all the testability you could wish for. Stop thinking of N0 as a separate construct from n1 – it’s a bad habit of warmies – and it leads you astray.

      • . Since I’m not one of those who believes testing it is worth lifting a finger for, I’m not really the one to provide it

        I think it’s fair for one to say they’re not interested in the question. Somehow, I don’t think that’s what you’re saying when you write things like “The true null [is] that there is no evidence of human influence warming the planet dangerously.” It seems more like you have an hypothesis that you simply stipulate must be accepted as fact without establishing any criteria whereby it can be falsified. And that’s valid as opinion. It’s just not science.

        Stop thinking of N0 as a separate construct from n1 – it’s a bad habit of warmies

        Assuming that what you’re referring to are H(0) and H(1) – the null and research hypotheses – no. Hypotheses have to be rigorous, and depending on how H(1) is formulated, H(0) may not be merely its logical negation.

        No hypothesis – neither a null nor a research hypothesis – is exempt from the necessity of testability. If it can’t be tested, it’s not a hypothesis.

      • steven,
        Thank you for your clarification.
        Here is the problem I see with that:
        As Trenberth and pals demonstrate, they take that claim and turn it into:
        1- a demand for crazy amounts of money.
        2- As you have documented, play fast and loose with their work.
        3- Enable the use of their work by people claiming exactly what I and others are pointing out: dangerous climate change, to demand policies which many of us believe are well documented as bad policies that do nothing they are claimed to do.
        If you can help untangle that, please do.
        Meanwhile, I see stopping what Trenberth et al are doing by making AGW basically a religious tenet as a good place to start.

      • steven,
        How can your conclusion possibly be correct?
        The whole point of the AGW movement is to save the planet from an ongoing worldwide catastrophe caused by AGW.
        How can it possibly be that this cannot be tested?

      • steven mosher

        hunter. What i’m pointing out is that to be testable a null has to specify metrics and numbers. very simple.

      • Trenberth doesn’t offer something testable either, by that standard No one seems upset very much over that until now. I was being symmetrical, in the worst case.

      • And what I have pointed out is that it is the task of the person framing the Alternative to do that.

        Until then, strictly speaking, we are dealing, as David Wojick says, with metaphor.

        Setting yourself the task of framing the Null, as a separate task from framing the Alternative is a nonsense, yet on the rare occasions when we see warmists deal with N0, that is what they invariably do.

      • John Carpenter

        “The issue is that Trenberth hasnt said anything that can readily be turned into a testable null.”

        I agree this is the way to refute his statement. His statement is too vague and undefined to explain what he envisions the new null hypothesis to be.

        “That humans influence the climate? how much?”

        Says it all… without further definition as to how much does human activity (not just human combustion CO2 emissions) contribute to the observed warming, we are going nowhere.

      • “That humans influence the climate? how much” Exactly – surely we can all agree that a null hypothesis has to be – er – null? I just don’t see how a statement which asserts a positive, causal relationship between phenomena can also be a statement of nullity. As Paul Dunmore has pointed out, this is a logical stricture, and is unaffected by evidence, however abundant or persuasive, one way or the other.

      • I just don’t see how a statement which asserts a positive, causal relationship between phenomena can also be a statement of nullity.

        What is a “statement of nullity?” It seems like you want to assert that H(0) can be just a presumption of ignorance… that “we just don’t know” can be presented as an hypothesis.

        It can’t.

        Every hypothesis – whether H(0) or H(1) must be testable. It must be falsifiable. Otherwise, it’s not an hypothesis.

        This is elementary.

    • David L. Hagen

      “my assignment as refuting Trenberth’s statement” vis
      “the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence”.

      The problem is that logically Trenberth’s statement is ill posed:

      1) Living necessitates continuous energy expenditure.
      1.1) I eat, therefore I affect my environment and change climate.
      1.2) I live, therefore I obtain and use energy, changing climate.

      2) What is happening?
      Every human affects climate.
      The critical issues are not IF but:
      2.1) By HOW MUCH?
      2.1.1.) A scientific null hypothesis:
      Nature will continue to vary as it has, including human influences.

      2.1.2 An scientific anthropogenic model:
      Increased energy use in substantially altering historic anthropogenic climate impacts.

      2.1.3 A scientific test
      Anthropogenic fossil energy use is changing climate by X with a probability of exceeding natural variation with historic anthropogenic impacts of Y.

      This must include all anthropogenic affects including changes in agriculture and economic growth with growing populations.

      2.2) SO WHAT?
      2.2.1) What are the consequences of anthropogenic effects on future natural climate variation?

      2.2.2) What are the probabilities of anthropogenic causes changing impacts of global cooling vs global warming?

      3) Agenda
      3.1) Trenberth’s unstated agenda and assertion is that:
      Anthropogenic global warming will cause catastrophic climate change with great suffering and massive economic loss.
      3.2) By a “slight of tongue”, Trenberth cloaks “anthropogenic global warming” as “climate change”.
      3.3) Trenberth then makes the illogical jump:
      Humans are affecting climate which will cause great suffering and financial loss.

      However: Natural climate change includes BOTH cooling and warming.

      4) So What

      Trenberth has the burden to prove that:
      Anthropogenic global warming cause greater suffering and loss than during nature driven global warming than global cooling periods.

      This requires accurately quantifying natural changes, anthropogenic effects, and human and economic consequences.

      5) Alternative Hypothesis
      Anthropogenic global warming is beneficial.

      5.1) Colder temperatures during the Little Ice Age caused much greater harm than warming during the Medieval warm period.

      Will we need all the anthropogenic global warming we can get to alleviate future periods of global cooling?

      5.2) Increasing CO2 and H2O increase crop production
      With a growing population, we need all possible additional food production.

      6) Transition from Fossil to Renewable Fuels
      The critical issue is that we are rapidly depleting light crude oil faster than discoveries, and faster than the transition to alternative fuels.
      The myopic focus on anthropogenic climate change is diverting attention and resources from our critical challenge of developing and transitioning to alternative transport fuels fast enough to prevent massive economic harm especially to oil importing countries. This will likely cause consequent human suffering greater than that posited by anthropogenic global warming.

      6) Conclusions:
      Anthropogenic global warming may be detectable.
      It is small compared to natural climatic variations.
      Fossil CO2 & increased H2O is more beneficial by increasing agriculture and reducing hardships during colder periods.
      We need all possible liquid fossil fuels for transport as we transition to alternative transport fuels.

      • David L. Hagen

        Increased energy use is substantially altering historic anthropogenic climate impacts.
        “Anthropogenic global warming cause greater suffering and loss during nature driven global warming than global cooling periods.”
        6) Transition from Fossil to Renewable Fuels
        7) Conclusions.

  6. I have a bit of time for Kevin ‘surely it can’t be decadal’ Trenberth’s ENSO work – only to the extent that all the ENSO proxy studies say different things about frequency and intensity but are interesting nonetheless.

    I wondering what the null hypothesis is. Carbon dioxide is not accumulating in the atmosphere? Measurement is fairly simple. People are not adding to the accumulation? Seems very unlikely given my liking for driving my wife Daisy around in her 4 litre Ford (there’s only one thing better than driving Miss Daisy). The accumulation doesn’t change the radiative flux at TOA? It seems otherwise from satellite spectral analysis. That climate isn’t chaotic – exhibiting sensitive dependence and abrupt and non-linear change? Seems hardly likely.

    So the great atmospheric experiment involves changing the composition of the atmosphere. The hypothesis is that this results in climate change. The null hypothesis is that it doesn’t result in climate change. But there are in fact questions on which science is at best uncertain. If we can’t answer a question convincingly – and I submit that we can’t predict climate convincingly – nothing is said about the hypothesis either way.

    To proceed from the point of a lack of proof that changing the atmosphere is harmful to a lack of proof of harm being a proof of no harm is not the null hypothesis at all – but the informal logical fallacy of argumentum ad ignorantum.

  7. Brandon Shollenberger

    I don’t get the idea of having one null hypothesis. Global warming is a complicated issue, so why would anyone try to reduce it to a binary issue? The only point I can find for adopting Trenberth’s null hypothesis is it gives a foundation for discussion. The people who accept it could move onto other topics for discussion, and the people who reject it would be segmented out. Of course, the people who would reject that null hypothesis are a small minority, and they rarely participate in these sort of discussions.

    If Trenberth really wants to make progress, he needs to explain the next step. If we all agree there is some influence on the climate from humans, then what?

  8. Is the climate reacting to only natural non man-made forcings with a certain amount of natural variability,


    Is it reacting to a combination of natural and man-made forcings with a certain amount of natural variability.

    Either one can be the hypothesis and the other the null.

    I really don’t think it matters.

  9. From most of the comments so far, it seems that it would be an unenviable assignment to defend the proposition that the null hypothesis that should be retained is one of zero human influence on climate. Trenberth has implied that to be the current null, but one could dispute Trenberth by claiming that the current credible null according to Lindzen, Spencer, and others is that human influence is too trivial to require action. That would more properly capture the nuances.

    • Brandon Shollenberger

      You say “the null hypothesis,” but why should there only be one?

    • Fred Moolten

      I have to agree with you on the “zero human influence on climate”.

      Earlier a lawyer brought up the “innocent until proven guilty” analogy, but that is a bit too vague and we are not talking about a courtroom here.

      “Innocence”, i.e. “zero human influence on climate” is an absurdity, so can be ruled out.

      But that is not the question here.

      We are discussing whether or nor the premise is valid that

      AGW, caused principally by human CO2 emissions, has been the primary cause of observed warming and represents a serious potential threat to humanity and our environment

      Trenberth states:

      Given that global warming is “unequivocal”, to quote the 2007 IPCC report, the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence.

      Trenberth’s logic is basically flawed:

      First of all “because IPCC said so” is an “argument from authority”, which is an error in logic.

      Secondly, the warming may have been “unequivocal”, but it is certainly not “unequivocal” that a significant part of of this observed warming was caused by AGW, let alone that AGW was the principal cause of the observed warming.

      The impact of natural climate forcing and variability is largely unknown. There have been several solar studies showing that a significant portion of the 20th century warming can be attributed to the unusually high level of 20th century solar activity (particularly in the first half of the century). Other studies show good correlation with ocean currents, which may or may not be influenced by solar variations. The role of clouds and how they may be impacted by the sun is also largely unknown.

      The current lack of warming, despite CO2 levels reaching record levels, has been loosely attributed to natural variability but is basically unexplained.

      In addition statistical studies have shown that the temperature record with its 30-year cycles of warming and slight cooling is more of a random walk and that there is no robust statistical correlation between atmospheric CO2 and temperature. Where there is no robust statistical correlation, the case for causation is extremely weak.

      And finally, there are no empirical data based on actual physical observations or reproducible experimentation, which corroborate the dAGW hypothesis.

      In summary:

      To postulate a null hypothesis of “innocence”, which states humans have had absolutely no influence on our planet’s climate is silly as very few people would argue for this case.

      To postulate a null hypothesis, of “guilt” which states that post Industrial Revolution man has had an overriding influence on our planet’s climate, which could represent a serious potential threat (the “dangerous AGW” hypothesis) is equally absurd, since this hypothesis is still uncorroborated, i.e. it has not yet been supported by empirical data based on actual physical observations or reproducible experimentation, nor has it successfully withstood falsification attempts.

      I’d say Judith should be able to put together a very convincing argument against Trenberth’s “null hypothesis” proposal.


  10. Bill Hunter

    “Given that global warming is “unequivocal”, to quote the 2007 IPCC report, the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence.”

    Seems to me that as a matter of fact the hypothesis has always been as wished by Trenberth and the only way for humans to have zero influence on their environment is for humans to be extinct.

    I see others have inserted the term “significant” into the equation. However, what is the definition of “significant”. That is a very relative term. Is 3 degrees going to make people significantly uncomfortable? Is 1/3rd of one degree?

    The proposal seems to have everything to do with what governments should allow people to do. If that’s the case, its really not a science question at all.

    CPAs are prohibited from having investments, direct or indirect, in the subject of their opinions. That standard is obviously impossible when the entity is the entire planet. But the special role of science in the policy arena is to provide supported facts and allow the democratic system to determine the significance of those facts.

    Your project here to better describe uncertainties are pertinent facts as is data that shows the weakness of a proxy.

    When one has to worry about such facts it really is only evidence of the weakness of the sum total of facts.

    • My view of ‘significance’ here is that of being an actual contributor, not that the effects of the contribution (or net magnitude of any/all contributors) itself is significant.

      Consider the Coriolis Force … It is ‘significant’ when permitted to act across large distances, but is ‘insignificant’ when factored into an equation that governs a sea-breeze (even though it’s clear from Rossby numbers that a ‘non-zero’ number can still arise for both situations).

  11. If you want to save the world perhaps the first step should be showing it needs saving. I can’t take his argument seriously because it is meaningless.

    • The Trenberth quote begins at this point/assumption. If everything goes back to first principles we won’t get anywhere and so there is no point discussing the issue. This is a forum to look at the “what ifs” in science proceedure and that is why, in my opinion, this is an interesting topic.

      It isn’t meaningless, you are simply disagreeing with the assumption, it therefore becomes meaningful IF, for example, what the IPCC AR5 report says is correct, THEN we can start thinking about the burden of truth. It is menaingful as a hypothetical question even if the IPCC reports are wrong. It could also become true by the standards of most people in this forum if new and powerful evidence came to light. There is no contradiction so it is meaningful (but, it MIGHT be based on a false assumption).

      • Paul, his null hypothesis is to show man doesn’t change the climate. I don’t think you can find one single person out there that doesn’t believe man changes the climate in at least one way or another be it GHGs or UHIs or irrigation or dam construction. We all know we change the climate. If he wanted to claim a null hypothesis that said show it wouldn’t cause catastrophic or at least dangerous change that would be a different issue. The problem with that is he needs to make a strong case before he can claim the high ground in such a manner. The models that show this may be the case haven’t shown themselves to be very accurate up to this point. The troposphere doesn’t exhibit a hot spot, the stratosphere isn’t cooling, the oceans are not accumulating heat, the warming has been 40% of that expected, and the models are inching close to falsification. Where does he gain his claimed high ground from? Imagination doesn’t count. The only response his current argument deserves is a : yeah, so what?

      • Steven,
        thanks for the response. I understand Trenberth’s point and I understand your point your first, which my comment is a response to. Your second point “Where does he gain his claimed high ground from?” would probably receive the response (from Trenberth at least) 97% of climate scientists. A bit unfair, I suppose, but that is what gives his comment about being “unequivocal” traction. The, admittedly, vague notion of “the onus of proof” or “the balance of probabilities” often rests on a much lower degree of expert agreement than 97% (BUT, of course, we know this expertise has been wrong in the past). Your point “yeah, so what?”, might be one way of responding but I think the burden of proof could be an interesting way of thinking about what needs to be done (or not) to call governments to action.

      • Yawn. 97% of climate scientists agree on what? Be exact and show your reference.

  12. John Kannarr

    While I think there are ample arguments against the proposition that the human influence on climate is of any significant magnitude, I don’t think it is likely that there is sufficient data at this time to claim to prove that there is no miniscule such influence. I suspect that any conclusion that there is any even miniscule influence (even what might actually be a beneficial influence) will still be turned into the usual trumpeted demands for stifling human freedom, for reducing man’s access to economical energy, and for imposing the green dictatorship for the foreseeable future.

    I have always seen the whole global warming debate as a series of issues, including, at the least:
    1) Is the planet warming?
    2) is any of that warming caused by human activities?
    3) Is that warming in any way a problem?
    4) Is there any reasonable action that can or should be taken with regard to any human-caused component of such warming, that will not be worse for humans that the warming itself?

    I think that debate of any null hypothesis less rigorous than the claim that “any human influence on climate is more catastrophic than any possible alternative action man might take” will ultimately be used by the CAGW forces as if that catastrophic version were indeed the one that had been debated and not falsified.

  13. Climate science seems to involve observations which are affected by many uncontrolled variables. In this context is it at all meaningful to argue over a null hypothesis? In medicine, biostatisticians came to the conclusion decades ago that the null hypothesis was of limited utility; what was important was measuring the magnitude of the effect and placing statistical confidence bounds on that. From multiple similar studies, trends could be assessed objectively and a better estimate of the magnitude of the effect (and uncertainty thereof) could be judged.

    • Rose Dlhopolsky


      I always felt that the metaphor of the earth as a patient is a useful one. You raise interesting points.

  14. Both essays would be written ‘blind’ – i.e., without seeing the others’ essay – but the two essays would be peer-reviewed by the same three experts.

    It’s probably worth pointing out that Dr. Trenberth won’t necessarily be “blind” to what’s written here. That isn’t necessarily a bad thing, because the challenge will be to mount a persuasive argument rather than spring a surprise on the other participant.

    • John Carpenter

      “Trenberth has apparently already submitted his essay…”

      If this is true, it don’t matter at this point.

      • Indeed, I’m sure Trenberth intended for Curry to be able to benefit from a discussion here. Or could be submit a revised version later? I guess he could. At which point it all becomes rather murky. But I’m glad for Judy’s openness and a sharp focus on this one key statement from the man, denier that he considers me.

      • Personally, I am less interested in “winning” this little debate than I am in putting forth a thoughtful response. As in the preparation of my congressional testimony and the response, my arguments definitely benefited from the exchanges at Climate Etc.

      • It seems you have already hit on the right response. That if the null is AGW true, then more effort should be spent to falsify. In order to falsify, some standard of falsification has to be made. If AGW is less than X percent of natural warming…. It is impossible to falsify “any human climate impact”, and a hypothesis should be falsifiable or it is just a paradox.

    • Well that would be interesting if Trenberth were to be influenced by anything written here :)

  15. The Trenberth statement relies on the authority of the IPCC. Placing aside, for the moment, that this is a logical fallacy known as the appeal to authority – the statement lacks credibility because it is a highly criticized authority subject to the whims of political influence.

    Perhaps there may be a reason to change the null hypothesis, but an appeal to the IPCC – especially after Climategate – is not a winning strategy.

    If there is a change to the null hypothesis, I would think it would have to do with the potential for danger or catastrophe.

    • I agree with this. The claim that the null should be AGW stands upon an Appeal to Authority, not verifiable empirical data of any kind. As such it is very much an argument about how many angels can dance on the head of pin.

      The hockey stick is broken! Of course the climate is warming since the last age and will continue to do so until the tipping point where we begin to slide into the next age.

      The real null hypothesis that involves AGW simply must include a level of significance in order to be testable. In other words, warming that includes AGW will be (for example) 0.3 degrees per decade.

      To simply say that ANY WARMING is evidence of AGW is again using an Appeal to Authority without any empirical data to prove it. It is totally untestable from the standpoint of AGW.

  16. Based upon the way that the scientific community treats climate change, they have already determined (to their financial satisfaction) beyond a reasonable doubt that humans are responsible for climate change and are milking it for all it is worth. Where other than the blogosphere does the scientific community treat climate change as not man made?

    Playing semantical games with the argument makes for good theater and a way for intellectuals to flex their egos but in the real world the scientific community has already reached a consensus which is beneficial to their interest and they are riding it to the bitter end. Reality, truth and the rest of humanity be damned

  17. Trenberth was referring to extreme events specifically. There seems little incontrovertible evidence that these have changed at all in frequency or intensity.

    In January I sent a photograph of a flood gauge to a site claiming catastrophic flooding in Central Queensland. ‘Well no, we might have crocodiles swimming in the streets but really this is the 4th largest flood in 100 years. No big deal.’ I was edited beyond conscionable bounds by a fanatic climate millennialist (not a hydrologist) thousands of kilometres away.

    The flooding was predictable from the 1980’s – as a result of analysis showing multi-decadal flood and drought dominated regimes. Here is the wettest day data for Australia – – it demonstrates decadal changes much better than recent extremes.

    It is far from proven that increases in greenhouse gases have had more than a marginal influence on recent global temperature. The satellite record tends to suggest otherwise. Or that any of the global ocean indices (and therefore hydrology) are obviously outside the bounds of natural variability even given the limited record. Even the surface temperature seems not to be unique in the Holocene.

    This is a long way of saying that the Hypothesis is simply the starting point of the scientific method. We have thesis, analysis and synthesis. The thesis has no scientific significance without analysis – which is the accumulation of data of the senses to confirm or falsify a hypothesis. We need to discount models – as simply noumenological products of the human mind and with other (technical) problems. The thesis can only be confirmed or falsified by data – and I submit that we have done neither.

    There are as well theoretical problems with the concept of global warming -as opposed to climate change. In a dynamically complex system with unknown control variables and multiple and little understood feedbacks – it seems to me to be by no means certain that climate will evolve steadily into a warmer state. Carbon dioxide may in fact have very little effect as a control variable in pushing climate to shift state – a small change lost amidst multiple other factors of which clouds, ice, UV and orbital irregularities are the obvious candidates.

    So we understand certain basic physics – but by no means possess the wit to predict the outcome of the great atmospheric experiment. It is uncertainty that prevails in not either confirming or falsifying the hypothesis of climate change.

    • Below is a comment I just posted on WUWT in response to the tread “Interannual Terrestrial Oscillations” That basically espouses my view of how the NULL hypothesis is driven and can therefor be predicted or forecast. That would leave the un-forecast changes available for explanation by the AGW hypothesis, if there appears to be any long term changes not accounted for by the cyclic patterns predicted by the planetary and lunar/solar cyclic effects.

      Unique to the field of AGW I have provided testable daily forecast results for a six year period, after observation and ongoing study of the original data sorting program, I am going to make revisions to upgrade the software to compensate for problems found in the original method I used to generate these maps. I think this is the way science makes progress, hypothesis, test, revise and retest, based on observable read time data fit of the hypothesis.

      What Trenberth is suggesting is backwards to real scientific method as practiced for a very long time.

      Richard Holle says:
      May 16, 2011 at 12:04 pm

      From my years of electronics communications work modulating and then demodulating signals that have been transmitted over a radio frequency carrier method, I can see ways the overall long term signals in the weather and climate data could be understood.

      If the longest wave pattern indicative of a composite wave pattern is used as the key for the demodulation of the packets of individual periods of oscillation patterns, then the over lapping of several periods of these patterns should give an indication of the reproducibility of the original signals.

      The inner planets and the outer planets have their own separate periods of harmonic oscillation as a result of billions of years of gravitational, tidal and electromagnetic interactions that are in a resonate pattern, the inner planets have a beat frequency of about 6558 day period with the outer planets settled into a period of somewhere between 172 and 179 years.

      Most of the cyclic patterns seen in the above graphs are a result of the shorter term periods of interactions of both the inner planets and the outer planets beating together and generating these short term affects in the global circulation patterns.
      As RC Saumarez says:
      May 16, 2011 at 10:14 am
      “We only have one climate record and therefore we cannot use an ensemble approach without segmentation of the data and thus limit the minimum observable frequency.”

      To solve this problem I looked at three patterns of the 6558 day period, overlaid them at the daily weather data level, and plotted the resultant combined signal for Precipitation, and temperature patterns for the USA, extended that cyclic interpenetration for a six year period, and plotted out maps to show the repeating reoccurring patterns in the global circulation, as a ( 6 year long stretch, we are now ~40 months into the posted 6 years long) forecast for part of the current repeat of the 6558 day long cycle.

      The precipitation patterns repeat well enough to show the locations of the repeat of the 1974 tornado out break we had this spring, and will also show the repeat of the severe flooding in Australia. The problem with this forecast technique is the outer planet ~172/179 year pattern is out of phase with the inner planet harmonics so it needs to be compensated for as well.

      The best method I have come up with so far would be to plot the data excursions from the expected past 6558 day patterns, by looking at the 5 to 10 day long disruptions caused by the current 6558 day period heliocentric (or Synod) conjunctions, as the corrective algorithms to adjust the inner planet patterns for the effects of the outer planet modulations.

      The Earth has a synod conjunction with each of the four major gas planets with a little over a year between impulses, (~+2 days for Neptune, ~+4 days for Uranus, ~+13 days for Saturn, ~+34 days for Jupiter) these past almost annual periods could be segmented and applied into the composite forecast as an adjustment algorithm to further enhance the pattern match, to increase the accuracy (~85%) of the inner planet 6558 day pattern to closer to 95% with the addition of the compensation for the outer planet interactive influences.

      I am in the process of rewriting the map making software to fine tune the resultant process to get a better repeatability of the temperatures as well. With the next posting of updated several year long series of forecast maps I will be including the areas of Alaska, Canada, and Australia (with enough hind cast [+6 months] to show how well it did with the flooding there). I will be adding a fourth cycle going back another 6558 day period to use the additional analog year starting in ~1938 for 2011.

      If any of the readers here have helpful suggestions on how this could be better accomplished, I would like to hear from them.

      • I think we can proceed only with great difficulty in system with multiple feedbacks and resultant non-linearity. What you are talking about are control variables. Especially solar UV – which seems to have a very significant effect on some regional indices and is modulated by the solar magneto.

        However, Earth systems vary abruptly as tremendous energies cascade through sub-systems – winds, oceans, waves, clouds, ice, vegetation and dust. The outcome is not predictable from the actions of control variables but would need to rely on a deterministic analysis of controls and feedbacks.

        I am not sure how that can done as physically plausible models themselves are non-linear but very different non-linear systems to that of climate. The actual mathematical tools available for analysis of non-linear systems are very limited. We can note the increase in autocorrelation before a tipping point (synchonisation) or noisy bifurcation (dragon-kings) – but that I think is about it.

        ‘The great revolution of nonlinear dynamics over recent decades has provided a wealth of information about the bifurcations that can destabilise a slowly evolving system like the Earth’s climate. These bifurcations are defined as points during the slow variation of a ‘control’ parameter at which a qualitative topological change of behaviour is observed in the multi-dimensional phase space of the system. The Earth’s climate is what dynamicists would call a dissipative system, and for this the bifurcations that can be typically encountered under the variation of a single control parameter are classified into three types, safe, explosive and dangerous (Thompson et al., 1994; Thompson and Stewart, 2002). The safe bifurcations, such as the supercritical Hopf bifurcation, exhibit a continuous supercritical growth of a new attractor path with no fast jump or enlargement of the attracting set. They are determinate with a single outcome even in the presence of small noise, and generate no hysteresis with the path retraced on reversal of the control sweep. The explosive bifurcations are less common phenomena lying intermediate between the safe and dangerous types: we simply note here that, like the safe bifurcations, they do not generate any hysteresis. The dangerous bifurcations are typified by the simple fold (saddle-node bifurcation) at which a stable path increasing with a control parameter becomes unstable as it curves back towards lower values of the control, and by the subcritical bifurcations. They exhibit the sudden disappearance of the current attractor, with a consequential sudden jump to a new attractor (of any type). They can be indeterminate in outcome, depending on the topology of the phase space, and they always generate hysteresis with the original path not reinstated on control reversal.

        Any of these three bifurcation types could in principle underlie a climate tipping point. But it is the dangerous bifurcations that will be of major concern, giving as they do a sudden jump to a different steady state with hysteresis, so that the original steady state will not be re-instated even if
        the controlling cause is itself reversed. So any future climatic tipping to a warmer steady state may be irreversible: a subsequent reduction in CO2 concentration will not (immediately, or perhaps ever) restore the system to its pre-tipping condition.’

        J. M. T. THOMPSON & J. SIEBER, Climate tipping as a noisy bifurcation: a predictive technique –

        There is no theoretical necessity for ‘tipping’ to be to a warmer state – but at rate are not predictable from the orbits of planets.

      • Thanks for the response, I found the linked paper by;
        J. M. T. THOMPSON & J. SIEBER, Climate tipping as a noisy bifurcation: a predictive technique –
        to be very helpful in thinking about the probability of the continuing ongoing entrapment of the attractors due to their long term stable resonance periods are a result of past influences that have set the outer bounds on the range of perturbations set by the interactions of the internal solar system dynamics sans the sudden injection of galactic scale shifts in magnetic strength or orientation in the greater galactic magnetic field topography.

        “”There is no theoretical necessity for ‘tipping’ to be to a warmer state – but at rate are not predictable from the orbits of planets.””

        When/if the total interactions between the harmonic resonance patterns established being considered, are not just tidal, and gravitational, but having electromagnetic parameters as well, and given consideration by some researchers of the possibility of a good hypothesis regarding spin orbit coupling, as an example of views that there may be galactic drivers of many natures that penetrate into the heliopause giving rise to currently unknown resultant driven reactions, that in the end will need to considered if this method is going to predict interactions past a couple hundred years at best.
        Thanks Richard Holle

  18. Where is the logic of the proposition? See …

    Given that global cooling is “unequivocal”, to quote any magic report, the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence.

    Noah says: Given that global raining is “unequivocal”, and the ark is sailing around the world, the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence.


    Or is he implying AGW is unequivocal, without saying it? In this case, he is basically saying: Given that I am right, the burden of proof is in your side. Looks childish, but that’s no news.

  19. Mike Keller

    Looks to me like the Trenberth represents the last refuge of corrupt science: “In the absence of our ability to actually prove our theory, we hereby deem it to be true. Therefore those who disagree must prove we are wrong”

    Why one would even seriously respond to Trenberths’ obviously devious and dishonest gambit is beyond me.

  20. Alex Heyworth

    Trenberth’s statement involves the most blatant switching of the subject under discussion. By his logic, global warming was unequivocal during the Paleocene Eocene Thermal Maximum, so the null hypothesis should be that this was caused by human influence. The fact of global warming and the causes of it are entirely unrelated subjects, so far as establishing the correct null hypothesis is concerned.

  21. The equivalent of the null hypothesis is that increasing CO2 has no effect on climate. Now, even Lindzen and Spencer would refute that saying doubling it adds about a degree. Therefore the current null is something that no respectable scientist believes is true and this makes it worthless as a null, being already unanimously thought wrong on the grounds of basic science.
    An alternative null would have to be a credible view held by at least the extreme scientists, such as there is no net positive feedback to increasing CO2. This cannot be proved wrong, since we are not dealing with pure mathematics, but likelihood can be given probabilities based on data as is the current practice in IPCC reports.

  22. Trenberth’s statement –
    “the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence”. was made to put those who oppose his views on the defensive.

    It espouses a ridiculous position – namely that the null hypothesis is presently that there is NO human influence whatever on climate. And in general, that fails as the null simply because few, if any of those who oppose him believe that to be true.

    Yes, there are a few outliers who claim NO human influence – but how many? One in New York and 3 in Tennessee, maybe?

    No – as I see it, the true “null hypothesis” is that there is no proof that the human influence is/has been harmful in terms of climate. Up to this point, the harm that has been projected has been a subjective judgment, generally made by those who lack the qualifications to make that judgment. And yes – I include most climate scientists in that company. How many climate scientists have either training or experience in the political arena – or in, for example, Public Administration – or in the practicalities of engineering solutions to real-life problems – or in risk management. How many are, so to speak, “fluent” in history or even in meteorology?

    In fact, the harm that has been presented has been, in general, nothing more than the extension of a particular fringe political philosophy that has used perceived harm and catastrophe as a weapon and a way to gain power for at least the last 60+ years. Aaron Wildavsky’s book “But is it True” is an investigation of the earlier efforts at projecting “harm” – and he found little evidence of truth in the various scares starting in the late 1950’s and ending in the early 1990”s. The ONLY one he found any reason to support was the Ozone Hole. Of Global Warming, his comment at one time was that it was the Mother of all environmental scares. But he found no real supporting evidence for it at that time – in spite of Jim Hansen.

    More than that, 20+ years of assorted CAGW related catastrophes have been projected – and have universally failed to materialize. Floods, hurricanes, drought – all have been projected to be worse than ever in the past – and have not been so. Insect infestations, glacier melt, Arctic ice loss – have all so far been squib rounds fired by the Gang that can’t shoot straight. Temperatures and sea level rise that are experiencing accelerating rates of increase have, in fact, done no such thing but rather have flattened.

    The ONLY supporting evidence for the entire structure of CAGW is that the CO2 level continues to rise (as expected) and that temperature has increased by some relatively small (and generally beneficial) amount over the last 150 years. And even the purported slight correlation between those two has failed for the last ten or more years.

    Has climate change happened? Of course. We no longer live in the icebox that was the LIA. And we should be thankful for that. And hope it doesn’t return.

    Do humans have an effect on climate – or at least temperature? Undoubtedly. Land use changes, UHI, aerosols, black carbon and yes, even CO2 emissions are bound to have some effect. I’ve seen the effect of black carbon and dust on glaciers – and it explains why some glaciers have melted. But it does not explain ALL of them. Nor has glacier melt been tied directly to GW except in the minds of the catastophists. To carbon output from coal plants? Yes – and that’s correctable. . To carbon output from forest fires? Yes. To GMAT? No.

    Has human influence been “harmful”? Define your terms – humans have done a great deal of damage to the planet – mines, erosion, cities(?), hundreds of thousands of miles of asphalt and concrete, forest fires, even dams could be considered “damage” if one were so inclined.

    But none of those have been successfully related directly to “climate” – in spite of a massive effort to make it so. An effort even larger and more expensive, in fact, than the Manhattan Project.

    So we’re back to the only “null hypothesis” that makes sense to me – Has or will the human influence on the climate done “harm” in terms of catastrophic events inimical to human life and welfare?

    It’s claimed that we’ve contributed to the rise in temperature for the last 50 or 60 years due to CO2 emissions. But the temperature rose prior to that – without those CO2 emissions. I’ve heard the rationale that the causes are different – but there is no proof of that – only speculation and computer models. And the studies that have been touted to “prove” that this warming is different from all past warmings have failed to convince, in part because they’re based on the models.

    At the same time there’s been little or no interest in even the possibility that warming may be beneficial – in spite of the experience of the last 150 years, in spite of the history that teaches that humans deal well with whatever changes happen – and that human society has always flourished in warmer times.

    Nor am I likely to trust computer models that are conceived, written and operated by those whose agenda and/or self-interest is tied to a particular interpretation of the output of those models. More – having worked with models, I know better than to believe that the model outputs are data – or that they’re definitive – or even objective.

    I’ve watched the conversations on Climate, Etc and other places – and what I’ve seen is that the basis for belief in CAGW is not generally data or fact, but a blind belief in a science that few understand, a misplaced faith in models and studies based on models and – fear. For my part, the data does not impress me as “catastrophic”, models “can be” useful but are not to be taken too seriously, studies based on models are simply speculation, and fear is not a way of life that I’ll accept.

    Wrt models – for those who haven’t seen it, I’d recommend this link as being of interest. Hulme’s take is a little more realistic than that of most conversations on the subject –

  23. The subject is too vague.

    If by human influence and global warming they mean the 2007 IPCC report then that is already out of date and wrong. If they mean people cause heat then he is correct.

    It’s not good for climate science either to make such a broad claim.

    • Besides his idea already seems to be the null hypothesis in schools and with most newspapers and journalist.

      His proposal has the same flaws as saying the debate is over.

      On some other points, Dr T at times has some almost Orwellian thoughts. He really dislikes the internet which promotes broader and open discussion. That seems to be certainly an area that contrasts with Dr Curry’s. He also wants the media to play a bigger role? Well the media publishes to sell newspapers and we have freedom of the press so what is he implying?


    Anthropogenic global warming claims that human emission of carbon causes global warming.

    Let us look at the global mean temperature and human carbon emission data.

    According to the global mean temperature data, in the last century, the globe experienced two global warming phases. The latest one was from 1970 to 2000, and the previous one was from 1910 to 1940 as shown in the following graph.

    As shown in the above graph, both periods had a global warming rate of about 0.15 deg C per decade.

    According to the human carbon emission data, in the second warming period, the average carbon emission was about 5 G-ton per year compared to about 1 G-ton per year for the first period as shown in the following graph.

    From the above results, human emission of carbon has no effect on the global warming rate. AGW is not supported by the data.

    There is no evidence for AGW.

  25. “The null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence.”

    This is not a scientific statement, it is a political one. I will leave to the scientists and academics the proper formulation of the null hypothesis, or hypotheses, with respect to climate science. But Trenberth’s statement was not about scientific analysis, it was about policy. When he says that AGW should be the new null hypothesis, he means that unless skeptics can “prove” the contrary, governments should formulate policy based on AGW, if not CAGW.

    It is just another attempt at framing the argument, based on the common perception that “He who frames the argument has already won.”

    In reality though, the “burden of proof” is not some ethereal, metaphysical concept. It is a real, practical term of art, that has real consequences in the world of law, politics and policy.

    In the law, the burden of proof is established by statute or common law. It has three elements. First, is the “burden of production.” That establishes which party has the obligation to first present evidence that a given fact/issue should be determined in that party’s favor. If the party with the burden of production does not have sufficient evidence, the other party need present no evidence at all, winning by default.

    Second is the “burden of persuasion.” This determines who has the ultimate burden of convincing the trier of fact on a particular issue. In criminal cases it is the state, in most civil cases, it is the plaintiff, but that can vary. Where the party with this burden of proof fails to offer sufficient evidence to convince the judge or jury, he loses. This is the sense in which Trenberth uses the term.

    Third, the “burden of proof” also refers to the degree of proof the party with the burden of persuasion must present. In other words, the state must prove guilt beyond a reasonable doubt in a criminal case, while a plaintiff must only prove his civil case by a “preponderance of the evidence,” meaning that his position is merely more likely than not true.

    But the whole concept is irrelevant unless you understand whom it is that the party with the burden has to convince. In litigation, the answer is obvious, you must convince the judge or jury depending on the type of trial. But in the climate debate, who is the jury? Who is the judge? Answering that question correctly moots the attempt by Trenberth to frame the debate.

    In the academic world, the realm of the IPCC, and the peer reviewed literature, the jury consists of the consensus scientists who control the university science departments, the IPCC and most of the peer reviewed journals. To this jury, Trenberth’s statement simply reflects the current reality. The consensus is the default position, the “null hypothesis,” in the academic world, the IPCC and the peer reviewed journals. This is not to say that it should be, only that that is the way it currently is.

    But again, Trenberth’s comment was directed to the world of policy. And there, his statement is more problematic. Among European governments, Canada and Australia, the null hypothesis is identical to that of the academic world, the IPCC etc. The burden of proof on changing policy there lies with skeptics.

    But the big question, the arena toward which I believe Trenberth’s comment was really directed, was the U.S. And in the U.S., the jury on the policy question is the electorate.

    Until recently, the president, majorities in both houses of Congress, and the Supreme Court were all on board with cap and trade , carbon taxes, and decarbonization in general. But there has been a sea change politically. As it stands now, convincing the politicians (which had largely been accomplished) will not be enough. The electorate is now paying attention, and will cast its votes in part on where particular politicians stand on the implementation of decarbonization policies.

    Here Trenberth’s plaintive claim is falling on deaf ears. It matters not the slightest whether every climate scientist, every blogger, and every progressive politician in the world agrees that CAGW should be the null hypothesis. In real world of U.S. politics, the political null hypothesis is that cap and trade, carbon taxes and decarbonization are not justified by the current state of the science. The burden of proof is on the consensus, to justify the policies they seek.

    So as to whether “the null hypothesis should now be reversed,” the answer is that in Trenberth’s world it already has been (whether it should be or not); in the real world, it has not and is not likely to be.

    • Gary, thanks for your insightful comments

    • Gary,

      Well done. The third aspect of your discussion of burden of proof is one which deserves a great deal more attention. While a voter can use any standard or no standard at all in evaluating the evidence, responsible voters should be asked to consider what kind of standard they ought to use. In criminal cases, we all understand that a jury is instructed to use “beyond a reasonable doubt”. That’s because the law recognizes society’s moral judgment that citizens should not be deprived of their life, liberty or property unless the evidence of their malfeasance is very compelling. Better that ten guilty men go free than to imprison one innocent man. Letting the guilty go free can prove harmful to society. But it’s a risk we choose to take because of our recognition of the importance of basic human rights. [A precautionary principle notion would be that locking up potential miscreants is safer than allowing the possibility that they will do serious harm to others. After all, look at all the bad stuff we think they’ve done. Even if we can’t prove this particular crime beyond a reasonable doubt, we know they’ve probably committed others. We even have proof they did, but we can’t use it in court.]

      This societal moral imperative should be instructive in considering policy options demanded by CAGW alarmists. What standard of proof should be required before enacting policies which will inevitably lead to deprivations of life, liberty and property of millions, if not billions of people? How compelling must the evidence be? Especially when these people haven’t been accused of any crime.

    • Anonymous Climate Guy

      It would be interesting indeed to know who the audience is… These “Three reviewers” … Would they happen to be Gavin Schmidt, Eric Steig, and James Hansen?

      • Anonymous Climate Guy,

        I would love to see the reviews, even if the identity of the reviewers were kept confidential.

      • Gary –
        If we see the reviews, we’ll likely have a good idea who the reviewers are.

  26. maksimovich

    Given that global warming is “unequivocal”, to quote the 2007 IPCC report, the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence.

    Firstly the statements presented by the IPCC were inconclusive for the arguments presented by Ghil 2001 (Hilbert problems for the 21st century)

    We have cast a bird’s-eye view on selected problems among the ten formulated in the Abstract. This view has clearly emphasized strong interconnections between them, both pairwise and all around. Unified methods of solution are provided by dynamical systems theory. Each problem requires the use of a model hierarchy for its solution.
    To conclude, we briefly address Problem 9. More precisely, we ask whether the impact of human activities on the climate is observable and identifiable in the instrumental records of the last century-and-a-half and in recent paleo climate records? The answer to this question depends on the null hypothesis against which such an impact is tested.
    The current approach that is generally pursued assumes essentially that past climate variability is indistinguishable from a stochastic red-noise process (Hasselmann, 1976),whose only regularities are those of periodic external forcing (Mitchell, 1976). Given such a null hypothesis, the official consensus of IPCC (1995) tilts towards a global warming effect of recent trace-gas emissions, which exceeds the cooling effect of anthropogenic aerosol emissions. Atmospheric and coupled GCM simulations of the tracegas warming and aerosol cooling buttress this IPCC consensus.

    The GCM simulations used so far do not, however, exhibit the observed interdecadal regularities described at the end of Sect. 3.3. They might, therewith, miss some important physical mechanisms of climate variability and are, therefore, not entirely conclusive.

    As northern hemisphere temperatures were falling in the 1960s and early 1970s, the aerosol effect was the one that caused the greatest concern. As shown in Sect. 2.2, this concern was bolstered by the possibility of a huge, highly nonlinear temperature drop if the climate system reached the upper-left bifurcation point of Fig. 1. The global temperature increase through the 1990s is certainly rather unusual in terms of the instrumental record of the last 150 years or so. It does not correspond, however, to
    a rapidly accelerating increase in greenhouse-gas emissions or a substantial drop in aerosol emissions. How statistically significant is, therefore, this temperature rise, if the null hypothesis is not a random coincidence of small, stochastic excursions of global temperatures with all, or nearly all, the same sign?

    The presence of internally arising regularities in the climate system with periods of years and decades suggests the need for a different null hypothesis. Essentially, one needs to show that the behaviour of the climatic signal is distinct from that generated by natural climate variability in the past, when human effects were negligible, at least on the global
    scale. As discussed in Sects. 2.1 and 3.3, this natural variability includes interannual and interdecadal cycles, as well as the broadband component. These cycles are far from being purely periodic. Still, they include much more persistent excursions of one sign, whether positive or negative in global or hemispheric temperatures, say, than does red noise.

    Secondly a significant number of “unequivocal” statements were merely subjective opinions the affixing of cardinal numbers to said statments does improve their scientific status.

    It is “unequivocal” that there are significant problems with GCM’s ar4 runs in both singularities such as volcanics,the middle atmosphere ,and co2 senstivity.

    The problematic issues with the stratosphere, and implications for future climate with AR4 were so poorly analysed that a significant number of parties required reanalysis and inclusion of an additional chapter in WMO 2010.

    The current chapter helps to place the Protocol’s climate impact within a wider context by critically assessing the effect of stratospheric climate changes on the troposphere and surface climate, following a formal request for this information by the Parties to the Montreal Protocol. As requested, the current chapter also considers the effects on stratospheric climate of some emissions that are not addressed by the Montreal Protocol,
    but are included in the 1997 Kyoto protocol. Hence, the chapter covers some of the issues assessed in past Intergovernmental Panel on Climate Change (IPCC) reports (IPCC, 2007; IPCC/TEAP, 2005). The current chapter is designed to provide useful input to future IPCC assessments.

    The troposphere and surface climate are affected by many types of stratospheric change. Ozone plays a key role in such stratospheric climate change, but other physical factors play important roles as well. For this reason, we consider here the effects on the stratosphere of not only emissions of ozone-depleting substances (ODSs), but also of emissions of greenhouse gases, natural phenomena(e.g., solar variability and volcanic eruptions), and chemical, radiative, and dynamical sratosphere /troposphere coupling

  27. Approximate annual global carbon emission in G-ton from 1900 to 2007 = 0.53*e^(0.0267*(year-1900))

    For middle of the 1910 to 1940 warming period of 1925, Carbon emission = 1 G-ton

    For middle of the 1970 to 2000 warming period of 1985, Carbon emission = 5 G-ton

  28. There are no independent and dependent variables here so we are not dealing with the situation for which the null hypothesis convention was created. Nor are we dealing with a simple between hypothesis testing. We are looking at a test between “wriggle fits” or models, consisting of observations and hypothesised relationships of those observations – in other words, after each new observation it is permissable to change both models. They will converge, call them A and B, but will A tend to converge on B or B on A, and how do you statistically measure the tendency to converge. It’s also perfectly concievable that the direction of convergence will reverse from time to time. That may inform the modification of relationship within any particular model but the expression “null hypothesis” and “burden of proof “ seem inappropriate to describe this paradigm.

  29. In coming back to this after the passage of some time, and in thinking about the challenge you have been issued, I think there are two levels to address Dr T’s paper on.

    The first is at the technical scientific method level. I think my comments that you quote stand the test of time (IMHO :)), and that in this light what Dr T writes is still a nonsense.

    The second is at the level of the polemic. And this is at the level that Dr T writes, despite (or perhaps because of) the audience he faced.

    For my money I would explicitly separate his utterances into these two categories and discuss the proposition separately in each context.

    For the first a systematic disassemble of the paragraph of his speech in which the comment appears and the one following would have him looking a fool (e.g. just read the nonsense about “placing the burden of proof on showing there is no human influence” and try and link that in some way to what scientists do wrt null hypotheses).

    But you do also need to address his arguments as polemic, because that is what they are, and that no doubt will be where he takes his argument. What he is really saying (IMHO) is that me and my mates want to say we are right about AGW (and by implication I want you the audience to sign up for this).

    This deserves a political response, namely that it rather speaks poorly of the AMS that they would let someone stand up in front of them and spout such rubbish about the scientific method to advance a political position without the audience throwing rotten fruit at them.

    You might have a more measured way to make this point. I personally think Dr T is a poor scientist at a number of levels having reviewed his output for 2010.

    • Thanks HAS, I have been thinking about separating out the scientific vs political null hypothesis, i hadn’t thought of framing this in terms of a polemic

  30. Michael Lowe

    I think that this issue is probably a political approach to dealing with the problem of “Cherry picking”. If anyone advances a hypothesis ( ie something other than the null hypothesis), it can be disproved by a single incompatible observation. We can look around and cherry pick the world for a fact that disagrees!
    However, this doesn’t apply to the null hypothesis. The idea that ‘the world just works as it always has, sometimes stable, sometimes unstable, not very influenced by CO2 but perhaps a little’ is so poorly defined that it is immune to cherry picking. Trenberth would like to appropriate the vagueness of the null hypothesis to make AGW immune to cherry picking and put the onus on others to make testable theories.

    Politically, the sceptic community would be wisest to not let him do this. On the other hand, it would be a good test of intellectual rigor to actually have to make a statable, testable hypothesis rather than spend eternity picking holes in others’.

  31. Paul Dunmore

    The null hypothesis is a concept from statistics, and it is reasonably well defined. Its essential feature is that, assuming it to be true, we can explicitly compute the probability distribution of some test statistic. We can then compare the observed statistic with that distribution. If the observation is strongly inconsistent with the distribution, then we can be fairly confident that the assumed null hypothesis is false. For example, if the observed value has a 1% or less chance of being observed if the null hypothesis is true, then we can be almost certain that it is not true. (This is a probabilistic version of the mathematical method of proof by reductio ad absurdum. But it is only probablistic – no finite number of observations can absolutely disprove the null hypothesis.)
    Dr Trenberth can choose any null hypothesis he likes, but only if he can determine the distribution of some observable test statistic that is implied by his hypothesis. “Humans have caused observable warming” does not support such a computation and is therefore not a possible null hypothesis. “The temperature response coefficient is 3.5C for doubling CO2” is, or might be if Trenberth is clever enough to do the calculations, a possible null hypothesis. Of course, it might be rejected by careful enough observations if the true coefficient is either 2C or 5C. Similarly, “the climate has not changed in 1000 years” is a possible null hypothesis, but “climate change is caused by natural variation” is not, because it does not provide a basis for computing a probability distribution.
    None of this is affected by the weight of existing evidence – it is required by the underlying logic of testing. At no point can one say “we now have enough evidence that different logic applies from now on.”
    Perhaps Dr Trenberth means something else by his “null hypothesis”, but if he has not defined what he means then it is impossible either to support or dispute his claim.

    • Paul – interesting post, but I’m struggling with “The temperature response coefficient is 3.5C for doubling CO2″ as a null. What would be the alternative hypothesis to such a null?

      • Paul Dunmore

        Tom – Any other hypothesis, most generally “The response coefficient is not 3.5C” (which would be true if the response were zero, or 2C, or 5C). Often we may have some more specific alternative in mind, such as “The coefficient is more than 3.5C” or even “The coefficient is 5C”. Some test statistics will work better (“have more power”) for one alternative than another; but their shared feature is that we must be able to compute their distribution if the null hypothesis is true. If we were comparing “The response is 3.5” with “The response is 5”, we could treat either as being the null, but usually the alternative is vaguely specified while the null is precise. In that case, only the precisely specified hypothesis can be the null.

    • John Carpenter


      “The null hypothesis is a concept from statistics, and it is reasonably well defined. Its essential feature is that, assuming it to be true, we can explicitly compute the probability distribution of some test statistic.”

      I struggled with a post last night, which I ultimately scrapped, trying to say what you stated so well. I may be wrong, but it seems to me you have to have a measurable, distinct statistical signal to accept or falsify a null, therefore the null must be stated in such a way that allows for such a measurement. Trenberth has not demonstrated this.

  32. Anything other than “unspecified natural variations” begs the question.

  33. “Given that global warming is “unequivocal”, to quote the 2007 IPCC report, the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence.”

    “Given”? Who gave? What was “Given”? “Where was it “Given”? When was it “Given”? How was it “Given”? It may also be appropriate to ask, in this day and age, why was it “Given”?

    There’s a lot of “give” and “take” in science in search of answers to questions unknown, but really, has anything been “Given”? Or is this a case of grabbing or “taking”?

    • Trenberth’s first burden should be to show where the IPCC said ‘unequivocal’.
      Is there a way to do a word search for that term?
      The second should be to ask Trenberth what he means by impact.

  34. Judy,

    Do you have any interest in disputing Trenberth’s assertion by pointing out the poor quality of the science? If I were debating him, I would insist that he not use any models which haven’t been verified and validated, that he not use data from instruments which fail to meet basic scientific standards or databases which lack minimal quality control. Or studies whose data, methods, code aren’t availiable or haven’t been replicated.

    Given the track record of highly visible studies which have been shown to have butchered stats and other errors, what I would consider sufficient “proof” would require a whole lot more than an assessment put together by a group of incompetents who all agree that their own studies need never be checked.

  35. Joe Lalonde


    The only actual damage man has done to this planet that cannot repair itself is the exchange of water for oil.

  36. Trenberth is using a bit of rhetorical trickery here. He says that the IPCC calls global warming “unequivocal”, but really they call the temperature increase of the 20’th century unequivocal and they don’t apply the term to mankind’s influence. They say it is “very likely” humans are causing it. NOT unequivocal.

    • Sam;
      Good catch!
      And, of course, the human CO2 impact is somewhere between “negligible” and “imaginary”.

  37. I agree with what a lot of posters have already highlighted- that in this instance there is more than one null hypothesis and that the wish to reverse is mainly political not scientific (and further that the IPCC and consensus are already operating under the ‘reversed null’ position).
    For my two cents the two null’s would be
    1- There is no human effect on climate wrt C02
    2- There is no significant human effect on climate wrt C02.
    With # 2 being the more likely of course.
    The null needs to be tightly confined to co2- there are obviously other human effects on climate (land use being the first to spring to mind), but they just muddy the issue. The focus of the IPCC et al is on carbon regulation- so here is where the discussion must lie (further supporting the ‘it’s a political not scientific issue’ statements made above).
    With regard to your submission- you’re in a far better position than I to comment on the specifics, but I’d be looking to include the following:
    -An examination of the ‘typical’ null hypothesis, it’s use (and reasons for) and why they are necessary.
    -A discussion on the reasons FOR the reverse of the Null Hypothesis. I.e. what scientific benefit does it give? If it doesn’t have a scientific benefit, why is it being proposed?
    -following on from that, a discussion on who the proposed reversal benefits. I.e. will there be a general benefit (i.e. it helps both sides) or is the benefit side-specific and further, is this linked to WHICH side is proposing it.
    -Finally I’d address the duplicity (if I may call it that) in Trenberths statement :
    “Given that global warming is “unequivocal”, to quote the 2007 IPCC report, the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence.

    As the first part (up to unequivocal) has no bearing on the second. The world has indeed been warming, but to suggest this proves it is man is incorrect. It’s a case of watching the pea under the thumble here. However, even if we assume that he mean the attribution of blame was unequivocal- he is still wrong- that needs to be addressed too.

    I hope you’ll post your article here too Dr Curry once finished? I’m sure it’ll be a fascinating read- Dr Trenberths too, if he agrees

    • Blast. formating fail in the cut and paste- apologies for the wall of text.

    • My general plan is to make drafts of my more relevant papers available pre-publication. However, on this one, I think it is best to wait until both papers are accepted in final form. And then probably wait until both are published (unless Trenberth makes his paper available online before that time). I think both papers should be discussed together.

  38. Brandon Shollenberger, Paul Dunmore and others have emphasized above that the idea of “the Null Hypothesis” is misplaced. The null hypothesis is always an alternative to a well defined hypothesis, and there should be a rational basis for choosing that particular hypothesis.

    For the climate change its not reasonable to pick the only unique hypothesis: AGW exists at some level, however weak. It certainly does, but that is not yet a reason to worry. Therefore we have two alternative paths that make sense:

    1. We do not define any specific hypothesis, and thus have no null hypothesis either. Instead we just try to find out whatever we can about climate change and uncertainties of this knowledge.

    This doesn’t contain any specific hypothesis, but its possible to formulate numerous different hypotheses and related null hypotheses, and get answers from the knowledge gathered.

    2. We think first what we want to decide, i.e. we wish to know, whether we have enough evidence to support a specific choice between alternatives. Then we have a different hypothesis and different null hypothesis for every such question.

    The proposal of Trenberth to invert the null hypothesis is just one way of expressing the opinion that the natural political action is to decide on a strong mitigation policy, and that those who oppose that should provide evidence that this is unnecessary. Thus his proposal has little to do with scientific procedures, but is based on an opinion concerning the present settings for decision making. He has likely given much weight on “the precautionary principle” in his judgment.

    This issue is also closely related to my previous comment in another thread

    and some postings on my own blog including this

    • Thanks Pekka

    • If I can repackage what you just said, once again, policy drives all. The question of which null hypothesis formulation makes sense is meaningless outside of the context of a preferred policy option.

      It seems like that’s becoming a law; no question is meaningful absent a policy context. And this is true regardless of what any of us want.

  39. I am a lawyer, so I see this issue as a shifting of the burden of proof.

    Trenberth wants to change the trial from proving guilt beyond a reasonable doubt to proving innocence beyond a reasonable doubt.

    So instead of the rain in Minneapolis today being caused by weather (the old burden of proof or null), it is now caused by human influence on the climate (the new null).

    Now, weather and climate have always varied – for many many natural reasons.

    Instead of trying to understand those, this new null ignores all natural causes (cosmic rays, volcanos, carbon black, land use changes, methane releases, etc. – and basically says humans caused whatever happens, by default).

    It makes no sense to say that all previous changes in the climate (pre-1850) were natural, but all changes in future climate are unnatural, unless you can prove otherwise beyond a reasonable doubt.

    Humans caused sea level rise to decelerate – the new null? How did we do that – did we stop putting out CO2?

    Humans caused a flattening in ocean heat content over the last 8 years – the new null? How did we do that?

    Humans have caused the temperature to rise slower than the models projected – the new null? How did we do that?

    Humans caused the sun to enter a quiet phase – the new null? How did we do that?

    Personally, I don’t think Trenberth’s position makes logical sense.

    Trenberth has the burden of proof to show that a change in the default burden of proof should flip – and just stating what the consensus is in AR4 is not proof sufficient to meet Trenberth’s burden.

    All of the things we don’t understand about the climate are based on our inadequate understanding of the climate – not on human’s influence on the climate IMO.

    • Thanks Rick, i like the “shifting of the burden of proof”

    • Excellent post- my thoughts far more eloquently put :-)

    • So with this shift, AGW would be assumed to magnify any weather event. For example, Tornadoes would be assumed to be 5% more powerful due to human caused warming unless you could prove otherwise, which would require you to be omniscient.

    • Bravo RickA!

      When warmists started calling sceptics denialists, I immediately knew that they’re projecting – they were the ones who denied climate change. You HAVE TO deny climate change (natural) in order to believe in CO2AGW.

    • Is it possible you misunderstand Trenberth’s position?

      How about all weather and climate is affected by the rise in CO2 caused by humans.

      Slightly different than all weather and climate is caused by the rise in CO2 caused by humans.

      Are we looking at a classic straw man argument?

      Can the skeptics prove the current climate is the same as 100 years ago, before the significant contribution of CO2? Within the range of natural variability that has existed for the last thousand years?

      • bobdroege,
        Can you show us the climate has ever been unchanged over 100 years?
        More importantly, has the last 100 years seen the climate change in ways that are unique, much less more dangerous, than the 100 years prior? 500 years? 1000?
        You can weave, but you cannot dodge.

      • How about the period between the medieval warm period and the little ice age for a period that the climate was stable?
        It was at least 100 years, maybe as much as 300 years.

        Can you show me a historical recontruction that shows a 1 C change within 100 years? That would be what you need to show a historical climate change greater than that which we are currently experiencing.

        If the current trends continue, considering the warmest year was 2010, 1998, or 2007 depending on which index you use, and all show warming trends, and that historically 3 C above the 20th century average was enough to melt the majority of the icecaps, I would say that was dangerous.

        The argument that the climate has always changed, and always will change, is not a very good argument for saying we should continue to change the climate.

      • Melting ice caps?

        Maybe now are you looking at a classic straw man argument

      • Are you implying that the icecaps have never melted before?

        Look for evidence of the Hothouse climate of the Cretaceous period.

        They have drill beneath the Greenland ice cap and found…

        Medieval Viking Helmets!

      • (It just seems to be going off topic. )

        So Greenland becomes more inhabitable …what’s the danger in that? The Vikings seemed to like it.

      • “what’s the danger in that? ”

        Sea level rise

      • bob –
        I don’t think they drilled beneath the ice cap. Those artifacts are found by digging in the MWP Norse middens.

        As for sea rise – you’re playing the catastrophist role. It’s not a pretty sight.

      • “I don’t think they drilled beneath the ice cap. Those artifacts are found by digging in the MWP Norse middens. ”

        You would be wrong about drilling beneath the icecap

        Of course they didn’t find any medieval viking helmets.

      • Make up your mind, bob – you said this –

        They have drill beneath the Greenland ice cap and found…

        Medieval Viking Helmets!

        Now you’re saying this –

        Of course they didn’t find any medieval viking helmets.

        What are you trying to say?

      • so is the Sea level rise the real danger? your target keeps changing. The sea has been rising for many centuries. People that live on the coast have dealt with it for centuries.

      • Jim Owen,

        One would think it was obvious that I was joking about finding Medieval Viking Helmets under the greenland ice cap.

        But two million years ago or so, there was significantly less ice on Greenland, and how much warmer was it then?

        Natural or artificial changes to the climate can happen.

        And in your opinion, I am playing the catastrophic role, why don’t you argue against the points I make rather than straw men of your own construction?

      • One would think it was obvious that I was joking about finding Medieval Viking Helmets under the greenland ice cap.

        Given some of the other things you’ve said, you might have been dumb enough to be serious.

        But two million years ago or so, there was significantly less ice on Greenland, and how much warmer was it then?

        Two million years ago – how many humans were there? Did they survive? Why do you believe that they survived with the level of technology that they had, but humans today cannot survive using the level of technology we have – or will have in the future (if people like you don’t short-circuit that technology? ?

        Natural or artificial changes to the climate can happen.

        We know that statement is half right. The other half is in question. Otherwise we wouldn’t be here, questioning each others intelligence, parentage, etc.

        I am playing the catastrophic role, why don’t you argue against the points I make

        What points? Have you made any coherent points? All I’ve actually seen is a rerun of AIT and a lot of goalpost moving.

        For example – you said –
        How about the period between the medieval warm period and the little ice age for a period that the climate was stable?

        It was at least 100 years, maybe as much as 300 years.

        What makes you think that period was “stable”? Who told you that?

      • This much warmer won’t be pretty.

        But it’s just more from the AGW gravy train.

        This quote is interesting, in that it doesn’t matter if the CO2 is natural or artificial, the result is the same.

        from the early Pliocene, when carbon dioxide levels remained close to modern levels for thousands of years, may indicate how warm the planet will eventually become if carbon dioxide levels are stabilized at the current value of 400 parts per million,” said Aradhna Tripati, a UCLA assistant professor in the department of Earth and space sciences and the department of atmospheric and oceanic sciences.

      • Bob –
        This much warmer won’t be pretty.

        If you’re getting your information from environmental/left/alarmist sites it explains a lot of what you’ve been saying here. Doesn’t excuse it, just points up your problem.

      • bobdroege,
        The period between the MWP (does it now exist again?) and LIA was ‘stable’?
        How is something stable when it is moving between warmth and a mini-ice age?
        Can you show a 1oC change in 100 years? the current period does not have that, if I recall correctly, so please show me I am wrong.
        irt 3C- How can there be an historical record of something you are claiming is unique?
        Sorry, but your entire line of quesiotning does not make sense.
        Please feel free to clarify and get back to us on this.

      • Look, I never said the Medieval Warm Period never existed.
        And I don’t think I have claimed the current warming trend is unique either.

        I am claiming that there was a flat period of temperature after the end of the MWP and before the start of the LIA.

        This chart shows a flat period between 1200 and 1400 AD.

        I specified 1 C in 100 years and I can’t show you a 10 C change in 100 years.

        There were periods that were 3 to 5 C warmer than today, you are aware of that, yes?

      • bob –
        Your Wiki chart doesn’t support your words.

        Have you REALLY researched what you’re saying?

      • You make my points precisely. Thank you very much.

      • There were periods that were 3 to 5 C warmer than today, you are aware of that, yes?

        With sea levels 80 meters or so higher?

        I don’t think you are making any points.

      • bobdroege,
        You are perfectly demonstrating an ever dynamic climate.
        There is no such thing as static climate on Earth.
        Which is my point, and which you have amply verified.
        Thanks again,

      • You think that’s a valid point?????????????

        But the question is whether or not man can change the climate.

      • You think that’s a valid point?????????????


        But the question is whether or not man can change the climate.

        No. That’s one incomplete question. There are others.

      • bobdroege,
        I don’t where the heck this will show up on the linked conversation. I am replying to your comment about my point.
        My point is that whether or not humans can change the cliamte is irrelevant. We always have and always will.
        The question is are we, with CO2, going to change the climate in ways more dangerous than other periods of climmate change?

      • Yes, that is possible and hopefully his essay will explain his position better.

        I think that “Ship of Theseus” is a paradox(amongst others) that can help express the difficulties in going back 100 years ago using models. It could be too gunky.

        The ship wherein Theseus and the youth of Athens returned [from Crete] had thirty oars, and was preserved by the Athenians down even to the time of Demetrius Phalereus, for they took away the old planks as they decayed, putting in new and stronger timber in their place, insomuch that this ship became a standing example among the philosophers, for the logical question of things that grow; one side holding that the ship remained the same, and the other contending that it was not the same.
        —Plutarch, Theseus

        wikipedia entry

        or “Heraclitus’ river”
        you can never step into the same river twice

      • bobdroege –
        Can the skeptics prove the current climate is the same as 100 years ago, before the significant contribution of CO2? Within the range of natural variability that has existed for the last thousand years?

        Your first question is nonsense. 150 years ago was the approximate end of the LIA. Climate HAS changed since then – and NOT due to CO2.

        The only time range you can claim CO2 to have had any effect is the last 50 years or so. And both the magnitude and rate of warming have been less during that time than during the previous 50-100 years when CO2 was not a factor.

        The conclusion therefore is – CO2 retards warming.

        Don’t take that too seriously – I don’t.

        OTOH – can you prove that the anthropogenic contribution of CO2 has had ANY effect on climate? I’ll even stipulate a small contribution to the present warming – very small.

        So far I’ve seen a lot of speculation and a LOT of crazed caterwauling, but nothing that stands up to rational examination and not one iota of actual linkage. If you have it, then trot it out here for all of us to see.

      • Looking at Hadcrut3, I get 0.6 C for the last 50 years, and about 0.4 C for the 50 years before that, so the magnitude and length are not comparable.

        But lets say they were both 0.5 C and 50 years.

        Then the last 0.5 C involves a bit more energy due to the fourth power term in the Stefan-Boltzman law. Which is why the CO2 effect is not linear.

        And I’m sure you didn’t mean to say

        ” I’ll even stipulate a small contribution to the present warming – very small. ”

        Because that means you agree with Trenberth and reject his new null hypothesis.

        That’s good news because that means you are not a “denier”

      • bobdroege:

        I don’t think I misunderstand Trenberth’s position, but correct me if you think I am wrong.

        My take on his position is that on any climate issue whatsoever – if you cannot prove there is no human influence (refute the null hypothesis by showing it is less than 5% probable or 1% probable or whatever), then we have to conclude that it (whatever the question is that is being looked at) is caused by human influence.

      • I think what he is trying to say is all climate and weather events are influenced by mankind, not that any specific event is “caused” by man.

      • But to what end are they influenced by mankind?
        We influence the climate/environment in The Netherlands by building levees and pumping out marshes and protecting them from tidal and storm surges.
        Is that good or bad?
        We turned the Great American Desert into the bread basket of the world, changing the climate by land use/vegetative changes. Were those changes significant, or destructive?
        Ancient Middle East peoples disrupted the climate by over grazing and other agrarian and deforestation practices. was that severe or benign?
        Aboriginal people in Australia used brush burning to change the land, drive game, and changed the climate. was that good or bad?
        Trenberth is not after those questions. He is after imposing CO2 policies by ignoring the question of whether or not there is any evidence that additional CO2 has influenced the cliamte to do anything outside the range of historical experience, positive or negative.

      • That’s right, Rick, you haven’t understood the point of argument.

        Trenberth argues that since science/physics has already established the human influence on climate, oceans, etc. (and Curry would not say otherwise) it makes more sense for Curry to have to show that there is no influence on water vapor and precipitation (i.e., intensification of storm activity/heavy precipitation) than to show that there is, because of basic physics/physical systems/physical relationships that constitute the global climate cycle. That is the specific argument at hand.

        The ‘null’ (pretend) hypothesis Curry would have to hold contradicts the evidence of current studies, logic – and frankly physical reality.

        As Bobdroege says, with Trenberth, the question of whether this or that storm is caused by human activity overlooks the basic fact that at this point in history both natural and human influence are involved.

        Now, from the perspective of policymaking on mitigation of global climate change through emissions reduction, the attribution of a specific heavy precipitation event to human-caused GHG’s is not an extra development in science that is needed to add to the burden of proof regarding the human influence on climate already provided by the current scientific evidence. However, from the perspective policymaking on adaptation, it may be helpful in future to develop much more specific attribution capacity in the science for the obvious reason that as part of the overall development of better forecasting of regional impacts/events, it improves specific risk planning and claims for aid.

        Hope that helps you.

      • Martha,

        1) You state:”…it improves specific risk planning and claims for aid.”

        In what way is it an improvement? if their predictions are all over the map and you can only state the risk in subjective terms that are heavily influenced by politics and exaggerations to sell newspapers then how is that an improvement? I would say it was a disservice to risk planning and claims for aid.

        2) Even the hard science of physics is not infallible. Take the Millikan drop experiment as once described by Richard Feynman:

        We have learned a lot from experience about how to handle some of the ways we fool ourselves. One example: Millikan measured the charge on an electron by an experiment with falling oil drops, and got an answer which we now know not to be quite right. It’s a little bit off because he had the incorrect value for the viscosity of air. It’s interesting to look at the history of measurements of the charge of an electron, after Millikan. If you plot them as a function of time, you find that one is a little bit bigger than Millikan’s, and the next one’s a little bit bigger than that, and the next one’s a little bit bigger than that, until finally they settle down to a number which is higher.
        Why didn’t they discover the new number was higher right away? It’s a thing that scientists are ashamed of – this history – because it’s apparent that people did things like this: When they got a number that was too high above Millikan’s, they thought something must be wrong – and they would look for and find a reason why something might be wrong. When they got a number close to Millikan’s value they didn’t look so hard. And so they eliminated the numbers that were too far off, and did other things like that…

        And they were able to do experiments with real result with climate science you can’t.

      • Martha:

        You say “the question of whether this or that storm is caused by human activity overlooks the basic fact that at this point in history both natural and human influence are involved.”

        I agree that both natural and human influence are involved.

        However, what is the breakdown between them?

        Of the .8 C warming since 1850, how much is from natural warming and how much is from human influence.

        In my naive opinion, unless more than half the warming is due to human influence I don’t see how anyone can argue to switch the null from natural warming to human influenced warming.

        I don’t think we have a very good handle on how much of the warming is natural vs human.

        I think the error bars or uncertainty on this issue are greater than the actual warming of .8C .

        Once we establish how much of the warming is human influenced, I would then be interested in a further breakdown of that human caused portion.

        How much of the human influenced warming is due to CO2, how much to carbon black, how much to land-use changes, how much to cow farts, etc.

        Only when we have an estimate of how much the CO2 put out by humans has contributed to the warming, can we even begin to do a cost benefit analysis related to mitigating CO2 output.

        I don’t feel that we have a good handle on the this breakdown, let alone just the human caused portion.

        So I still think it is on Trenberth to show that the human influence on climate is greater than the natural influence, before we even think of switching the null (i.e. the burden of proof).

      • John Carpenter

        “it makes more sense for Curry to have to show that there is no influence on water vapor and precipitation (i.e., intensification of storm activity/heavy precipitation) than to show that there is”


        “the question of whether this or that storm is caused by human activity overlooks the basic fact that at this point in history both natural and human influence are involved.


        “….the attribution of a specific heavy precipitation event to human-caused GHG’s is not an extra development in science that is needed to add to the burden of proof regarding the human influence on climate already provided by the current scientific evidence.”

        As you are aware Martha, weather is not climate and so will never be a development that science will be able to use as “proof” regarding the human influence on climate.

        Regardless of that undeniable fact… when you author a peer reviewed journal article demonstrating a measurable human component to any weather event… send me a signed re-print so I can frame it and hang on my wall.

        Good luck with that!

    • John Carpenter

      “I am a lawyer…”

      Can’t live with em… Can’t live without em…

      Today I can live with em… ;) Nice post Rick, well stated.

    • RickA,

      The only problem I have with describing what Trenberth is attempting as “shifting the burden of proof,” is that it implies the burden of proof was first acknowledged to rest on the CAGW proponents, and should now be “shifted” onto skeptics. But I don’t know that any null hypothesis has ever been articulated by the consensus community. In other words, I don’t know where they ever acknowledged that they bore the burden of proof, let alone described what that burden might be, or how they met it.

      Trenberth essentially wants to claim that the consensus has met its burden of proof (ie. falsified the original null hypotheses) without ever stating what that burden (or null hypothesis) was. One of the primary complaints I see from skeptics better versed in the science than I am, is that there has never been a null hypothesis against which CAGW could be tested. Trenberth is not so much shifting the burden of proof in the traditional meaning of that term, as he is trying to skip the issue entirely for the CAGW hypothesis.

      For an example, in a civil rights employment case, the plaintiff has the initial burden of proof of showing that a particular employment policy had a discriminatory impact; that is his prima facie case. Once the plaintiff presents evidence to support that prima facie case, the burden shifts to the defendant to present evidence that the policy had a legitimate business purpose. What Trenberth is essentially trying to do is skip the requirement for establishing a prima facie case of CAGW, or even defining what the terms of that case are.

      In the common legal usage of the term, what Trenberth is trying to do is not so much shift the burden of proof, as rewrite the rules of procedure (or in this case, of the scientific method), which is much more radical.

      • I’d call it akin to pigeonholing

      • Gary:

        You could be right – Trenberth might be trying to change the rules of procedure.

        However, I always thought that the null was natural variability caused the warming.

        So, given that the climate varies due to both natural and human influences (I don’t think anybody disputes that), isn’t it on Trenberth to prove the human influences outweight the natural, before we switch the null from natural to human influenced?

      • Rick,

        If it were accepted that “natural variability is causing recent warming” was the initial null hypothesis, I would agree. For skeptics that is the null I would think. But I don’t think the consensus community ever accepted that (or anything else) as a beginning null hypothesis. The papers Dr. Curry and Trenberth are writing are based on his statement, which simply assumes that the consensus has already proved (to themselves apparently) that human influences outweigh the natural.

        So in the real world, you are surely correct. In the world of climate science, I think Trenberth’s attempt to rewrite the rules has to be resisted. It is as though, in the middle of a trial, the plaintiff realizes that he does not have sufficient evidence to meet his burden of proof. So he proposes that the rules of civil procedure (in this case the scientific method) be changed to require the defendant to bear the burden of proof.

        In the case of climate science, as I wrote above, I think the real jury in the U.S. is the electorate. What Trenberth wants is for the scientific community to combine and convince the politicians and the public that CAGW has been established and that public policy should follow unless skeptics can now disprove CAGW.

        In other words, Trenberth’s statement is just another rephrasing of “the science is settled,” dressed up as a legalism, and trotted out to see if anyone will buy it in this new form.

      • If it were accepted that “natural variability is causing recent warming” was the initial null hypothesis, I would agree. For skeptics that is the null I would think.

        What would the test be for “natural variability?” Would that “variability” be a “variability” in some measurable quantity? Or is it merely a statement of ignorance: “well, we don’t know for sure what is causing the warming, so let’s just call it natural?”

        I agree that no research hypothesis, let alone a null hypothesis, has been presented for anthropogenic climate change as a whole. Neither has either been presented for the notional “natural variability.”

        Steven Mosher is right. Until some rigor is imposed on this discussion, it will continue to go around and around.

      • Steven Mosher is right. Until some rigor is imposed on this discussion, it will continue to go around and around.

        What? The IPCC doesn’t supply that? :-)

        Actually, I agree with you (and Steve) on this one.

      • The IPCC doesn’t supply that? :-)

        No, but then again they don’t purport to.

        Part of the problem is epistemological. How would one structure a falsifiable null to Newton’s theory of gravitation, for instance?

        every point mass attracts every single other point mass by a force pointing along the line intersecting both points and this force is proportional to the product of the two masses and inversely proportional to the square of the distance between them

        Our understanding of gravity is based on a number of hypotheses, each testable, each falsifiable. Gravitation is not itself an hypothesis, but an observation. The same can be said of anthropogenic climate change. The phenomenon is too vast to be verified or falsified by a single test. However, a number of verified or falsified hypotheses can accumulate to a body of knowledge that could be called a “theory.”

        The details matter.

      • PDA,

        Asking me what “the test for ‘natural variability’ would be” gets things exactly backwards. I was referring to Trenberth’s statement: “Given that global warming is “unequivocal”, to quote the 2007 IPCC report, the null hypothesis should now be reversed.”

        Reversing the null hypothesis would require that there be something to reverse, ie. an original null hypothesis. My whole point was that no such null has been established as far as I have read, a point with which you apparently claim to agree.

        But you in fact seem to adopt Trenberth’s position. “[I]s it merely a statement of ignorance: ‘well, we don’t know for sure what is causing the warming, so let’s just call it natural?’” You apparently would answer no (as would Trenberth). I on the other hand would say, yes, when it comes to public policy what else should we assume?

        And of course it is a species of ignorance. That’s the whole point of a “burden of proof” (or in this instance a null hypothesis). In the absence of additional, sufficient evidence (ie. where we are ignorant of the ultimate facts), what is the default position we should take?

        It does not seem to be in dispute that the climate has been changing without stop throughout the history of the Earth. And I don’t find “natural variability” in this context to be as complicated a phrase as you seem to. I take natural in this instance to be anything other than anthropogenic, and variability to be a synonym for change. So where is the complication? Why would the “quantity” of a variable be proof of its causation, particularly on a planet with as wide a range of climate history as this one has?

        Trenberth’s “shifting the burden of proof,” the “uncertainty principal,” the “consensus,” the “science is settled,” these are all framings of the same central argument of CAGWers – that they know enough now to justify massive social and economic change. Your suggestion that assuming natural cause over anthropogenic is “ignorant” (or at least more ignorant that the contrary), is really nothing more than Trenberth’s comment restated. If you don’t think climate change should be assumed to be natural absent evidence to the contrary, perhaps you could explain why?

      • If you don’t think climate change should be assumed to be natural absent evidence to the contrary, perhaps you could explain why?

        Logically, this is a perfectly reasonable position for a layperson to take. We’re talking about hypotheses, however, and the fundamental quality of an hypothesis is that it can be tested. There simply is no “assuming” involved: either one can show statistical results at (e.g.) 5% significance, or one cannot.

        I do not think that “there is no human influence on climate” is a testable hypothesis, because it’s too vague. Similarly, I do not think that merely reversing that hypothesis makes it any more valid. I suspect Trenberth was intentionally being provocative, but I don’t think taking such a casual attitude about the null hypothesis was helpful on his part.

        Again, the null hypothesis can’t be “we don’t know.” That’s not an hypothesis. That’s just ignorance. There can be a whole plethora of nulls – “climate sensitivity is less than 3.5° K,” “cloud feedbacks are not positive,” etc. – and confirming or rejecting these can accumulate to an understanding that humans are or are not affecting climate. But that understanding is not in and of itself an hypothesis.

        That’s the point I’m trying to make.

      • PDA,

        As far as in a scientific/academic sense, I don’t see any problem with your position. My first comment on this thread began with the following:

        “I will leave to the scientists and academics the proper formulation of the null hypothesis, or hypotheses, with respect to climate science. But Trenberth’s statement was not about scientific analysis, it was about policy. When he says that AGW should be the new null hypothesis, he means that unless skeptics can “prove” the contrary, governments should formulate policy based on AGW, if not CAGW.”

        So for scientific/academic purposes, I take no issue with what the requirements are for a “null hypothesis.” I did google the term when this thread first started, and found about 13 different definitions of the term, depending on the context in which it was being used – statistical, financial, etc. I found this example on one site: “‘Hyperactivity is unrelated to eating sugar.’ is an example of a null hypothesis.”

        As I noted in my initial comment above, I was using the term in the sense that Trenberth was, a political null hypothesis. And it is all well and good to say that he got it wrong. But a large part of the “scientific” community is taking it seriously, so your objections on the misuse of the term might better be directed to them.

        Here is the problem I have with “technical” objections like this. When Dr. Curry first posted the thread, I did not see any AGW or CAGW adherents objecting to Trenberth’s use of the term in the political sense. Only when skeptics began to argue with his use of the term to frame the debate, did I begin to see objections about how we/skeptics are misusing the term.

        I suspect that the problem AGW/CAGW proponents have with forming a null hypothesis for their position is that it would make their position falsifiable, which is what I understand the purpose of a null hypothesis to be. And I don’t think those who advocate the AGW/CAGW position have any intent of allowing their beliefs to be so tested.

        While I am no scientist, this reluctance to agree to what a null hypothesis should be for AGW/CAGW is to me no different from the defense of climate models by claiming they do not need to be validated, or claims that peer review of the published literature does not need to ensure the correctness of the article being published.

        It seems to me that climate scientists are all about the rigor, until it comes to their own arguments.

      • As I noted in my initial comment above, I was using the term in the sense that Trenberth was, a political null hypothesis. And it is all well and good to say that he got it wrong. But a large part of the “scientific” community is taking it seriously, so your objections on the misuse of the term might better be directed to them.

        Well, I don’t know if I can speak to paraphrases of arguments supposedly made by other people. I’ve expressed my opinion and my understanding in this thread, the topic of which is “the null hypothesis.” I think there’s some misunderstanding – on both sides – about what an hypothesis is and is not, so I’ve pointed that out.

        I suspect that the problem AGW/CAGW proponents have with forming a null hypothesis for their position is that it would make their position falsifiable, which is what I understand the purpose of a null hypothesis to be.

        What I’ve expressed is that there cannot – logically – be a null or a research hypothesis for something so broad as the idea of anthropogenic climate change itself. Insofar as Trenberth suggests that there can, I think – again – that he is in error. Are there people who advance this position? Who are they? How have they expressed it? I generally find it more interesting and productive to engage specific arguments made by specific people, rather than broad-brush characterizations made by means of paraphrase.

      • GaryM | May 20, 2011 at 11:19 am |

        I was using the term in the sense that Trenberth was, a political null hypothesis.

        Dr T goes to great lengths to suggest he is making a scientific argument e.g. he goes on to talk about Type II errors etc, and his speech is all about what the science should be doing (although I concede that some would say that climate science is a political endeavour rather than a scientific one).

      • I think KT made it pretty clear that the IPCC was the governing authority; the ‘jury’ as you say. The question becomes, how does a UN agency become a scientific authority? Are they trading on the name of the UN?

        Show of hands. How many consider the UN to be authoritative, simply because they’re the UN?

      • ChE,

        I think Trenberth, and the consensus community in general, would love the IPCC to be the jury. But if that were the case, the litigation is over and they won. The IPCC is fully on board with decarbonization, in all its massive, intrusive, statist glory.

        But if Trenberth really believed that was the case, his whole argument would be superfluous. Why would he be arguing now for shifting the burden of proof, if the jury that mattered had already returned a verdict in his favor?

  40. This argument about whether there is one or more null hypotheses seems to me to miss the point – there are as many nulls as there are alternatives, and the alternative inescapably frames its null.

  41. Trenberth did a couple of neat tricks in reversing the null.
    First he put the burden of proof on those who question his work, which means, if his position is accepted, that he is in the same position as a religious organization, forcing doubters of the faith to prove a negative.
    Second, he redefined AGW as ‘any impact on the climate by man’.
    That is very clever, since humans have had impact on the climate since sheep over grazed in the Middle East and farmers started diverting water and changing vegetation and land use. The issue, as presented by his life work and the richly paid body of researchers has had to do with CO2 causing a world wide climate catastrophe. Not ‘any impact’, and not vague attribution of that impact. It was always about GHG’s, and CO2 was always the ghg to be worried about. This cleverness by Trenberth now lowers the bar and changes the game from ‘calamity’ to anything at all. Environmental impacts go from tiny to massive. In a rational world, we would spend large money where there is large risk. We would expend effort where we get substantial return. Trenberth’s new game allows him to continue ot demand huge sums of money to study what is measurably very insignificant, and to pretend that anyone questioning him is wicked and contemptible.
    The first on its own merely makes puts his work on the same level as a televangelist. The second puts him in a position where he gets to claim great victory over very little and bill us huge sums of money.

    • I agree very much.

      It’s Orwelian to call it AGW and to actually mean (A)CO2GW, because the issue is (A)CO2 and note any anthropogenic influence.

      • Edim,
        Thanks. To clarify my point a bit further, the question I am really asking is this:
        Is CO2 added by humans making the climate any more dangerous than it already was?
        As people in the Bay of Bengal, the US Gulf Coast, Tornado alley, and basically everywhere else in the world can attest, cliamte/weather is plenty dangerous.
        Is the increasein CO2 making it more so?
        So far the answer is clearly, “no”.
        Trenberth apparently, by his need to reverse the question, knows this at least at some level and is avoiding the issue.

    • randomengineer

      Correct, the notion of “any” impact is meaningless since all species have some effect (e.g. exhalation, deforestation,etc.) By changing the rules one changes the dicussion.

      Trenberth’s “null” is correct in the common sense manner of noticing that species impact their environments. But outside some basic common sense observations it’s utter crap, thus his “null” is range limited. A range limited null is useless.

      • randomengr,
        By claiming that the game is about ‘any’ impact, Trenberth is fuelilng the misantropic creeps who write books like “Time’s Up!”, the call for xenocidal terrorism against the industrial world.

  42. John F. Pittman

    The shifting of the burden of proff occurs by polemic with Given:Unequivocal. The problem is that this statement is offered as proof for the rest of the polemic. As pointed out, the actual wording of AR4 is very likely. The conclusion was based on a priori Bayesian structure that was used to support the conclusion of unequivocal. But these a priori are jsut as subject to be wrong as any other hypothesis. As others have pointed out this reduces the argument to a conflict with science as practiced. One of the intersting aspects of using the legal model as Gary outlined, is that part of the burden of proof is that scientists work to be considered must be as practiced. A case involving intelligent design was ruled as not meeeting the burden was that the persons involved in the ID were not scientists and DID NOT DO the science. As others have pointed out, his polemic is that they no longer need to do the science. This is not accepted for sacience nor in other realms such as law. So, the real problem is that Trenbeth is offering a static conclusion which would preclude thefalsification of any one of the a priori, and thus is not science.

  43. “One of the supreme ironies of Dr T saying we should adopt “AGW true” as the null means that climate science should then be focused on disproving this. This is of course the skeptics’ position and it’s the failure of the believers to do this that leads to much of the angst.”

    I don’t think you have a leg to stand on with this statement.

    The believers have made no attempts to, for example, to “model climate without the influence of CO2”

    I’m sure that all the “believers” would certainly like to be wrong about AGW.

    It’s like if there was no cancer, I would be unemployed. I would certainly prefer that state of affairs. I could find another career.

    • bobdroege,
      You would be wrong in that hope.
      Many misanthropic people have been attracted tot he AGW community. The AGW community has done little to nothing to seperate these disturbed people from the community.

    • I think a better analogy would be that if someone were to say that cell phones are the cause of cancer and that is the null hypothesis. If patient A has cancer it would now be presumed to be due to cell phones, you have to now prove it was not due to cell phones. Maybe patient A never had a cell phone, it doesn’t matter, you still have to prove (get witnesses and whatever) he didn’t and have contact with cell phones.

    • Rob Starkey

      It is interesting that many assume that a warmer world is worse for humanity overall, in the long term, with absolutely no relaiable evidence to support their fear.

    • bobdroege

      That was simply a methodological observation. If you set Ho as “AGW is true” you should then set up experiments to falsify this, eg show that ACO2 increase has been limited, or that CO2 sensitivity is below the threshold required to make the temp increases attributed to AGW, etc etc. I don’t see a lot of climate scientists trying to disprove this Ho (in fact most like Dr T tend not to be experimentalists anyway).

      Where you get an advance in scientific understanding is at the point when you manage to disprove the H. It is this that makes Dr T’s speech so amusing from a scientific method point of view. He is urging people to spend their time disproving AWG. I suspect what he really wants to happen is for science to disprove the Ho that AWG is false.

      Among other things he obviously misunderstands that there is no presumption in favour of Ho, and in fact explicitly argues in his example about precipitation that if you can’t falsify Ho the experiment somehow adds weight to Ho. In his example there is insufficient information to make a statement one way or the other about whether precipitation has increased by 5%. You or I might conclude we have learned little, but Dr T thinks where you have insufficient information to say anything useful, this somehow can be used to add weight to the Ho of your choice.

      Save me.

  44. I have two related points that encompass some of the preceding discussion:

    1) Consider a large sample, A, with a standard deviation > 0. If the null hypothesis is that A differs from itself, A, that hypothesis cannot be falsified even in theory – another way of saying that we can’t disprove a human influence if it can be allowed to be infinitessimal. This type of null hypothesis is, of course, not one currently embraced within legitimate climate science discussions. If a null hypothesis is to be reversed, it must represent one of the claims actually advanced as to magnitude and consequence of human impact, neither of which is claimed to be zero. A useful exercise would therefore be to specify a true “current null hypothesis” among a number of similar possibilities. The two forthcoming articles could serve a purpose by focusing on a realistic definition of the current null, or even a set of the same. This will also be an ultimate basis on which societal decisions are grounded.

    2) I mentioned above that Kevin Trenberth would have discussions in this thread as a source of information that he might use in refining his own article before the deadline (even though he has already submitted a draft). My personal view is that this would be a good thing, because it would permit both him and Dr. Curry to address the advantages and disadvantages of the same, realistic proposition(s) rather than a strawman argument his quoted statements imply. Both participants have the ability to engage in an informative discussion/debate if they agree on what it is they are disagreeing about, and it would be a shame not to have that happen. I recommend that Dr. Curry mention this to the journal editors in case they might then choose to inform Dr. Trenberth about the existence of this thread. Each author can remain “blind” to the actual contents of the other’s paper, but it wouldn’t hurt for each to have some idea of what subject will be discussed.

    • Fred,
      It sounds like you are starting to realize the problem Trenberth has created for himself. You are far more diplomatic and kind to Dr. Trenberth than his treatment of skeptics.

    • To extend the above concepts, I suspect that Kevin Trenberth may not insist on a literal interpretation of “no human influence” and will accept as an alternative “no human influence significant enough to require a concerted effort to counter its effects”. That distinction may prove not to be particularly problematic for him. My guess is that he will focus more on the question of why that null hypothesis, however phrased, should be abandoned in favor of its opposite, the existence of significant influence. His argument will be based on probability and impact – that there is now enough evidence to render significant adverse impact likely, and that the human and economic cost of inaction would outweigh the cost of effective action. This is a familiar argument, of course, but Trenberth is likely to cite evidence in its favor, at least as regards the impact if not the cost/benefit calculation. “Reversing the null” is simply a way of saying that if we can’t disprove significant human influence, we have enough information to conclude we are better off assuming its existence than its non-existence.

      If this is indeed a focus of Trenberth’s article, I see it as an area in which Kevin Trenberth and Judith Curry are not diametrically opposed, but rather disagree in terms of the magnitude and pace of future actions as a function of our certainty or lack of it about climate impacts. The forthcoming articles would allow them to define, and perhaps narrow, the boundaries for determining what to do as more climate data emerge, how rapidly to do it, and which data to pursue with vigor. This would be a more useful debate than one focused on the literal meaning of Trenberth’s claim.

      • Fred,
        There is nothing in r. Trenberth’s public statements, or leaked climategate communications, that leads someone to think he is willing to back down or compromise his position as you suggest at all. Do you have personal knowledge for your assertion, or is it a hopeful conjecture?

      • It’s a tentative prediction, Hunter. Kevin Trenberth is smart. He is also shrewd, and I expect both qualities will lead him to emphasize what he perceives to be the relevant climate evidence rather than to insist on a literal interpretation of “no human influence”. He may not, but it is what I would do in his circumstance.

  45. randomengineer

    You’re asking the wrong question.

    This is akin to arguing whether the glass is half empty or half full — an engineer will tell you that the problem is that you have the wrong glass.

    It’s obvious that the biosphere affects climate and vica versa so the correct question to ask via null hypothesis is whether human influence is capable of overriding the natural order, i.e. human influence by itself is able to make (adverse) changes.

  46. Michael Larkin

    Given that global warming is “unequivocal”, to quote the 2007 IPCC report, the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence.

    This statement is puzzling. Assuming global warming to be unequivocal, that says nothing about its causes. There is broad agreement on all sides that global warming has occurred. The hypothesis postulated is that this is primarily due to human influence that will have net negative outcomes for humanity.

    Trenberth seems to want to be saying that the preponderance of evidence is in favour of deleterious human influence (“deleterious” must be implied, because if it weren’t, where’s the problem?). He’s asserting that this hypothesis has assumed the status of greatest likelihood by dint of accumulated evidence.

    Until this hypothesis came along, I wonder if there was ever much emphasis on the counter-hypothesis that “human influence” (specifically CO2 production) on climate is negligible, which may have been mostly tacitly accepted. Indeed, it may not have occurred to at least some scientists to entertain a different possibility.

    If so, then the formulation of the deleterious(d) AGW hypothesis is what created emphasis on the counter-hypothesis by those who thought it unlikely. And this counter-hypothesis is, I suspect, what most people have in mind when they are talking about the null.

    In practice, it doesn’t matter a hoot what one calls the dAGW hypothesis or its counter-hypothesis (or any other related hypothesis). Many people already believe one or the other and are intent on refuting the opposition. What it boils down to is that one faction believes the accumulated evidence is sufficient to be able to say dAGW is a FACT rather than a hypothesis, and another that it is sufficient to be able to say the exact opposite. I’m not saying that there aren’t other factions with more nuanced views, but the principal acrimony is between the two factions I’ve mentioned.

    Nothing is going to stop the battle playing out to the bitter end. Sooner or later, it will be resolved one way or the other. Trenberth is just posturing, doing more of what he and his ilk have been doing for quite some time. It’s the old magical thinking: get a certain formulation of words to be acknowledged (at least by those who wield influence) as the status quo, and somehow that will prove something to be so. It’s more a matter of scoring points than anything else, and as such, exacerbates scepticism, thus being counterproductive.

    • D@mn WordPress… maybe Judith can fix my stupid typing. :(

    • GregP 5/17/11 at 11:49 am, Null hypothesis

      Some epistemologists define the Bayes hypothesis test as the decision method for applying subjective knowledge. But epistemologists are not even uniform in their definition of epistemology. To some, it is the study of all knowledge. To others, it is the study of demonstrable knowledge, turning epistemology into the study of objective knowledge, which in my book is science.

      Contrary to Wikipedia and its implications, the Bayes hypothesis test is one of the most elementary concepts in Systems Science, and not an iota of it is subjective. See Van Trees, H. L., Detection, estimation, and Modulation Theory, Part I, Section 2.2, Simple Binary Hypothesis Tests, esp. note Fig. 2.8, p. 36. That figure shows two, overlapping bell shaped curves of equal variance, separated in the mean. The one on the left is the conditional probability of an event under the null hypothesis, H0, and the one the right is the same under H1. This is a powerful concept, and its application is the subject of the second chapter. The point here with respect to the Wikipedia treatment is that the probability density under the two hypothesis are equal in importance: both are essential. A scientist does not assume H0 and pursue establishing H1.

      In some schools, literally for the most part, as distinguished from industrial science, the scientific method includes Popper’s falsifiability criterion, as prescribed in Wikipedia. Safe to say, this is never true in practice, not in academia, in science, nor with IPCC. It is an artifact of Popper’s objection to induction. He assumed that the scientific method involved induction by sampling, and as everyone knows, scientists and philosophers alike, that cannot reach a valid conclusion. Instead, science is strictly deductive. It never says all ravens are black, but that if you have a raven, the odds are it’s black. Ditto swans and white.

      The hard part about the null hypothesis that AGW doesn’t exist is getting anyone paid to be a climatologist to even test it. It’s not so difficult, though, requiring either one of just two bits of evidence. (1) ACO2 does not accumulate significantly in the atmosphere to be a cause, and (2) the entire instrument record of global warming is predictable from the Sun. Both have been established to a few. The null hypothesis exercise needs to concentrate on one or both of those two.

    • a false dilemma

  47. Michael Larkin and GregP, you are stating very clearly what the scientific situation is. However, unfortuately, our hostess cannot bring herself to state, unequivoically, that you are right. She will probaly ignore the obvious, and keep on suggesting that CAGW is right. How she will write her presentation, I have no idea, but I suspect she will take no account of what you have written.

  48. Robert Ayers

    I second the remark of Michael Larkin. If the as-stated formulation is correct, namely
    Given that global warming is “unequivocal”, to quote the 2007 IPCC report, the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence.
    there is a huge gap because “anthropologic” does not appear in the initial phrase.
    Global warming was “unequivocal” in 1930. And in 950. And in 10,000 BP.

  49. Craig Loehle

    The default position being taken as the climate models are true seems to rest on the “it is just physics” as some commenters here are fond of saying (e.g., M. Tobis). But the climate models are not “just physics” except in the most vague way. There is no experimental physics of clouds or laws of clouds or empirical relationship with a tight fit that enable one to quantify the light scattering of clouds nor how they come and go, where they are on Earth, or how many there are. None of that. And yet clouds are key to the entire dynamics. To take another example, albedo affects both reflection and also absorption of heat by the earth’s surface. Absorption warms the earth and heat is then lost by reradiation which occurs by a black body at the fourth power of temperature. But the models do not represent the earth by every square mm of land surface albedo, but by grid-scale averages—but note that the 4th power radiation from a mean albedo will almost never equal the radiation from the detailed actual surface. Similarly for evaporation. In all these cases, crude kludges (approximations) are used. Sorry, but this is not “physics” it is modeling. The GCMs are not a theory, but a calculation tool. They can not be “refuted” the way a theory can, nor “proved right”. They have degrees of usefulness depending on your purpose. We might say they have (or lack) sufficient reliability, skill, and lack of bias to make X decisions on, but they can not be considered “true” because too many of the computations are approximations of spatially heterogeneous processes or are fixes to numerical instabilities that result from grid-box discretizations. Further, the inputs to the models are very vague and approximate and the different models do not even use the same inputs (particularly for aerosols but also for volcanoes). The models differ in their outputs, but could not do so if they were “correct”. To the extent that they agree, they do so with different inputs and equations. It is stunning to me that Trenberth could make such an assertion about the null unless he does not understand all of the above.

  50. Craig Loehle

    To make another comparison, in physics we have various well-established theories, such as Maxwell’s equations or black-body radiation or something. These have been established by experiment with exquisite precision. Show me a single thing that the climate models get precisely right. Nothing. They are vaguely correct based on very generous eyeballing of their output (and only on some things). In what sense is this “proven”?

    • Rob Starkey

      Craig– Is their anyone who defends today’s GCM’s as accurate, or reasonably accurate in any meaningful way? I do not believe so. People can debate that details of how GCMs are built, but until someone develops a model that can accurately forecast rainfall and temperature at a regional level, then the models have very little practical value overall, and no value for policy makers.

      • Craig Loehle

        If GCMs are not accurate, it is hard to see how it could become the “null”. And it is the GCMs that lead to the “likely” diagnosis by IPCC. We accept many laws of physics and chemistry because they perform so precisely, not because they “seem right” or we like the output.

      • Bad Andrew

        “Is their anyone who defends today’s GCM’s as accurate, or reasonably accurate in any meaningful way?”

        Steven Mosher and every other Warmer that I know of.


      • Rob Starkey

        What Steve seems to “defend” is the process of creating GCM’s and that some of the results are within the expected margins of error at a “global level.” Unfortunately, the margin of error in these models is so high that it is possible to claim almost any result as being within the model’s expected outcome.

        I again suggest today’s global GCM’s are virtually without value. Policies are enacted and enforced here on planet earth by individual nation states and not by a global government. Unless and until we have reliable models that can accurately predict what the rainfall and temperature will be at the local or regional level as a function of specific CO2 levels the models are not of value to policy makers. Today, the models actually are so inaccurate in forecasting regional rainfall that they help to prove the case of those not believing in AGW.

        This issue is why I repeatedly ask what it is that makes people “believe” that a warmer world is necessarily bad for humanity overall, over the long term. I have researched, but have found zero reliable data to support the conclusion.

      • Bad Andrew

        “I again suggest today’s global GCM’s are virtually without value.”

        I totally agree. Now how do we get error spreaders like Steven Mosher to be more honest?


      • Bad Andrew,
        I argue with steven mosher quite a bit, but one thing I believe he is, is honest.
        I think we will be all better off in granting as many as possible in this credit for good faith and integrity, and save the condemnations for those who truly deserve it.
        My bet is the ‘crisis’ promoters are running out of steam and credibility quickly.
        Let’s try and plan ahead. Mosher has been a good guy in this from everything I can see.

      • Gotta agree with hunter –

        I don’t always agree with Steven Mosher, but I’ve never seen him be dishonest. You don’t have to like him or agree with him, but he deserves respect.

      • Bad Andrew

        Jim and Hunter,

        I have. He likes to try and tell people what their arguments should be, based on his esoteric “knowledge”. He has no more knowledge of why the climate behaves the way it does than anyone else, but he acts like he does. That’s dishonest, in my opinion.


      • Bad Andrew,
        This blog war has brought out less than the best in many of us.
        Steven may be prickly, but I would disagree that he is dishonest.
        He’s got the analytical tools to see through the ‘team’ smoke screen and give a credible description of what is there.
        That he believes AGW is significant is far different than the idea that it is *the* world crisis to deal with .
        I respect your take on this, but I respectfully also disagree.

  51. I am a bit confused about T’s claim, or claims. In his abstract he says “Given that global warming is unequivocal, the null hypothesis should be that all weather events are affected by global warming…” This may not be true but it makes certain amount of sense. I thought the issue of his talk was specifically attribution of extreme events to warming. To that extent, the issue of whether the warming is natural or anthro is not raised by the claim about the null hypothesis. Human influence has nothing to do with it.

    The quote at the beginning of this thread however, is entirely different. It refers to “human influence.” Moreover, it does not say influence on what, so it is meaningless as it stands. If he is claiming that the new null is human influence on extreme events, solely because the new null is that warming influences weather, then he has a huge non sequitur going. Simply put, the fact that it is warming (if it is) in no way makes human influence an established or null hypothesis. He has jumped the logic.

    So it seems like he started with something at least plausible, that warming changes weather, then made a huge jump into AGW being a null hypothesis. It does not work.

    As a logician I also have to point out that “null hypothesis” is here being used in a merely metaphorical sense. No wonder therefore that there is so much confusion, because metaphors are not science. “Null hypothesis” is a technical term is statistics. No one here is talking about actual mathematical statistical analysis so the technical concept does not apply. T is using a metaphor as though it were a scientific claim. Several people have made this point but not as technically as it should be made.

    • Craig Loehle

      If I say that David Wojick has an influence on human culture (no doubt true) that does not mean it is significant (no offense, David). In the same way, it is of course true that humans affect every aspect of the climate, but that does not mean the earth is going to burn up. Mixing qualitative and quantitative “effects on” is disengenuous.

      • No offense taken Craig (time will tell on my influence, ha ha).

        But it occurs to me that T is committed to AGW so for him the distinction between recent warming and human influence does not exist. In order to understand someone it helps to know what they believe (however strange). So what I see as a huge leap he sees as settled science. Perhaps this needs to be pointed out to him.

  52. Given that global warming is “unequivocal”, to quote the 2007 IPCC report, the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence.

    Did I miss something here, or did Trenberth try to pull a fast one by conflating GW with AGW, in addition to the obvious argument from authority? Unless there’s some missing context that changes this, it seems like an extremely clumsy argument, failing on some fairly basic logical grounds.

  53. Judith Curry

    You have asked for thoughts on the “null hypothesis” question. Here are mine.

    As you point out, Kevin Trenberth has suggested:

    Given that global warming is “unequivocal”, to quote the 2007 IPCC report, the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence.

    Before addressing this suggestion directly, I think it is worthwhile to look at how the traditional “scientific method” is practiced. I do not believe one should look for bypasses or shortcuts around the traditional “scientific method” in the name of post-normal or post-modern science.

    An essay “An Introduction to Science” discusses the application of the “scientific method” as follows:

    The scientific method is practiced within a context of scientific thinking, and scientific (and critical) thinking is based on three things: using empirical evidence (empiricism), practicing logical reasoning (rationalism), and possessing a skeptical attitude (skepticism) about presumed knowledge that leads to self-questioning, holding tentative conclusions, and being undogmatic (willingness to change one’s beliefs). These three ideas or principles are universal throughout science; without them, there would be no scientific or critical thinking.

    The scientific method involves four steps geared towards finding truth (with the role of models an important part of steps 2 and 3 below):

    1. Observation and description of a phenomenon or group of phenomena.
    2. Formulation of a hypothesis to explain the phenomena – usually in the form of a causal mechanism or a mathematical relation.
    3. Use of the hypothesis to quantitatively predict the results of new observations (or the existence of other related phenomena).
    4. Gathering empirical evidence and/or performance of reproducible experimental tests of the predictions by several independent experimenters and properly performed experiments, in order to either validate the hypothesis, including seeking out data to falsify the hypothesis and scientifically refuting all falsification attempts.

    How has this process been followed for AGW or, more specifically, for the premise that the observed warming since the modern record started has been caused principally by AGW and that this represents a serious potential threat to humanity and our environment (let’s call this the “dangerous AGW” hypothesis or dAGW, which Trenberth would like to see as the “null hypothesis”)?

    Step 1 – Warming and other symptoms have been observed.
    Step 2 – AGW (primarily from CO2) has been hypothesized to explain this warming.
    Step 3 – Models have been created based on the hypothesis plus interpretations of selected paleoclimate data and model simulations have estimated strongly positive feedbacks leading to forecasts of major future warming
    X Step 4 – The validation step has not yet been performed as there have been no empirical data, which would give robust support to the dAGW hypothesis, and there are many basic uncertainties. The two most important ones relate to a) the impact of natural climate variability and forcing factors and b) the sign and magnitude of the net overall feedback that could be expected to occur, which could either amplify or moderate the warming expected from a climate forcing.

    Confounding this dilemma is the observation that our planet has not warmed recently despite increasing levels of atmospheric CO2, thereby putting in question the dAGW hypothesis. To date there has been no satisfactory explanation for this recent lack of warming, nor has it been refuted scientifically.

    Until the validation step is successfully concluded, the dAGW premise remains an “uncorroborated hypothesis” in the scientific sense. If the above-mentioned recently observed falsifications cannot be scientifically refuted, it may even become a “falsified hypothesis”.

    So we are very far from declaring dAGW the “null hypothesis” as it has not yet been confirmed by empirical evidence from actual physical observation or experimentation, i.e. it has not been validated following the “scientific method” .

    Just because the IPCC, a political body, has declared in its last summary report

    Warming of the climate system is unequivocal

    does not mean that the this warming has been “unequivocally” caused principally by humans or that the dAGW hypothesis has been scientifically corroborated, and certainly does not mean that it should become the “null hypothesis”, scientifically speaking.

    It still has a very long and bumpy way to go before it deserves this status (if it ever will).

    These are my thoughts on this, Judith.


  54. Looking the thread over, it appears that many people are having the exact same reactions. When that happens, it’s usually a good bet that there’s something to those reactions, to wit,

    1) T pulled a switcheroo between GW and AGW,
    2) T appealed to authority,
    3) T created a straw man by saying no human influence.
    4) All of this to justify what is by Popperian logic unjustifiable.

    What else did I miss?

    • Very good and concise summary, ChE.

      Need one say more?

      This would be a good outline for Judith Curry’s testimony.


  55. Tony Hansen

    Can there be ‘any’ agreement on what the existing null really is?

    • Further to you point, Tony:
      1. Is there a specific Null Hypothesis (Ha0) that KT has nominated as falsified?

      2. Is there agreement among the Climate Community that Ha0 was a valid, non-trivial, realistic hypothesis and not a strawman?
      2.5 When was the Ha0 first published in a peer reviewed journal?

      3. Is there agreement that falsifying Ha0 was worth the world’s treasure? [“is that all you have to show for 7.5 years of work?”]

      4. Before we falsify Ha0, what new hypothesis are we replacing it with? People, this is a package deal. You must agree that Ha0 is falsified and that its replacement Hb0 is a valid, testable assumption. KT must not be allowed to define Hb0 by default.

      I see none of the above in evidence.

      • Tony Hansen

        Thanks for the reply
        1. I have never seen ‘exactly’ what the old one was. Or ‘exactly’ what the new one is for that matter.
        2. Until it can be defined how can anything be done with either the new or old
        3. It would be nice to see some work done on clearly describing what it was that they were trying to disprove
        4. I don’t need a new one especially.
        FWIW my guess was that the old was -‘natural cycles where stuff happened (don’t know why) and then after a while some/most/all of them reverted back ( don’t know why)’.

  56. son of mulder

    I can go with the null hypothesis if it is tracked with an index something along the lines as follows:-

    Each year calculate the ratio of the annual cost damage from climate related events like hurricanes/tornadoes/floods/droughts/freezes/forest fires to the annual cost of damage from non-climate natural events like earthquakes/tsunamis/volcanoes.

    The insurance industry probably already has had such data for many years. The change in gradient will give some indication of changing climate impact vs an ever present background of non-climate natural disaster.

    As population and wealth grow so this ratio will tend to cancel out such effects.

  57. Here is one more idle perspective on the subject of this thread.

    I’m familiar with WIREs Climate Change. The WIREs reviews are excellent – provocative, thoughtful, detailed, grounded in evidence, both empirical and theoretical – and non-polemical. I’ve read some of Mike Hulme’s writings, which impress me the same way.

    Unless informed otherwise, I would assume that the proposed Curry/Trenberth articles are intended to share the aforementioned characteristics. I doubt that WIREs is looking for a brawl – Trenberth vs Curry. Rather, I expect they want Judith Curry’s perspective and Kevin Trenberth’s perspective on the issue of how to frame a climate change null hypothesis (or hypotheses) and the implications of different choices. If this is correct, they are not expecting disagreement on every point, but rather shared views on some aspects and differences on others. I also expect that the readership would hope for that same outcome, and would be disappointed by articles designed mainly to fault the opinions of the other contributor, although they will certainly be interested in well-reasoned differences of opinion.

    Because the readership is well-informed, they are more likely to welcome a focus on the particular hypotheses in question than on theories of hypothesis testing, or the proper way to conduct science, both of which they probably already understand quite well.

    Well, that is my perception, and it may be quite wrong, but if Dr. Curry is uncertain about her assignment, she might wish to check with Mike Hulme or others to determine what they expect, and how much latitude she has to express herself as opposed to refuting somebody else. I would not proffer any of this advice unsolicited, but since Dr. Curry has asked us for our views, I offer this as something she might wish to consider.

    • Fred, why is it necessary for you or any of us to guess or assume what is the “Null Hypothesis” that is to be debated between KT and JC? Guessing nature is hard enough. Why should we stand for guessing what is supposed to be up for serious debate?

      I am astounded that after 200 comments, there does not yet appear to be a simple paragraph:
      Resolved: That Null Hypothesis XXXXXXX was successfully falsified. Further research and debate should now assume YYYYYYYY as the Null Hypothesis.

      Arguments for the resolution by KT and unnamed helpers.
      Arguments against the resolution by JC and unnamed, unsourced helpers.

      Dr. Curry, do you know if other people besides yourself were invited to argue the point?

      • Stephen – Based on my knowledge of what appears in WIREs, I don’t think they are seeking a “debate” in the sense of an argument, but rather a pair of contrasting perspectives on the issue. Because Judith Curry has solicited our views, that is mine. She will make up her own mind on what use to make of the suggestions from any of us.

      • Fred, I go by what I read:
        to invite you to develop a reasoned and lightly referenced argument for and against the proposition highlighted in the extract below – “the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence”. . . Both essays would be written ‘blind’ – i.e., without seeing the others’ essay –
        They do not state what the null hypothesis is.

        I note also that Using Kevin Trenberth’s speech … as the framing for these proposed essays is hardly neutral turf. Framing the question is as important as the answer.

        This game is rigged. Three Card Monty with KT dealing. Yet Dr. Curry chooses to play.

      • This game is rigged. Three Card Monty with KT dealing. Yet Dr. Curry chooses to play.

        If you don’t play, you can’t win.

      • Jim, Las Vegas must love it when you come to town.

        Remember the ending to the movie Wargames ?
        “The only winning move is not to play.”

      • Stephen –
        I don’t go to Las Vegas except to pass through to get to the mountains/desert. For a period of time (long ago) I ate only because I played poker. If I didn’t play – I didn’t eat. I didn’t miss many meals, but I have no need or desire to go back here.

        And if Dr Curry doesn’t play, then KT wins by default. I don’t think “not playing” is a real option here.

      • “….That Null Hypothesis XXXXXXX was successfully falsified. Further research and debate should now assume YYYYYYYY as the Null Hypothesis.”

        This runs fundamentally counter to my understanding of N0, and conflates it with “the consensus”. As Paul Dunmore has pointed out, no weight of evidence can either reverse the position of N0 & N1, nor provide grounds for “assuming” a “new” N0. The N0 for any particular N1 is inseparable from it, derives its metrication from it, is inapplicable to any other N1, and is eternal.

        Some N0 are of course so decisively disconfirmed that further testing of them is deemed futile and unnecessary, but that doesn’t mean that we should (or even can) “retire” them. And that certainly doesn’t apply to anything I’ve seen in climate “science”, beyond the rudiments, most of which have been around for about a century.

      • What if N0 is “Man does not cause climate change.” and N1 is not stated. JC loses, KT declares victory. We need an N1. Who writes it? It could be, “Man DOES cause Climate Change.” which will be twisted into “Man Causes All Climate Change” I can see the headlines in USAToday now. We need specificity on N0 and N1.

        Take for instance the G0 Hypothesis that General Relativity is False: Gravity does not affect light. You measure the position of a star above the limb of the moon during a solar eclipse. The star has shifted position, in accordance with GR. It is time to falsify G0 and replace it with G1. G1 must be specified a priori. G1 cannot be that “stars move.” Nor that Stars can move in a gravitational field. It must be that “Star positions will change by an amount as predicted by General Relativity.” Specificity matters.

        So far, this little KT framing exercise has been anything but specific.

      • “What if N0 is “Man does not cause climate change.” and N1 is not stated?”

        This is an impossibility. A null hypothesis cannot exist in the absence of its alternative – that’s why it’s called “alternative”, and not “optional”.

  58. Without reference to this particular issue, the Precautionary Principle doesn’t tell us, when you must abandon the tried and true the best option is its diametric opposite; it tells us when you must abandon the tried and true, the best option is generally the position most like the tried and true.

    If I develop a communication deficiency in English (which I imagine many could believe of me), I do not automatically endeavor to communicate solely in Mandarin, a language I have no familiarity with whatsoever.

    My career as a computer programmer fizzles as the programming languages and computers I work with become obsolete, I’m not going to be able to enter the field of brain surgery at the same level of competency on the argument that they’re just two different types of thinking machines.

    So why would anyone, moved off the former position by unequivocable facts take up the exact opposite position, rather than the next nearest reasonable position?

    How long did it take the discourse to form around the framework of the old viewpoint, and how much yet remains more relevant to its nearest viable alternative than to the completely alien contrary case?

    It’s probably more productive to all sides of the science to remain near the established paradigm, climbing the ladder of discovery one rung at a time, rather than skip over them all to that far end, abandoning the common idiom and starting all over from scratch.

    That said generally, many comments in this thread to give one pause to consider the topic in more detail.

  59. What is the specific text of the nominated Null Hypothesis that the debate is about?

    If the Null Hypothesis is reversed, then reversed to what statement?

    There were 100 uses of the word “Null”, now 103+, in this page and I did not see the specific statement. Sorry if I missed it.

    In terms of a general strategy for the debate, I suggest creating several parallel Null Hypotheses, replacing the word “man” with, “Sun”, “volcanoes”, “Algae”, “Earth’s Orbit”, “Urban Heat Islands” and “undersea volcanoes”. Challenge the proponents why these Null Hypotheses should not equally be reversed. Then ask, “What is the status of our science with all these reversed Null Hypotheses in play?”

    If “anthropomorphic” is used instead of “man”, then fight fire with fire and use “volcanogenic” and other twelve dollar words.

  60. Stephen Pruett

    As noted by Karl Popper, hypotheses cannot be conclusively proved, but they can be conclusively disproved. So, Trenberth’s position in which the null hypothesis is that there is AGW cannot be definitively proved (although an abundance of consistent evidence would be highly suggestive) but can be definitively disproved. On the other hand, if the null hypothesis is that warming is not primarily anthropogenic, this could not be conclusively proved but could be conclusively disproved (depending on the results of course). So, I would think Trenberth would prefer the null hypothesis that warming is not anthropogenic, since contradictory evidence would definitively disprove the null hypothesis, which seems to be his goal. The strangeness of openly preferring a particular outcome when beginning hypothesis testing is another issue.

  61. Jim Macdonald


    Your comments on the scientific method are right on the mark.
    About the only thing quantifiable by the AGW proponents are the predictions by the models. Without the absurdly wide margins of error, they clearly have been wrong, especially in the last 11 years. This being the case, they keep moving the goal posts, pushing their predictions of calamity
    further into the future.

  62. I think Judith is misquoting Trenberth slightly.
    The version quoted at the top is a non sequitur even by Trenberth’s low standards.

    The abstract of his paper says :
    “Given that global warming is unequivocal, the null hypothesis should be that all weather events are affected by global warming …”
    – nothing about human influence.

    In the main text of the paper he says
    “Given that global warming is “unequivocal”, and is “very likely” due to human activities to quote the 2007 IPCC report, the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence.”
    This makes slightly more sense than the version Judith wrote, but is still complete nonsense. Obviously, the burden of proof is on the IPCC and activists like Trenberth to justify their claim that the warming is ‘very likely’ due to human activities, which they have completely failed to do.

    • PaulM,
      Which is why Treberth and the team are so anxious to turn tables on the dispute.

    • KT is attempting to say that the (A)GW hypothesis has been proven. This ignores the fact that you can’t ‘prove’ an hypothesis, only disprove it. This is basically a restatement of the “science is settled” argument by Al Gore. No wonder they wanted a ‘better’ communication of their side, and this is just one example.

      The time is fast approaching when definite proof of a broken hockey stick must be admitted by all. The lack of warming in conjunction with rising CO2 is going to disprove KT’s null regardless of all the unequivocal claims by the IPCC.

      Plus, using a 2007 claim in a current paper while ignoring the last 4 years of temperature anomalies is more than a little questionable.

  63. I am starting to believe that anthropogenic CO2 is causing global warming. Luckily, the scientists who predicted the coming ice age in the 70’s were also correct, but the additional CO2 produced in the past forty years has saved us from the approaching cold. Yes, I’m being tongue-in-cheek, but can Treberth prove me wrong? A whole lot of people can believe something, but unless it can be proven, it is still only theory.

  64. Gary Mirada


    I think Trenberth is playing games. Personally I would not lower myself to his level. AR4 acknowledges that the level of scientific understanding on a number of climate components is low or very low. I think from memory that clouds fall into this category. It seems that we are starting to learn a bit more. Early indications from the CLOUD expriment hint at a cosmic ray effect as does this – . Untill we have a better understanding of how the climate works all this null hypothesis stuff is just hot air.

    kind regards


    • “Untill we have a better understanding of how the climate works all this null hypothesis stuff is just hot air.”

      Gary this is exactly wrong. How about – “Until Climate “science” gets clear about its hypotheses, and is prepared to find their null, rather than torturing the data until it gives in and allows them as positive finding, all this climate alarmism is just hot air.”

      Unfortunately this thread has wasted a lot of time (and hot air) looking for “The single Null Hypothesis for climate science” (there is no such thing – just a series of alternatives, each with its null), rather than looking at the way Climate “Science” has treated the null in its past experiments (mostly “Null what?”) and how it can be persuaded to deal properly with experimental evidence in future.

      • Gary Mirada


        cant say I disagre with you

        eloquence is not my strong suit

        and generally pushed for time these days to participate fully

        so I will leave it to you to keep up the good work

  65. Given the false science in Trenberth’s Global Energy Budget, I would have nothing to do with such a charlatan.

  66. Bad Andrew

    “I would have nothing to do with such a charlatan.”


    It’s my suspicion that once a person becomes a component of the machine, leaving or interfering with the machinery gets a little complicated.


    • Bad Andrew

      We could do a little experiment. Dr. Curry could start a “Judith Curry renounces AGW” thread, which just states that she now doubts that AGW is true.

      We would all observe what kind of responses there are, (here and perhaps elsewhere) and report back.

      My mental model (which I am always 95% certain is correct) says the results would be interesting, to say the least.


      • Rob Starkey

        I do not believe that would be a likely action by Dr Curry, because in fact humans are having an impact on warming the planet.
        What is really in dispute is the degree or rate of that warming, whether or not that warming is really bad for humanity, and what specific actions should be taken.

      • Rob,

        Of course it’s not a likely action by Dr. Curry. That’s not the point. The point is that it would be an interesting experiment.

        I mean, there are vast numbers of people who do not believe in Global Warming, and are free to say so and no one around them cares when they express that belief.

        It would be interesting to see what would happen when a person like Dr. Curry who is evidently not as free as others to express such a belief, does.


      • Rob Starkey

        Andrew– I would disagree that there are “vast number who have studied the issue and do not believe in Global Warming”.

        I do believe that there are vast numbers who do not believe that the issue warrants the proposed actions advocated by people like Hansen. I would guess that you would acknowledge that additional atmospheric CO2 will (all other things being constant) warm the world somewhat. The real question is how much (since all other things are not constant) and what that really means for humans. IMO- we do not really yet understand the relationship between additional CO2 and temperature, and more importantly do not understand the relationship between CO2 and regional rainfall.
        Overall, today; additional CO2 is more beneficial to humanity than harmful. Ask someone in a poor country who wants electricity what they think

      • Rob,

        You misquoted me. I didn’t anything about studying the issue. I’m talking about expression of beliefs. If you want to talk about something else, that’s fine, but I want to talk about what happens when people express their beliefs.


      • Rob Starkey

        Andrew- when people express beliefs based upon either ignorance or stupidity what is the appropriate response?

      • Rob,

        It depends.

        But I’m not interested in the philosophical questions. I’m interested in a specific experiment being conducted and observing what the results are.


  67. Richard Klem

    It’s not for nothing that it’s called the “null” hypothesis. For example, according to UC Berkeley: “The null hypothesis is typically that something is not present, that there is no effect, or that there is no difference between treatment and control.”
    Because a null hypothesis is merely the absence of the relationship claimed in the hypothesis, anything that makes a claim about the state of the world is a hypothesis not a null hypothesis. This rather limits the choice of null hypotheses, which is a good thing. Otherwise, a “scientist” could choose a null hypothesis that was so off base that it could be rejected when compared to practically any hypothesis. Trenberth’s assertion that, in the past, the null hypothesis has been that “there is no human influence on climate” may be an example of such an attempt.
    Trenberth assertion shows a lack of understanding of statistics. He seems to think that there are only two possibilities, namely, humans do influence climate or they do not. This may make sense in logic, but no statistical test can be that precise. Instead, the most that can be measured is the probability that the hypothesis being tested is true in the universe from which the data is sampled. If that probability is great, e.g. more than 95%, the null hypothesis is rejected and the hypothesis is supported. Trenberth seems to be attempting to lower the bar so that it is easier to support global warming hypotheses.
    Trenberth’s new proposal would lower the bar still further. By claiming that global warming hypotheses are actually null hypotheses, he would have us flip the probabilities. For example, instead of having to show a 95% probability that “global mean temperature had been stable for thousands of years”, he would only need to show a probability of greater than 5%.
    Comparing recent temperature changes to ‘natural variation’ is another problem. It just adds the problem of determining what is natural. I cannot think of any similar thing in science. As suggested by the UC Berkley cite above, normally the question is, what is the effect of something, e.g. does gravity curve light or does an increase in CO2 cause an increase in global mean temperature?

    • “The null hypothesis is typically that something is not present, that there is no effect, or that there is no difference between treatment and control.”

      In other words, as I observed earlier, that a null hypothesis must have the quality of nullity. And yes, PDA, it is a presumption of ignorance – or, better – of “nothing known”. That’s its point. Furthermore, the null requires, in principle, no metrication, since it will inherit its parameters entirely from its Alternative.

      • as I observed earlier, that a null hypothesis must have the quality of nullity. And yes, PDA, it is a presumption of ignorance – or, better – of “nothing known”. That’s its point.

        Please provide some support for this, other than mere assertion. It doesn’t comport with any meaningful definition of ‘hypothesis’ I’ve ever been exposed to.

      • PDA do you read before you shoot? I was RESPONDING to a reference in Richard’s comment. Since you can’t be bothered, here it is

      • Sorry, I still don’t get how “something is not present, that there is no effect, or that there is no difference between treatment and control” equates to ignorance.

        These are positive, testable propositions.

      • OK PDA, what does failure to disconfirm that:

        ““something is not present, that there is no effect, or that there is no difference between treatment and control” ” tell us? How, precisely – and remembering that you have not proven the null, merely failed to disconfirm it – does it diminish our ignorance about the relationship between phenomena studied?

        Paul Dunmore earlier gave an instance (sensitivity to 2XCO2 is/is not 3.5C). But that is a hypothesis about a relationship whose existence is not in dispute (in his instance), and is merely an attempt to measure it. It is a hypothesis about the extent, not the existence, of a relationship. In the case of the latter, I concur with UC Berkeley’s prescription.

      • This is getting circular. It’s best if discussions on a particular point are kept in their respective threads.

        In the thread you referred to above, I was talking about the quality of an hypothesis, null or otherwise. You have invented the quality of “nullity” and ascribed it to H(0), but “we don’t know” cannot be a null hypothesis.

        Now, a given experiment can certainly have an ambiguous result, and can (often does) have no effect on our ignorance of a phenomenon. But before one can say that they have “failed to disconfirm” an hypothesis, one needs to define that hypothesis rigorously, in a manner amenable to testing. As Paul Dunsmore pointed out, H(0) can in many cases be more than just the inverse of H(1).

      • The fact that you have not seen the word “nullity” before doesn’t mean that I invented it. Most people would see that a term which has the descriptor “null” should have the quality of nullity, but as you keep reminding us this is climate “science”.

        But since you are new to the term, it’s not surprising that you misunderstand it, and keep insisting that it equates to “we don’t know”. It doesn’t, but I don’t have the patience for any further attempts to explain to you why not.

      • Most people would see that a term which has the descriptor “null” should have the quality of nullity,

        Show me any use of the word ‘nullity’ in relation to ‘null hypothesis’ and I will gladly retract what I wrote above, and thank you for enlightening me.

        you misunderstand it, and keep insisting that it equates to “we don’t know”.

        I make no such insistence. You explained it to me by writing:

        a null hypothesis must have the quality of nullity. And yes, PDA, it is a presumption of ignorance – or, better – of “nothing known”

        If you can explain to me how it is a presumption of ignorance – or, better – of “nothing known” somehow means something different than “we don’t know,” I’d appreciate it.

      • Ok I’ll try one more time.

        “We do not know” is an ASSERTION of ignorance.

        “We PRESUME we do not know” is a quite different statement, and the logical premise from which all experiment must proceed.

        Presumption/assertion – different – OK?

      • I think Fischer used the term “null” to indicate that this was the hypothesis to be “nullified”.

      • Sorry, I realised when I looked at it “Fisher”

      • HAS “I think Fischer used the term “null” to indicate that this was the hypothesis to be “nullified”.” This is a very obscure way to use language. Why do think he meant this? Did he say so? The natural way to read “null hypothesis” is that it is a hypothesis described by the adjective “null”, particularly when it is paired with “alternative hypothesis”. Hence my insistence that it must have the quality of nullity.

        For what it’s worth, my SOED gives several definitions for nullity, of which the third is my usage: “the condition of being null or nought; state of nothingness”.

        While I was at it I looked at the adjectival (it can apparently exist as a verb, although clearly not in our case) definitions of “null” – and found – “Amounting to nothing, non-existent”. This is my understanding of Fisher’s usage.

      • TomFP May 21, 2011 at 12:21 am

        You forced me to do Google on [Fisher “null hypothesis” nullify] and turned up a number of references (including Wikipedia) but this one gives some explicit references to Fisher’s work and is a useful discussion of some of the issues being discussed on this thread

      • It seems quite similar to the law of parsimony

      • thanks HAS and it appears that Fisher construed his term in a very specialised way. Perhaps if he had anticipated its widespread coinage he might have taken care to choose a formulation which did not invite (as he implicitly admits) etymological confusion. Nonetheless, the usage I inferred seems to have found wide favour, and puts me in good company, not least with Truran himself, apparently, for much of his career, hence the “corrective” character of his paper.

        And the usage Truran complains of doesn’t seem inherently incompatible with Fisher’s, as he reports it. For instance he describes as “bizarre” the formulation “we must retain H0 until we gather further evidence and can draw a conclusion”. This seems a bit strong. He makes it clear that his concern is to avoid a form of Type1 error:

        “…seems to assume that an alternative hypothesis will eventually be accepted: it is only a matter of having enough evidence.” Quite so – climate “scientists” N.B.

        He then goes on to show that all that would be necessary (and it emphatically is necessary) is to substitute “unless” for “until” in the offending statement for the proper presumption of ignorance to be restored.

        And as Teddy implies, isn’t a further objection to the choice of a null which assumes the existence of a causative relationship that it violates the principle (I’m not sure it’s a law) of parsimony? At the outset of an experiment, are we not bound (if we accept Occam) to give logical priority to the statement which contains the fewest assumptions?

      • There appears to be confusion about the “null hypothesis” concept. A null hypothesis is typically of the “no effect” type, but it needn’t be. It is basically the hypothesis judged best to assume true unless it can be shown to be highly improbable – in other words, when its truth or falsity remains uncertain, it is wiser to assume it to be true than false.

        Clearly in the case of anthropogenic climate change, the judgment as to whether it is wiser to assume we are significantly affecting the climate adversely, or that we are not, involves political as well as scientific assessments. If the null is that human effects are real and significant, and if that hypothesis cannot be falsified, the implication is that we should make a serious effort to blunt the magnitude and impact of our greenhouse gas emissions. If the null is that human effects are insignificant, failure to falsify that null could conversely be used to justify a decision to refrain from major mitigation or adaptation efforts.

        Elsewhere in this thread, I have predicted that Kevin Trenberth will not focus on the definition of a null hypothesis, but will instead cite climate data as evidence that it is wiser to assume significant human effects, unless proven otherwise, than to assume the lack of these effects. I believe that an informative pair of papers from Drs. Trenberth and Curry would address that issue, including both the nature of the data and the confidence they deserve, rather than semantic quibbles about the meaning of “null”.

      • There is also a Bayesian element to this issue. Readers might want to revisit the thread on Probabilistic Estimates of Climate Sensitivity for a flavor of this element, since a cutoff value of climate sensitivity might be proposed as a criterion for significant human effects (e.g., at least 1.5 C/CO2 doubling). In most cases of hypothesis testing within science, a P value can’t be translated into the probability that a hypothesis is true, because the prior probability of its truth is unknown and a “principle of indifference” that assigns it a value of 0.5 would almost never be warranted. In the case of climate data, one could try to use existing data to assign a prior probability or a PDF to the truth of an anthropogenic effect hypothesis, and then reason that the posterior probability derivable from further evidence should utilize that prior in arriving at a conclusion. This would certainly lead to arguments about the prior data, and would hardly settle the issue, but it might also, in my view, put the evaluation of new evidence on a slightly firmer basis than an assumption that we knew nothing about climate influences until the latest study was reported. I expect Trenberth would argue that current evidence justifies a fairly high prior probability.

      • David L. Hagen


        Clearly in the case of anthropogenic climate change, . . . if that hypothesis cannot be falsified, the implication is that we should make a serious effort to blunt the magnitude and impact of our greenhouse gas emissions.

        That appears to be the politically correct answer – but on what basis?
        Cost/benefit analyses appear to show that that is the LAST thing we want to do? See Copenhagen Consensus, 2008.
        Why should we pursue that illogical irrational route?
        That appears to assume that earth was previously in pristine “equilibrium” and that we must maintain that state at all costs?
        What is the evidence for that? Temperatures appear -8C to 2C.

        Mitigation appears to dismiss all humanitarian projects and require that we “control climate” at unimaginable costs?

        Why should we pay US$1,900 trillion dollars per degree to control climate?

        Why require mitigation without addressing adaptation?

      • David – You quoted only part of my comment, but I cited both sides of that argument in order to suggest that it is the evidence on that point that would dictate the choice of the most prudent hypotheses to evaluate. I don’t happen to agree with your conclusions, but the purpose of my comment was not to give my perspective on the cost/benefit calculation for climate change mitigation.

        Regarding your last sentence, I know of no-one who doesn’t believe in the need to address adaptation as part of our approach to climate change.

      • Fred –
        Regarding your last sentence, I know of no-one who doesn’t believe in the need to address adaptation as part of our approach to climate change.

        I do. There are those for whom mitigation is the only answer because they do not consider adaptation to be a definitive enough step.

      • We must travel in different circles, Jim. More seriously, mitigation and adaptation are synergistic, because adaptation is most efficient when what it adapts to is not changing so rapidly that what is adaptive today will be maladaptive ten years hence. If mitigation slows the pace of undesired changes, they will be easier and less expensive to adapt to.

      • Please name one mitigation strategy that has worked.

      • Mitigation and adaption are both synergistic and competing for scarce resources and capital .

        In the case of Climate Change, the Mitigation by law is so expensive, that adaption crowed out. In a free market, adaption would be the preferred method by capital and labor efficiency. In a facist marketplace, political currency holds sway and mitigation is forced upon the subjects.

      • Fred –
        We definitely travel in different circles. For reasons I won’t explain here, I sometimes travel with people whose political philosphy is light years away from mine. And in some cases, yours. Won’t claim that most of them are friends – or that they have any common sense – or that they know enough about science to even spell the word. But they DO have some definite hardcore opinions that have been spoon fed to them by the environmental organizations among others. And total decarbonization RIGHT NOW is a popular theme even though they have no idea what it would mean – to them or anyone else.

      • Fred,
        While waiting on you to find a mitigation strategy success story, please consider also showing us any evidence that adaptation will be more difficult in the future than it has been in the past.

      • I think people here need to familiarize themselves with fuzzy logic:

        This problem can’t be properly stated in Boolean logic, because it’s a question of attribution, but can be properly stated in fuzzy logic. Of course, once you’ve reformulated the question in fuzzy logic, the answer becomes a question of shares of gray.

        And BTW, to be perfectly clear, the fuzzy logic formulation isn’t a question of probability, but a certain question of attribution. This isn’t a question of Baysian probability, it’s a much simpler question of % natural and % anthropogenic.

      • Exactly.

      • Only problem is, of course, that there’s no proof (or falsification) in fuzzy logic, only shades of attribution.

        Lawyers figured this out a long time ago, which is why there’s guilty and not guilty, and nothing in between. The fuzzy logic gets applied in the sentencing (extenuating circumstances, etc.), but you’re either guilty or not. Partial guilt would create a mess, even though that’s more or less what extenuating circumstances amounts to. Many academic legal papers have been written about this.

      • Saying that a null hypothesis “presumes ignorance” might be a bit awkward, since the target of this kind of statistical inference is usually reality, not belief. When an hypothesis gets rejected, it’s because there we found no fact of the matter to sustain it.

        Of course, we can then say that we know a bit more. But we do not need to know that much more: in a frequentist setting the explanatory power of hypothesis testing is quite minimal, for the null hypothesis merely orients the burden of proof, or refutation if we are to take seriously the myth of falsification. In this setting, it sounds more appropriate to talk about the presumption of inexistence, i.e. “there is no relationship between such and such”.

        If we can equate inexistence with ignorance, we might hold the solution to a very old quandary in the interpretation of probabilities.

        As far as I know, the expressions “presumption of knowledge” and “presumption of ignorance” have been used by Peter Strawson (1964), to designate what information considered known and unknown in a pragmatic situation:

      • 1 “reality, not belief”?
        What does “the presumption of ignorance” have to do with belief? I think you are confusing it with “agnosticism”.

        2 The example you link to relates to presuming ignorance in others regarding a message the subject wishes to impart. The presumption that must be in the mind of the experimental scientist is of his OWN ignorance, a well as everybody else’s. It is, of course, an artificial state of mind, adopted for experimental purposes, applies only to the phenomena being studied, and is not one that comes naturally to humans. Which is why we’re not all scientists, and why even fewer of us are good scientists.

      • To believe, to ignore and to presume belong to some kind of epistemic mode, in one way or another. Presuming is akin to taking for granted, and thus akin to believing. This is no only plausible in our mundane way of seeing things (notwithstanding the knowledge paradox) but also has technical currency, for instance in the realms of belief revision, abductive reasoning, and belief function, the latter being an interesting approach to statistical inference:

        The concept of abductive reasoning might be the most natural way to interpret a null hypothesis as a “presumption of ignorance” into a logical framework:

        Considering that a null hypothesis is usually couched in a frequentist setting, I find it not really surprising that eyebrows are reaised when we try to explain it by turning to the epistemic mode.

  68. maksimovich

    One of unfortunate arguments is that KT has used is this is applicable to extreme events.

    In a significantly powerful review paper (Ghil et al 2011) the arguments as posed in a binary way, are not as simple as he thinks nor ameanable to archaic methodology that has been to justify a number of spurious arguments eg Lenton.

    All science is subject to revision ,and it might be approriate for a number of the “schools” of thought such as the IPCC to bring there understanding of mathematical physics at least into the later part of the 20th Century,

    Extreme events: dynamics, statistics and prediction

    From the introduction.

    Extreme events are a key manifestation of complex systems, in both the natural and human world.Their economic and social consequences are a matter of enormous concern. Much of science, though, has concentrated until two or three decades ago — on understanding the mean behavior of physical, biological, environmental or social systems and their “normal” variability. Extreme events, due to their rarity, have been hard to study and even harder to predict.

    Recently, a considerable literature has been devoted to these events; see, for instance, Katz et al. (2002), Smith (2004), Sornette (2004), Albeverio et al. (2005) and references therein. Still, much of this literature has relied heavily on classical extreme value theory (EVT) or on studying frequency size distributions with a heavy-tailed character. Prediction has therefore been limited, by-and-large, to the expected time for the occurrence of an event above a certain size.

    This review paper does not set out to cover the extensive knowledge accumulated recently on the theory and applications of extreme events: the task would have required at least a book, which would rapidly be overtaken by ongoing research. Instead, this paper is trying to summarize and examine critically the numerous results of a project on “Extreme Events: Causes and Consequences (E2-C2)” that brought together, over more than three years, 70–80 researchers belonging to 17 institutions in nine countries. The project produced well over 100 research papers in the refereed literature and providing some perspective on all this work might have therefore some merit.

    • Thanks for this link Max.

      It will take me a bit of time to get my head around it – and perhaps a little longer for a more general recognition of radical paradigm change that is central to the application of chaos theory to weather and climate.

      Does it provide a new methodology for predicting extreme events and is this materially different to the power series distributions (typically a log-Pearson type 3) commonly used in hydrology? Could be interesting if our 10,000 year forecasts for dam spillways are very wrong.

      Although I would disagree with you on Lenton et al 2008 – whose paper includes a reasonable listing of feedbacks potentially leading to chaotic bifurcation but with a content that is very much only in principle. Although there are extreme limitations on what is known about potential tipping points the new paradigm does provide an opportunity to reframe the climate wars – clearly that nearly everyone has been dead wrong about the essential nature of Earth systems using simple causality rather then dynamical complexity.

      But the essence of the Ghil paper revolves around extreme events that occur in the vicinity of tipping points, noisy bifurcation, catastrophe in the sense of René Thom or dragon-kings in the sense of Sornette. Both Lenton and Ghil posit the same dynamic processes in climate. As I say – it is the new climate paradigm and one that does not allow anyone to wriggle out of the potential (or the risk) for abrupt climate change as a result of anthropogenic greenhouse gas emissions. The appropriate policy response is I believe as we have agreed on previously.

      The real problem with Lenton et al 2008 is that it focuses too exclusively on greenhouse gases as control variables in chaotic Earth systems. It puts a whole new light on the null hypothesis. Earth systems are not chaotic? Not terribly likely.

      ‘It is often known, from modelling studies, that a certain mode of climate tipping (of the oceanic thermohaline circulation, for example) is governed by an underlying fold bifurcation. For such a case we present a scheme of analysis that determines the best stochastic fit to the existing data. This provides the evolution rate of the effective control parameter, the variation of the stability coefficient, the path itself and its tipping point. By assessing the actual effective level of noise in the available time series, we are then able to make probability estimates of the time of tipping’ Thompson and Sieber (2010)

      A number of other sub-systems with multiple states – commonly known as oscillations – exist in a synchonised planetary system (how could it be otherwise) with multiple control parameters.


    • Exceedingly cool!

      Bonus almost a dozen pages of references!

      I now have my weekend reading taken care of.

      Much better than any more droning on about the dull hypothesis, or wasting time with another pirate sequel.

    • bingo on spotting the ghil paper, i’ll do a post on this, looks very good

      • Judith Curry

        A thread on the Ghil study and what it means to the suggested “null hypothesis” that AGW will cause an increase in extreme weather events would be interesting.

        I like the authors’ concluding remarks:

        the yes-or-no answer to the “Great-Gatsby” question is a definite “maybe”, i.e. “it depends”


        one person’s signal is another person’s noise

        Does not sound like a firm basis for a new “null hypothesis”.


  69. I recommend that CE participants have a look at Trenberth’s full statement:

    It’s only six pages long and KT hits most of the talking points and conceits of the climate change movement: the hacked emails, “deniers”, the complicit media, the importance of avoiding scientific debate plus amusing self-serving cartoons at the end and a loving dedication to friend and colleague, Stephen Schneider.

    This is a scientist propagandizing in activist the mode and tradition of Dr. Schneider.

    • All of which begs the question if Dr T is OK with lying for science like like his late pal Dr. Schneider?

      • The fact that Trenberth refers to the Climategate emails as “hacked” and the media as “complicit” demonstrates that Trenberth is opting to be “effective” over being “honest” — ah, the terrible horns of the Schneiderian Dilemma!

        Would any of the orthodox care to explain why scientists like Dr. Trenberth should be trusted?

      • hunter

        You ask

        Is Dr. T OK with lying for science like his late pal Dr. Schneider?

        You could have added Dr. James E. Hansen, another ex-scientist turned activist.


  70. It’s the usual Orwellian technique, in this case redefine the scientific method to avoid the burden of evidence. Typical among agw mongers.

  71. Clearly from the Trenberth statement, his implied null hypothesis currently is that there is no human influence on global warming.
    That is, despite adding 40% to CO2 it has had no warming influence.
    That is, CO2 has no warming influence per se.
    Is this a viable hypothesis, is the question. As I argued above, no credible scientist holds this view, making it worthless as a hypothesis, something like the hypothesis that the moon is made of cheese. A worthwhile hypothesis is one that has not already been rejected.

    • Whether the CO2 increase is due to human emissions is in fact debated. Human emissions are a tiny fraction of the annual CO2 flux so it is not a trivial issue. So once again there is a conflation of hypotheses here. Is it warming? Is the warming due to the GHG increase, especially CO2? Is the CO2 increase due to human emissions? Are the extreme events due to the warming? I don’t think there is a null hypothesis in this mix.

      • In the materials I usually read, the question of the dominant anthropogenic contribution to atmospheric accumulation is considered well established on the basis of isotopic measurements. What do you consider to be reputable literature discussing weaknesses in this particular assertion? (or hypothesis, if you will?)


      • Rob Starkey

        Paul– please reference the isotopic measurements you reference. Are you referencing C12 and C13??? Care to get into how and when these measurements have been done and the margin of error?

        I am not saying that humanity is not putting vast amounts of CO2 into the atmosphere, but you will not be able to demonstrate it very well (imo) based upon isotopic data.

      • Rob – I’m not sure why you think the human origin of post-industrial CO2 increases is a subject of much uncertainty, since the conclusion resides in the convergence of multiple lines of evidence that include measurements of C14, C13, C12, atmospheric oxygen, volcanic activity, and records of industrial emissions. For the conclusion to be seriously in error requires all of these independent sources to be seriously wrong in the same direction. Do you have literature references for a different conclusion that supports a major alternative major source for the rise in atmospheric CO2 of about 40 percent?

        Some data and multiple references regarding the C14 evidence are provided by CDIAC at Atmospheric Carbon-14 Data. A general discussion of the convergence of evidence is given by Richard Alley in his video presentation – see Richard Alley – Segment 6>.

      • Fred
        Please reread what I wrote. I did not question that humans are “a” or even “the predominate” source of additional atmospheric CO2. I questioned the abiliy to prove that conclusion using isotopic data. C14 has been an unreliable measure since the late 40’s and measures using C12 and C13 have large margins of error. It is not difficult to “estimate” human emissions and that is why I am not doubting that humanity has contributed to an increase in atmospheric CO2. I challenge you to demonstrate the math and the source of the data to show what percentage of atmospheric CO2 was due to humanity in 2010.

      • Rob – It’s not a question of whether human contributions are the “predominant source” – that’s essentially an understatement. If you read the references and sources I’ve cited as well as the review cited by Pat Cassen, I think it will be clear that for the almost exclusively human contribution to be refuted would require more than measurement uncertainties in C12/C13 ratios (although Pat Cassen’s reference addresses the methodology quite well). It would require a quantitatively significant, multidecadal alternative source of rising atmospheric CO2 that could be reconciled with all the data on each isotope as well as atmospheric oxygen, volcanic eruptions, and industrial emissions. While an unidentified source of this type could I suppose be considered a theoretical possibility, it is not only implausible in the absence of evidence, but also unnecessary in the presence of strong evidence for the human role at the extent deducible from the multiple lines of evidence available. If, however, you do have a physically plausible alternative in mind, you should identify it, along with evidence for its existence. Otherwise, this seems to be a sterile argument.

        I do, however, recognize the point of your suggestion that year to year fluctuations in the rate of CO2 change can involve flux changes between atmospheric and terrestrial or oceanic sources that add to or subtract from the human contribution. These fluctuations do not appear to be huge. More importantly, however, the multiyear and multidecadal trends can only be plausibly explained by the human contribution. If the question is whether the year 2010 increase was exclusively human, the answer is that it may not have been, but was probably close. If the question is whether the 40 percent rise since preindustrial days has been exclusively human, for practical purposes the answer is yes.

      • I should qualify my own comment on the 40 percent increase. As far as we know, the “airborne fraction” (percentage of emitted CO2 remaining atmospheric) has not been changing greatly over the past century, and if any change is occurring, the fraction is perhaps increasing very slightly due to greater saturation of the oceanic sink. However, I’m not sure we can exclude a small contribution to the value of the airborne fraction from solar warming during the earlier parts of the twentieth century that reduced the solubility of CO2 in seawater. This would be a non-anthropogenic contribution to atmospheric CO2. Recent solar changes have probably operated in the other direction, but again are likely to have affected the airborne fraction only minimally, leaving the balance due almost entirely to anthropogenic contributions.

      • Fred–Please show that I am mistaken by providing a link to any data/calculations that show what % of the total CO2 content was due to humans in any given year. If the data was that reliable via isotopic data it would be readily tracked and published. I have gone through the assumptions and calculations using C12 and C13 and I simply ask….what was the margin of error in the conclusion?

        Although the base issue of humans impacting the climate does not change, I challenge you to produce the data I asked for. From what I can tell it is one of the many myths about the “good data” on the topic of human caused climate. Please no more generalities on the topic and claim that it should be believed

      • Rob – I believe your question has been answered. For any individual year, we can’t precisely identify the human contribution – we do agree on that. Over the course of multiple decades, we can accurately state that the human contribution has accounted for all or nearly all the observed rise. This is based directly on multiple lines of evidence, and indirectly on the absence of a plausible alternative that I’m aware of. If you can identify an alternative compatible with the multiple lines of evidence, we should reconsider.

      • Pat Cassen

        Rob – A nice review of how carbon isotopes are used to identify the components of the carbon cycle, including the anthropogenic contribution, can be found here

        The 14C identification of the fossil fuel contribution was detected by Suess (Science, 122, 415) in 1955, and verified many times since.

      • Pat- I am familar with the basic concepts and have looked at the math. Ask someone using C12 / C13 to show you the data they used to determine what they believe the human contribution %was to total atmospheric CO2 for the last (X) years. Then ask what the margin of error is for each year’s calculation. From what I can tell you really can not use isotopic data to determine this accurately at all.

      • Pat Cassen

        Rob – You are right: determining the precise anthro contribution by 12C/13C alone is only marginally possible. Overall budgetary considerations are currently more quantitavely restrictive, but the isotopic data can be used to rule out other ‘unidentified’ sources. These matters are discussed in the review paper I cited, and references therein.

      • Rob – well, that’s actually an interesting question. But what’s the bigger picture here? I recall (from a course I took nearly 10 years ago from Inez Fung) that isotopic evidence was key to establishing the anthropomorphic fraction; Fred reports lots of additional lines of evidence. I took a look at the IPCC’s AR4 briefly and didn’t find the direct citations, so it will take me a little more work to get up to date.

        But – to return to the “null hypothesis” question, sort of: I asked David for citations that question the conclusion that anthropogenic emissions are responsible for the rise in atmospheric CO2 concentrations. You don’t offer any such citations either, just an argument that C13 and C14 measurements wouldn’t be sufficient to reject the null that “Anthropogenic emissions are not the primary source […]”. But this seems largely irrelevant; I think that the role of anthropogenic emissions in atmospheric buildup is one of the sub-questions in which “burden of proof” can clearly be shown to lie with the “skeptic” position. Do you hold otherwise?


      • Bad Andrew

        The burden of proof always lies with the party making the claim. Always.


      • Paul–I am not questioning that humans have caused an increase in atmospheric CO2. I am saying that you can not use isotopic data to demonstrate what % of total atmospheric CO2 is due to humans with enough accuracy for the measure to be useful. You brought up the topic. I think you have to agree that what I stated is accurate

      • As I understand it the CO2 flux is on the order of 150-200 bty. Anthro emissions are reaching 8 bty. The annual increase is around 3 bty. If someone is claiming that the annual increase consists of anthro emissions it is news to me and I would like to see that claim.

        The usual AGW claim is that the huge non-anthro flux is somehow magically in equillibrium so the tiny anthro addition is causing the increase (minus the mysterious missing sink, a fabrication of AGW). Isotopic measurements are basically irrelevant to this causal claim. The most they can show is that anthro emissions exist and are increasing, which no one denies.

      • David– good summary. The idea that “non human” factors that contribute to CO2 changes remain relatively constant has been shown to be an incorrect assumption. It has been learned that soil bacteria release significantly different amounts of CO2 due to varying stimulation.

      • Rob – I’m not sure the magnitude of the human contribution is controversial enough to demand extensive discussion in its own right, but I’m intrigued by your arguments nonetheless. First, though, I would ask you to cite specific references and quantitative data to support them, rather than merely assertions.

        Second, regarding soil bacteria, I wonder whether you are confusing “human activity” with “products of fossil fuel combustion”. Human activity, including changes induced by fossil fuel CO2 in temperature and soil moisture, can then further influence the biosphere in ways that affect CO2, but it’s still a human contribution. Here’s my question: Let’s assume (improbably) that soil bacterial quantity and activity has changed markedly for reasons completely unrelated to anthropogenic climate change. Can you describe very specific changes in bacterial mass and metabolism that over the course of a century would not only significantly increase atmospheric CO2 concentration but also increase the ratios of C12 to C13 and C14 in the manner that has been observed (we’ll leave out the bookkeeping from industrial records that also must be accounted for)? Please describe what is being metabolized (the carbon source) and how the input and output would change the isotopic ratios. Numerical examples that close the metabolic budget would be helpful. We can then analyze how much this is likely to affect conclusions about non-human contributions on a long term basis as opposed to year to year fluctuations, where we agree many factors can be involved.

      • Rob Starkey

        Fred wrote –“ we can accurately state that the human contribution has accounted for all or nearly all the observed rise.”

        My response- This (imo) is a pretty good example of the thoughts and positions of those supporting the argument that increased atmospheric CO2 is a dire problem. They often seem to take weak science and then claim that it is really solid until confronted with facts.

        Fred’s statement is actually scientifically WRONG. What we can actually determine is that over the last 50 years:
        a. humans are emitting large amounts of CO2
        b. CO2 has gone up
        c. It is highly likely that human CO2 emissions were the leading cause of the overall CO2 rise as there are not other sources that we have identified as having increased their CO2 emissions in the amounts required to make the overall rise.

        We can not however claim with any certainty as Fred states as fact that all or nearly the entire observed rise was due to humans. We have inferred that, but not proved that.

        Fred then wrote– “This is based directly on multiple lines of evidence, and indirectly on the absence of a plausible alternative that I’m aware of.

        My response- I generally agree. Humans are probably responsible for most of the CO2 rise based on the observations of the CO2 we are emitting, but to make the claim that Fred has made would require the ability to make accurate measurements of total CO2 at any point in time and then to accurately determine what percentage of that total CO2 is of human origin. That folks we can not accomplish at all and people who claim it can be done based upon isotopic measurements are in error.

        In response to the increase in CO2 emissions from soil-

        I don’t like the source, but it was handy and I was to lazy to pull other data

      • Rob, David – Notwithstanding the limitations in the precision of isotopic measurements, no source sufficient to explain the increasing CO2 trend has been identified that produces the observed:
        (1) decreasing O2/N2
        (2) decreasing 14C/12C
        (3) decreasing 13C/12C
        (4) increasing pH of the oceans

        Indeed, elementary budgetary explanations that invoke fossil fuel burning (and deforestation) readily account for these observations. Moreover, if the observations were due to an imbalance in the natural sources and sinks, it would be a remarkable coincidence that such imbalances occurred at precisely the time that anthroprogenic sources were of (more than) sufficient magnitude to also account for them.

        But, I suppose, it never hurts to entertain other hypotheses.

      • “increasing pH of the oceans”? I thought they were supposed to be “acidifying”?

      • Tom – Pat’s analysis is one I agree with, but in case at this late hour, he doesn’t respond immediately, I expect he will appreciate my inferring that he intended to say “increasing ocean acidity” rather than “increasing pH”.

      • Thanks, Pat.

        Rob: In fact if you had read my original post carefully, I didn’t make any specific claim about WHICH isotope data was the basis for this conclusion; I was referring to a class I took a decade ago, and browsing AR4 a few years ago. The focus on C13/C14 was your move. As Pat notes, several different isotope ratios lead to the conclusion that humans are largely responsible for the increase. Does this give enough precision “to be useful”? That is, of course, not a scientific question. But tt seems to me to be pretty clear that the likely answer to “will CO2 concentrations stabilize if we don’t stop fossil fuel emissions?” is “not till they’re much higher”.

      • Which isotope tell us how much CO2 in the air is from Coke and Pepsi?

      • Rob Starkey


        Again, I am not disagreeing that humans are a major source of CO2 emissions, but I do fault your conclusion that you know humans are definately the source for all the increases in atmospheric CO2. Ist- you can’t really use C14 measurements at all as they have been unreliable since the late 40’s. C12/C13 use was started because C14 could not be used and those measures have large margins of error.

        I am sorry to get in the way of you preconceived notion that you know that humans have caused all the increase in atmospheric CO2…..but we can not prove that, we can demonstrate we are probably the predominate cause, but that it all.

      • Perhaps someone can explain how these isotope statistics demonstrate causality of the global CO2 concentration? I don’t see the reasoning.

      • My view is that they are not a direct demonstration, but they provide tests on scientific understanding of the carbon cycle. Thus they provide indirect support for the estimates based on the knowledge about carbon cycle.

      • Noting the sub-threads above, I would also raise the issue that the carbon budget of emission says that we have put twice as much CO2 into the atmosphere as the amount it has risen, and also note that Henry’s Law establishes that the ocean should acidify too, and given that it is, it cannot be a source of the CO2 in the atmosphere because there is a net gain in the system.

      • There is no AGW science, but there is climate science that can be used to make estimates on the strength of AGW. Similarly there is no direct scientific research on the increase of CO2 in the atmosphere, but there is research on carbon cycle, i.e. on all important processes that determine the changes in CO2 concentration.

        The research on carbon cycle has built up a satisfactory knowledge on what’s happening. The empirical observations made in atmosphere, oceans, ice sheets and on biosphere have been important in this process. The isotope studies are an important part of this research. Combining that with theoretical knowledge on ocean chemistry and on known sources of carbon from burning fossil fuels and other human activities is enough to give a clear picture on main trends.

        Basically we know that the natural carbon cycle consists of very large annual flows, but leads to only rather small changes in the stored volumes of carbon. We don’t have any natural storages that allow for large releases on the time scale of decades (excluding such volcanic activity that has not been observed in actual recent history). There may be significant year-to-year variations, but they must even up to a large degree, and there may be large changes over thousands of years, but on the scale of decades no known process can have produced changes comparable to what we have observed at Mauna Loa. We have also good reason to be confident that no such processes exist, but have not been observed.

        Everything fits well with the view that essentially the whole change seen at Mauna Loa is due to anthropogenic influence. While this change in atmospheric CO2 is about 57% of CO2 emissions over the period since 1959 (or 55% over a recent 10 year period), it’s not correct to say that 55-57% of the emissions (or some value corrected for other contributions) stays in the atmosphere, as the addition is due to the annual activities, but removal depends on the cumulative difference of the concentration from an equilibrium value, and as this equilibrium value is not unique, but dependent on the way the system boundaries are drawn. Taking all processes into account the removal depends on the history of carbon cycle over a very long historical period. The specific value of 55% (or whatever it is when defined differently) is true only for this history with its increasing CO2 emissions. For a different emissions history the number would be quite different and it’s going to be different in future.

        The main factors of the increase in the CO2 concentration are well understood. The extent of emissions from fossil fuels is so large compared to the uncertainties that there is no space for serious doubts concerning the general picture. There are significant uncertainties in the details of natural processes, but not on a scale that would change significantly the role of fossil fuels.

        This specific issue has similarities with other aspects of climate change. Presenting broadly drawn hypotheses and null hypotheses is likely lead only to misdirected fighting on something beside the really important issues. It’s better to avoid the whole concept of null hypothesis and formulate the knowledge and its uncertainties more directly and comprehensively.

      • Some questions on CO2 residence time:

        Table 1 of the IPCC TAR WG1 report says that the CO2 lifetime in the atmosphere before being removed is somewhere between 5 and 200 years with the footnote: “No single lifetime can be defined for CO2 because of the different rates of uptake by different removal processes.”

        Narrowing this down a bit, the long-term residence time of CO2 in our climate system has been estimated to equal a half-life of around 100-120 years. (Data presented at the Yale Fomum by Zeke Hausfather).

        (I am not referring to the short-term lifetime of 5-15 years, as summarized by Tom Segalstad)

        This equates to a decay rate of 0.58% of the concentration annually or 2.3 ppmv/year at today’s 390 ppmv.

        Human activities (fossil fuels, cement, deforestation) add 30 GtCO2/year today. This calculates to an increase of 4.5 ppmv/year.

        But Mauna Loa records tell us that we are only seeing an actual increase of 2.2 ppmv/year (roughly 49% of the “human emissions”).

        So are the “missing” 2.3 ppmv “leaving our climate system” in accordance with Zeke Hausfather’s data?


      • The best way of understanding the removal of the CO2 from the atmosphere may be the compartment model, which is based on the existence of many different storage places for the carbon. Some of these storages are in direct contact with the atmosphere like the uppermost layers of atmosphere and the biosphere are, while others are in contact with other storages, but not directly with atmosphere.

        Furthermore many of the storages have complex dynamics. Some plants grow immediately faster in higher CO2 levels, while others react more slowly.

        The uppermost ocean and some parts of biosphere are the main basis for the fastest removal, while deep ocean and sediments are at the opposite end of the spectrum. The combined effect is far from exponential, but can be described adequately by a sum of several exponentials (e.g. four) as long as the concentrations do not differ too much from the present values. The multiexponential parameterizations are based on more complex theoretical models, and are thus not fully verified or accurate. They give, however, a rather good and adequately verified description on what to expect.

      • In the first paragraph: “uppermost layers of atmosphere” => “uppermost layers of oceans”.

      • Pekka, this seems to be mostly hand waving, to the effect that we know what is going on. But we do not. We have no idea what the CO2 sources and sinks have done over the last 100 years in, say, the Gulf of Mexico, or the Ontario boreal forest, or the Indian ocean, etc. Given that these are mostly biological phenomena I expect oscillations, not steady states.

      • Details are not known, but the dominant amounts are clearly dominant, and the emissions is on of those. Playing with the much smaller uncertainties is irrelevant compared to that.

        We know the main contributions beyond reasonable doubt.

      • In logic this what is called argument by assertion. I do not agree. Our emissions are certainly not dominant. They are quite small actually.

      • They are twice the change, and they are the only major positive contribution, when natural processes are netted in the sense that transfer across the ocean/atmosphere boundary is netted and growth and decay of plants is netted.

      • It’s really trivial. The carbon atoms do not disappear. We add them to the armosphere. Half of that is removed. We know enough about the amounts of stored carbon to tell that all other main storages are growing or changing much less, not diminishing strongly.

        There is no alternative for anybody not really in denial of facts.

      • Rob Starkey

        Pekka- I hope you noticed that I have in no way denied the basic point that humans are emitting large amounts of CO2 and that those emissions might possibly have an impact on the planets long term climate. I merely get frustrated when people claim something as a demonstrated fact, when in reality it is far less.

      • Rob,
        I think that our views are not very different. We both seem to agree that the human influence is essential, and we both seem to agree that some of the more detailed claims are not well placed.

      • Suppose we were to take this as a test case, and to try to explore the arguments related not to any arbitrary null hypothesis, but to a specific question like “to what precision and with what confidence can we estimate the human contribution to atmospheric CO2 accumulation”?

        I don’t mean this as a short term exploration on this blog, but as a structured conversation with the experts in the field, and potentially with invited “neutral” judges.

        I’m particularly interested in how participants could believe such a process to be useful. It seems to me that in this very short interchange here, there is a pretty strong presumption that neither side will have its mind changed by the other. But I don’t think that needs to be the case.


      • world population in 1959 was 3 billion now we are at 7 billion. Humans breathe lots. Not to mention crop yield increases. co2 can not be reduced. Increases may even be beneficial.

      • The physicist Howard Hayden argues that ocean warming is the source (and cause) of the CO2 increase. He has a nice argument but unfortunately it is not on-line, as he publishes a paper newsletter, the Energy Advocate. If we ever do a technical thread on this I will try to bring Howard in. His primary evidence is a very strong correlation between SSTs and CO2 levels. His mechanism is the generally accepted fact that water expels absorbed CO2 when warmed, or something like that. Let’s just leave it that this argument exists, which is my only point.

      • But we know that CO2 has increased also in upper ocean. Thus it’s a sink, not a source.

      • Not at all. If more CO2 is being produced, hence the increase, and it is a source. One really needs to know what is going on, and we do not. That is my only point.

        Nor do I believe we know that CO2 has increased in the upper ocean globally over the last 100 years. That is even more preposterous than claiming to know the global average temperature, far more preposterous.

      • We may not know empirically CO2-levels over all periods, but we know, what is going on now and and what has been going on for years. That’s enough for the principal effects.

        There are certainly uncertainties, when we are looking at effects that are on the level of a few percent of the total increase over the last 50 years (Mauna Loa period). There are also uncertainties on that level or a bit higher in the estimates on the future persistence of emitted CO2 in the atmosphere.

        But none of this provides any credible alternative to the general picture that the increase in CO2 levels over the last 50 years is essentially in whole due to the anthropogenic influence.

      • Here is the basic point. There are thousands of sources and sinks (technically trillions or more). If some increase and some decrease then the net change is caused by all combined. No single source, such as human emissions, causes the net change. And we have no idea how all these sources and sinks are changing. It is very simple. Moreover, most of these sources and sinks are biological so steady state is their least likely state.

      • David –
        As a biologist, I’m a very good engineer. So, please correct me if I’m wrong, but isn’t CO2 one of the byproducts of biological decomposition ?

        If so, I can physically show anyone who cares one of the major sources – tens of thousands of linear miles/millions of square miles of source that is natural – not anthropogenic.

        What I’m seeing here is an argument by assertion that only ACO2 is THE major source because other sources cannot be quantified. And a lot of handwaving about correlation being equivalent to causation.

        And I don’t believe that. Correlation is not causation – certainly not until the physical process can be explained in detail. And even then – there are multiple precedents for the explanation having been wrong.

      • Jim – When a discussion begins to repeat itself, my usual inclination is to refrain from adding to the repetition. In general, therefore, I would let readers refer to the evidence and numerous reference sources cited above by Pat Cassen, Paul Baer, Pekka Pirila, and myself, which converge to the conclusion we draw that human activity accounts for all of the 40 percent increase in atmospheric CO2 observed since preindustrial days, with the possibility of only minor exceptions. When that evidence is examined in its entirety, the human contribution is affirmed quantitatively, and no mechanism for an alternative major source of the increase becomes tenable. None has been advanced within these discussions.

        I’m responding to you, however, because you asked a question rather than making a claim. To answer your question – yes, biological decomposition generates large quantities of CO2. The larger question, obviously, is how this fits into the carbon budget of the climate system.

        In nature, CO2 enters the climate system essentially through volcanic activity and related emissions from beneath the Earth’s crust, both terrestrial and submarine. It leaves through subduction zones, mainly as carbonates, which are heated through the volcanic process and returned to the system as CO2, thus completing the cycle. Volcanic activity proceeds over the decades at a level of only about 1 percent of industrial CO2 emissions, and even major eruptions observed during the past century have changed atmospheric CO2 trends only minimally and transiently. Equally important, the process is not oxidative, and so the commensurate reduction in atmospheric O2 we have seen with rising CO2 requires an oxidative process for its explanation that volcanism can’t provide.

        Biology shifts carbon within the climate system, but it can’t change the total. The CO2 emitted from biological processes must come from somewhere. The source is atmospheric CO2 that is photosynthetically incorporated by plants (including oceanic plankton) into the carbohydrates and derivatives that comprise plant material. This in turn is respired by plants, animals, and other life forms back into atmospheric CO2, as you mention. As a natural process, could this account for a significant fraction of the 40 percent increase, which in total amounts to more than one trillion tons of CO2?
        Such a phenomenon would require an enormous, non-anthropogenic, long term net reduction in total plant biomass on the planet, because without such a reduction, CO2 would merely be cycling rather than accumulating. Not only has this not been seen, but the predominant loss known to have occurred has been anthropogenic, due to deforestation.

        There is another problem with invoking massive plant loss as an unidentified alternative to the anthropogenic contribution to CO2. Carbon isotope data (obtainable in biological materials such as tree rings that can extend back long before the atomic bomb explosion in 1945) demonstrates that atmospheric CO2 has exhibited a reduction in the C14/C12 ratio commensurate with a rise in total CO2 concentration to an extent indicating that the additional carbon coming from any postulated unidentified source (e.g., outside of deforestation) must have been lacking in C14. C14 is generated constantly from atmospheric nitrogen, and so plants utilizing atmospheric CO2 will incorporate C14, albeit less efficiently than C12.

        However, plants that have died, become fossilized, and remained underground for millions of years will contain no C14, because its half life is only about 5000 years. The dilution of atmospheric CO2 by CO2 with no C14 can therefore have come from fossil fuel combustion exclusively or almost exclusively. I expect that a minor additional source is carbonate in rock material used for cement manufacture, but that is of course also an anthropogenic process.

        The argument has been made that there are uncertainties in all the relevant measurements – of recorded industrial emissions, isotope ratios, atmospheric oxygen, and the like. That is undoubtedly correct, and is a good reason to conclude that during any short interval, we may have inaccurately estimated a carbon budget that is imbalanced for a variety of reasons. As a reason to doubt that over the long term, anthropogenic activities have been almost the entire contributor to the observed rise in CO2, it requires us to have incorrectly estimated many separate lines of data in the same wrong direction, while neglecting an unidentified major source of CO2 for which no evidence exists. This is possible, of course, but it should be subjected to the principle of Occam’s Razor.

        I will leave it there, with the expectation that those who have not asked questions will continue to conclude what they wish, but that I may helped answer the question you asked.

      • Rob Starkey


        You wish to say the exchange became repetitive, but imo what happened is it was shown that science can not really claim that the entire rise in atmospheric CO2 is due to humans. I posted a link to studies the showed:
        “The researchers found that soil respiration had increased by about 0.1 percent per year between 1989 and 2008, the span when soil measurement techniques had become standardized. In 2008, the global total reached roughly 98 billion tons, about 10 times more carbon than humans are now putting into the atmosphere each year. The change within soils “is a slow increase, but the absolute number is so large, even a small percentage increase is quite a bit,” says Bond-Lamberty.”
        People who wish to claim that increasing levels of atmospheric CO2 is dangerous often wish to minimize the truth that many of their claims/opinions are not supported by science. To state that you can demonstrate that humans are responsible for all, or virtually all of the CO2 increase over the last 50 years is not correct. To claim that it can be proved by isotopic data is even more inaccurate.
        In spite of all of the above, it does not change the basic question- “are human CO2 emissions harmful for humanity overall over the long term to the extent that steps should be taken today to reduce or eliminate said emissions.”

      • Rob – My response above to Jim should answer your question – see the distinction between the rate of carbon cycling and carbon accumulation. Non-human processes can accelerate cycling, but as I explained above, can’t account for a significant component of long term accumulation, because in a cycling system, the carbon they add to the atmosphere must come from the atmosphere. Also, although it’s a side issue, I believe the article you refer to considers the increased respiration to be consistent with anthropogenic warming.

      • David L. Hagen

        Fred Moulton:
        You assert: “the process is not oxidative, and so the commensurate reduction in atmospheric O2 we have seen with rising CO2 requires an oxidative process for its explanation that volcanism can’t provide.”

        Biodegradation and natural peat fires provide oxidative CO2. e.g. See Indonesian Peat Fires
        “For instance, during the drought associated with the El Niño phenomenon in 1997/98, up to 2.57 gigatons of carbon was released from the wetlands of Indonesia alone. . . . When one uses the new data to estimate the level of CO2 released for all of Indonesia in 2006 – a year with a weak El Nino, in which rainfall was relatively low – one comes up with a figure of up to 900 million tons. This value exceeds the total emitted from all sources in Germany in that year, and corresponds to about 16% of all emissions attributable to deforestation worlwide. ”

        On unidentified sources, there is growing evidence that solar modulation of cosmic rays affecting clouds has a much greater effect than IPCC estimates, with consequently smaller anthropogenic causes. See <a href=, May 19, 2011.

      • David Hagen – I addressed your point in my response to Jim Owen.

      • David Wojick 5/19/11 at 2:53 pm, Null hypothesis

        The central AGW argument in the IPCC version begins with estimating the global rise in CO2 from MLO measurements in the last half century, and then attributing that rise in CO2 to human activity. The attribution that MLO is global is a separate problem for later.

        IPCC begins its attribution story with the Revelle factor, RF, also called the buffer factor, formulated by Revelle and Suess in 1957 on the threshold of the development of a surge in atmospheric CO2 by Revelle protégé Charles Keeling. RF was a numeric, defined as the ratio of ACO2 emissions to CO2 in the air divided by the ratio of ACO2 dissolved to CO2 in the surface layer of the ocean. It is not a consequence of any law of physics. Revelle and Suess could not establish a value for RF that fit their model of the air-ocean carbon cycle.

        In 2006, Zeebe and Wolf-Gladrow redefined the denominator of RF as the ratio of the increase in Dissolved Inorganic Carbon to the total DIC in the surface layer, now called RF0. The denominator became an analytical expression, so the authors derived an explanation for RF0 based on equilibrium carbonate chemistry for DIC. They also provided a pair of charts for RF0, along with a chart for Henry’s Law coefficient of solubility, which silently showed that RF0 was identical to a change of scale of the solubility coefficient for all CO2 in water.

        In the development of AR4, IPCC reported the Zeebe & Wolf-Gladrow results, but when a reviewer/contributor criticized the results, IPCC deleted and concealed references to the equivalence between RF and solubility. Nevertheless, IPCC continued in AR4 to rely on the RF as its primary method to attribute the bulge in atmospheric CO2 to humans, citing Zeebe & Wolf-Gladrow as authority, and ignoring that RF0 no longer was peculiar to ACO2. That is, under the RF model, natural CO2, which exchanges between air and surface at more than an order of magnitude greater than ACO2, would be as susceptible to accumulation in the atmosphere.

        The primary model for attributing the observed bulge in atmospheric CO2 to human activity was invalid, but IPCC concealed its failure.

        Additionally, IPCC’s reliance on Zeebe & Wolf-Gladrow equilibrium carbonate chemistry requires the surface layer to be in thermodynamic equilibrium, which requires simultaneous mechanical equilibrium, chemical equilibrium, and thermal equilibrium. None is satisfied in the ocean. The ACO2 accumulation in the atmosphere is too small to be measured; ACO2 acidification is too small to be measured; and the ACO2 contribution to the global average surface temperature is too small to be measured.

        In a second attack on attribution, IPCC claimed that the observed decline in atomic weight of atmospheric CO2 showed that lighter ACO2 mixing with heavier atmospheric CO2 was evidence that ACO2 was the cause of the bulge at MLO. IPCC appeared to be claiming that the weight of atmospheric CO2 started with the weight of natural CO2 in 1750, but it never supplied a mass balance analysis to show that this isotopic attribution was reasonable. (δ13C for natural CO2 is zero, for ACO2 is -27.2‰, and for the atmosphere is now over -8.1‰.) IPCC did not pursue that mass balance angle. Instead, it turned to the claim that since the early 80s, the observed lightening paralleled the estimated rise in ACO2 emissions.

        At the same time IPCC claimed that the rise in MLO CO2 just matched the decline in atmospheric oxygen, showing the rise was due to fuel combustion. IPCC created a pair of graphs that showed the parallel progression, but each was manufactured by putting the two records on left and right ordinates, and then scaling the right ordinate to make the records parallel. This is humorously called chartjunk, but it is deception and unethical.

        IPCC also claimed weakly that the bulge at MLO was caused by man because it was coincident with the industrial era. This is the fallacious art of using coincidence as a substitute for cause.

        In summary, in spite of quantifiable opportunities to attribute humans to CO2 observations, none worked, and IPCC resorted to concealment and manufactured evidence.

        The complexity in the climate story is manufactured. The null hypothesis that the Sun causes the observed warming is true, and IPCC’s conjecture, H1, that ACO2 is the cause is false because (1) CO2 does not accumulate in the atmosphere, and (2) the observed increase must be due to natural causes. We thus have a short cut through the immense maze of climate complexities to a quick and convincing answer: AGW does not exist.

  72. If you want to anticipate what someone will say on a topic, it never hurts to read what they have said in the past. In the “null hypothesis” case, Kevin Trenberth made a more detailed statement than the one cited in the invitation received by Dr. Curry. Here is an excerpt:

    “Prior to the 2007 IPCC report, it was appropriate for the null hypothesis to be that “there is no human influence on climate” and the task was to prove that there was. The burden of proof is high. In general in this case, scientists assume that there is no human influence and to prove that there is requires statistical tests to exceed the 95% confidence level (5% significance level) to avoid a chance finding of a false positive. To declare erroneously that the null hypothesis is not correct is called a type I error, and the science is very conservative in this regard about making such an error. Scientists are thus prone to make what are called type II errors whereby they erroneously accept the null hypothesis when it is in fact false.
    Given that global warming is “unequivocal”, and is “very likely” due to human activities to quote the 2007 IPCC report, the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence. Such a null hypothesis is trickier because one has to hypothesize something specific, such as “precipitation has increased by 5%” and then prove that it hasn’t. Because of large natural variability, the first approach results in an outcome suggesting that it is appropriate to conclude that there is no increase in precipitation by human influences, although the correct interpretation is that there is simply not enough evidence (not a long enough time series). However, the second approach also concludes that one cannot say there is not a 5% increase in precipitation. Given that global warming is happening and is pervasive, the first approach should no longer be used. As a whole the community is making too many type II errors.

    So we frequently hear that “while this event is consistent with what we expect from climate change, no single event can be attributed to human induced global warming”. Such murky statements should be abolished. On the contrary, the odds have changed to make certain kinds of events more likely. For precipitation, the pervasive increase in water vapor changes precipitation events with no doubt whatsoever. Yes, all events! Even if temperatures or sea surface temperatures are below normal, they are still higher than they would have been, and so too is the atmospheric water vapor amount and thus the moisture available for storms. Granted, the climate deals with averages. However, those averages are made up of specific events of all shapes and sizes now operating in a different environment. It is not a well posed question to ask “Is it caused by global warming?” Or “Is it caused by natural variability?” Because it is always both. It is worth considering whether the odds of the particular event have changed sufficiently that one can make the alternative statement “It is unlikely that this event would have occurred without global warming.” ”

    What seems apparent to me is that he is not recommending a new null hypothesis simply stating that climate change in some vague, general way, and to some unspecified extent is attributable to human influences. Rather, he suggests that a set of observable phenomena at a specified level could be the subject of null hypotheses. These might include the frequency and intensity of precipitation or heat waves, as well as other phenomena that occur naturally but whose magnitude or frequency could be subject to anthropogenic influences. He further suggests that the reason for this change is based on climate data, including increased atmospheric water vapor as a function of higher temperatures, which is why he states “the odds have changed”.

    My purpose in quoting this is not to provoke another argument about the validity of his opinions – that has already been argued extensively – but rather to provide what may be a clearer picture of the position he will take in his forthcoming article. It is certainly more nuanced than a simple “let’s reverse the no-human-influence null hypothesis”, and probably more worth addressing when Dr. Curry writes her own piece on the subject.

    • thanks for this, Fred.

    • Trenberth still begs the question of ‘how much?”
      And avoids the apparent answer very diligently.

    • Fred, I think you have offered the perfect defense. What Trenberth means has yet to be said (by him). We will just have to wait until he says it.

    • Fred M: Dr. Curry did link to the full piece by Trenberth — and what a sloppy, partisan, and frankly bigoted work it was, complete with self-congratulatory cartoons at the end. It’s hard for me to understand how a thinking person could take it as anything but a polemic.

      For Dr. Trenberth, other null hypotheses have been reversed to include that the Climategate emails were “hacked”, climate change skeptics are “deniers” and are not to be debated, and the media is “complicit in the disinformation campaign of the deniers”.

      I’m not a climate scientist and I can only go so far in criticizing their work, but when I read someone like Dr. Trenberth speaking of things I do understand and doing so in such a biased, faulty way, I must ask why is he to be trusted at all?

    • …which is why he states “the odds have changed”.

      Fred M: But what does this mean? As hunter asks, “How much?”

      Not long ago you got into trouble here for arguing — for far too long — that global warming had changed the odds for the Japanese earthquake via a few centimeters of sea level rise.

      Is this the line of argument you and other climate change advocates wish to continue?

      • Huxley – Regarding your two comments, I do agree with you that elements of Trenberth’s extended address have a polemical tone to them that I find troubling. However, the part I quoted deserves to be taken seriously whether one agrees with it or not, and I expect Dr. Curry will do that, even though she probably has some disagreements with Trenberth’s overall perspective. The “odds have changed” element entails the familiar concept that when a claim first surfaces, it needs to be substantiated to be credible – i.e., it must be considered unproven – but once evidence accumulates, the burden of proof may shift at some point to those to reject the claim. The point at which a shift is warranted becomes a matter of judgment based on evaluation of evidence, in this case evidence for an anthropogenic influence and its consequences, but the principle that a burden can shift is a legitimate one. Quantitation is one of the elements Trenberth discusses in the piece I quoted (e.g., a percent change in precipitation), and is relevant in that some of this thread had assumed that Trenberth was proposing a new null hypothesis with no quantitative aspects.

        Regarding the earthquakes, you may have misremembered the thread, because I came to no firm conclusion on the possible miniscule relevance of sea level rise, but information I’ve since seen provides some support for the possibility in that a lower tectonic plate supporting ocean weight might be pushed downward more than an upper tectonic plate anchored by land. I’ll refrain from rehashing the subject, though, because I see its practical significance as negligible.

        One interesting sidelight about a part of his presentation not quoted above involves the question of whether the Climategate emails were “hacked” or “leaked”. I have no dog in that fight, but it seems to me that neither of these can be a clearly preferable null hypothesis because either one is simply a guess. “Leaked” may be slightly more likely, but I can visualize a circumstance whereby an insider gossiped about the emails to an outside hacker who then decided to get them to the public. It might be best to describe the event as an “unauthorized release” of the emails, because both “hacked” and “leaked” have an advocacy flavor to them.

      • Fred Moolten

        no matter how nuanced he presents it, the basic problem with Trenberth’s proposal is

        a) the premise that AGW has been the principal cause of 20th century warming (and thus represents a serious potential threat) has not been validated by empirical data based on physical observations or reproducible experimentation

        b) the premise that AGW is the direct cause of recent severe weather events has also not been validated by empirical data based on physical observations or reproducible experimentation

        c) the recent cooling of our planet, despite CO2 concentrations increasing to record level, has tended to falsify the premise that CO2 is a primary driver of our climate; until this apparent falsification can be scientifically refuted, the “dangerous AGW” premise remains an uncorroborated hypothesis at best; if this cooling should continue for another few years, the “dangerous AGW” hypothesis will become a falsified hypothesis.

        Trenberth is nervous about the “unexplained” recent lack of warming (which he referred to as a “travesty”), because he sees that time may be running out for dAGW.

        So, before that happens, he is trying to bypass the rules of the scientific method in his favor.

        This has nothing to do with science and everything to do with politics.

        In a scientific court of inquiry he would be tossed out on his ear.

        Judith’s very best argument is the scientific method itself.


    • Fred: Scientists have been dealing with the problem of Type I and Type II errors for a long time. The only reason Trenberth is advocating a change is political and the only cause of climate change for which he wants to change the rules is man. Trenberth says man’s role in warming is proven (at least 0.35 degC, anyway) and we are making to many Type II errors not attributing other forms of climate change to man. Take this year’s floods on the Mississippi River, anthropogenic GHG’s must be responsible even though the carrying capacity of saturated air for water rises only 7% per degC or 2.5-5.0%. The sun clearly plays a role in global warming: AR4 says that solar forcing in the 20th century was significant and approximately 10% of total forcing. (That’s several times bigger than the change in absolute humidity.) The coldest temperatures in the last millennia occurred during the the Dalton and Maunder minima. Is Trenberth saying that we are making too many Type II errors by not blaming the sun for climate change? No, the new rules will only apply to anthropogenic causation. Natural variability makes it difficult to invalidate climate models that make predictions disagree with observations, such as amplification of warming in the upper tropical troposphere. Is Trenberth saying that we are making too many Type II errors when we don’t judge these models incapable of making useful predictions about future climate? Of course not. Climate scientists are making too many Type II errors ONLY when they are studying anthropogenic effects. For Trenberth, after AR4 successfully convicted man of a trivial amount of global warming, man is guilty until proven innocent (by his new null hypothesis) of ALL unusual weather phenomena.

      • Of course Trenberth’s argumentation is on policies, not on science. The principal conflict is political, not scientific. Only very few skeptics are active based on standard scientific incentives, and the pro AGW scientists take activist positions to influence policies. On both sides many people try to frame their stances as being on science, but that is only tactics (intentional or not).

        The scientific issues on climate change are such that very few scientists would even mention null hypotheses except, when testing specific details. The concept is brought to the discussion much more by the skeptics in their tactics of trying to make the issue appear an issue on science in a way favorable for their cause. In this the misrepresent the way science is really related to the issue on policies. The activist scientists are equally culpable in misrepresenting the proper role of scientific knowledge in decision making, but they do it usually without reference to null hypotheses, although there are similarities in, how they wish to derive policy conclusions from statements about scientific knowledge.

        There is no doubt that Type II errors are commonplace. Emphasizing them is at the basis of the Precautionary Principle. It’s correct to take both types of errors into account, when deciding on policies and specific actions. This is done all the time in political decision making, there is nothing new in that. Trenberth didn’t introduce anything really new in his presentations, he is just reformulating same ideas the pro action party has been using as their main argument all the time. The argument is in that sense logically valid although I dislike this reformulation, but being logically valid is not by itself that the argument is stronger than opposing logically valid arguments. Judging even semiquantitatively the strengths of various arguments is essential for making best possible policy decisions.

      • Adding to above.

        While scientists do not explicitly discuss very often the concept of null hypothesis, it’s implied in every analysis that results in a confidence interval. The null hypothesis that is rejected states that the confidence interval is not valid. From the point of view of statistical analysis there is a difference between testing a hypothesis specified fully before the experiment and determining the confidence interval, but the issues are closely related and methods developed to handle the differences correctly.

        Most of the determinations of confidence intervals concern some specific details, whose importance for the more general conclusions is not straightforward. The confidence intervals and levels of certainty presented as main conclusions in the IPCC reports can, however, be interpreted to represent evidence against certain null hypotheses. The most obvious example is the IPCC/AR4 conclusion that the equilibrium climate sensitivity is very unlikely (less than 10% likelihood) to be less than 1.5 C.

      • Actually I disagree.

        The principle dispute is scientific. The sceptic is saying that the AGW hypothesis is unproven (i.e. H0 that global temperature increases aren’t predominantly natural hasn’t been rejected at reasonable levels of confidence), and this is perhaps why there is some focus on the null in sceptical circles.

        I should add that Trenberth is making a rather unique contribution to science here, he is suggesting that you can reduce type II errors by simply changing the null – even if you also can’t nullify the new null.

      • HAS,
        Reading your detailed comment further down on this thread, we do indeed disagree.

        My view is that the broad statement “AGW is true” and the related related null hypothesis “AGW doesn’t exist” are not part of good scientific discussion, because the issue is not an independent scientific theory or hypothesis, but a side result of climate science. From the scientific point of view there are many better formulated hypotheses concerning changing climate. The AGW is a concept for policy argumentation. Science provides information on AGW, but science has not created the concept and the concept is not internal to science.

        Inverting the null is guaranteed to invert also the Type II errors. What was Type II error for the earlier setup may become Type I error or may disappear from the list containing both Type I and Type II errors, while new opportunities for errors are created.

        There is no unique argument of logic to tell, what is a appropriate null hypothesis. The choice is always determined on the question posed, and there is no limit on, how many questions may be posed. What Trenberth is proposing is to disregard the proposal of skeptics and start to apply the proposal of those supporting the precautionary principle as expressed by UN, when UNFCCC was created. My view is that neither should be used in the way proposed by its supporters.

      • thanks pekka, i like your last para, KT’s motivation seems to be in the context of the precautionary principle

      • I suspect the issue comes down to what you call a “science” – is for example economics in some shape or form a “science”?

        For my part if science and the scientific method(s) can’t help answer the question “to what extent does man made CO2 cause temperature increases?” I think we are indulging in what might be called a “high redefinition” of the word “science”.

        On the precautionary principle I agree that it is a red herring here, particularly in the context of type I and II errors. If Dr T is saying he wants to avoid the possibility of AGW at all cost then of course he should avoid ACO2 at all cost. The choice of H0 and the science makes no difference.

        Finally I must add that while I completely agree that you can pick your null as you see fit, what you can deduce having done so is then constrained.

      • In this discussion I have in mind the model of physics for science, because the atmosphere is a physical system. For the implications of using physics as the model I use my knowledge on, how highly valued scientific research in physics has been done in practice. That seems to be a much better guideline than mutually contradictory views of philosophers of science (they are often interesting, but not so useful, partly, because they are contradictory without resolution of the differing views).

        My view is that the physical climate science related to AGW cannot be presented well on the basis of hypothesis testing, because there are no specific precise hypotheses that summarize the relevant scientific knowledge. The structure of the knowledge is far more complex consisting from innumerable pieces of information as well as continuous functions and distributions.

        When we move from the science to decision making, we must accept that each decision is discrete. At some point it’s finally defined as choosing one out of two or more discrete alternatives. At that point we must accept or reject some policy related hypothesis. Science has produced some background information for that decision, but science alone cannot tell, what’s the right decision.

        Here I side strongly on the role of “The Honest Broker” for those who bring the scientific knowledge to the decision making process. “The Issue Advocates” confuse the process as they don’t even try to be objective, but are willing to misrepresent the relative weights of opposing arguments. I don’t approve relativism, but I consider it essential to understand that even those who try to be honest brokers will always be subjective. This makes the task of the well intentioned decision makers difficult, but that is an unavoidable fact of life.

      • I would be of the view that the atmosphere is a complex stochastic system and therefore demands inter alia the use of techniques drawn from statistical analysis, rather than the more deterministic end of physics.

        In this it shares a lot of characteristics with some of the engineering and social sciences for example (as an aside I get a wry smile when I hear people say climate science is unique because we only have one experiment, and think about the way social scientists leap on those rare longitudinal studies to help them understand things like learning and criminal behaviour).

        Because of its stochastic nature climate science should not just be concerned with estimating parameters, it should have a strong focus on things like their distribution – something regrettably you don’t see a lot of. Climate is uncertain and therefore the only thing predictable about it might be the bounds on that uncertainty, but you can still get still get a working hypothesis out of this – complexity and stochastic behaviour don’t prevent scientific study.

        Finally three points on decision making.

        First I’d observe that there is nothing wrong with a bit of applied science – science directed to a specific purpose.

        Second I’d have to say it is a mistake to say decision making is discrete and binary. Responses can (and are all the time) tailored to the uncertainty in knowledge.

        Third good decision making processes can (and does) take account of bias in the professional adviser – and you see that in the political response to climate science in a number of jurisdictions.

        So I think rigorous science has much to contribute to understanding AGW and the response if any that might be required.

      • HAS;
        but does rigorous science have much to contribute to Catastrophic Anthropogenic Climate Destruction studies?

      • HAS these are good points you are making in terms of reducing type I, II errors by changing the null

      • If that’s indeed true (and I agree with you that it is), then not only the WG1 question about warming comes into the question, but also the WG2 questions about consequences. It’s at least theoretically possible that a 2 or 3 degree rise will have no serious negative impacts. So the null hypothesis in this case implies something about the WG2 questions as well. IOW, that there’s moderate warming – OR – moderate consequences (or by DeMorgan’s theorem NOT (high warming – AND – severe consequences).

        Which has the effect of increasing the probability of the null hypothesis.

  73. Batheswithwhales

    “Given that global warming is “unequivocal”, to quote the 2007 IPCC report, the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence. ”

    In the quote above, Trenberth seems to assume that the question in contention is whether there is global warming or not, meaning that the old null was that “there is no global warming”, and that the new null would be “there is global warming”. Which of course is nonsense.

    As we know it is about the degree of anthropogenic causation. And this is where it gets a little more interesting, for what would be the new null hypothesis? That precisely ALL of the modern warming is caused by human emissions? More than the present warming if it wasn’t reduced by other factors? A certain fraction of it? Most of it? We have heard all of these from different sources.

    It is impossible to disprove a general notion, which is basically correct in principle, but where the actual amount is in contention even among the proponents of the mainstream view.

    It is all word games anyway. If the old null was “all climate change is natural”, then no one took any notice of it anyway. It was not useful in any way. And neither will new null hypothesis be useful. Trenberths mission here seems to be another: he wants to change the dynamics of the debate.

    Instead of having to defend IPCC mainstream from being attacked from all angles, he proposes that the attackers come up with new theories to rival mainstream theories, thereby making the establishment the attackers and the outsiders the defenders. It would of course be much more comfortable.

    (Which would also make former deniers the new Coolists, and the former warmists the new deniers in the process. Bet he didn’t think of that!)

    • “Instead of having to defend IPCC mainstream from being attacked from all angles, he proposes that the attackers come up with new theories to rival mainstream theories, thereby making the establishment the attackers and the outsiders the defenders. It would of course be much more comfortable.”


  74. “the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence”.

    Oh, I get it. They specified H1, not H0. In other words
    H1 := “There is human influence [on X]
    By inference:
    H0:=” There is no human influence [on X]
    X is not directly specified, but X = [Global Warming] we can conclude in context with p>=.95 /sarc

    So, in the spirit of admitting and correcting mistakes, let me own up to a mistake that the Null Hypothesis was in front of my eyes all the time. You did have to realize H1 was clearer than H0, and you had to tease out the GW, but it is there. I just didn’t believe anyone would be so daft to propose this as a serious null hypothesis.

    H0: There is no human influence on Global Warming.

    Has been falsified. Therefore the new null hypothesis is
    H1: There is human influence on Global Warming.
    (please send Nobel Price c/o Ken Trenberth).

    I wonder what peer reviewed journal this first appeared in. I also wonder if it was ever part of any UN organization Charter.

    Good Grief! With such an H0, KT does not even have to commit himself to the sign of the “influence.”

    • Assuming H0 is “There is no human influence on Global Warming.”

      My first recommendation to Dr. Curry is:
      “Don’t Play. The Game is Rigged. You have been set up for a sucker punch. Either that, or you will be accused of throwing the game to the opposition. You cannot win this argument.”

    • Recommendation #2
      H0 was falsified decades ago. When man burned high sulfur coal spewing SO2 into the atmosphere, we felt it. Jet aircraft add measurable ice particles to the stratosphere. Urban Heat Islands rais the temperature of their environment. Their environment is part of the Globe, therefore UHIs influence Global Warming. Humans make Urban Heat Islands. QED.

      Humans influence Global Warming, for the colder and the warmer.

      “Is this all you have to show for a decade of work?” We want our money back.

    • Recommendation #3 harkens back to my 5/18 00:02 post.
      “Ken, in the interest of time for us all, I concede immediatly H0 is falsified. In fact, I thought it was falsified when we created Earth Day, but what do I know?
      While I have your attention, maybe we can clear up a bunch of other old business today:

      I think the following other Null Hypotheses should be formally falsified the same way with your concurrence and support”
      Hs0: “There is no Solar influence on Global Warming.”

      Hsv0: “There is no Surface Volanic influence on Global Warming.”

      Huv0: “There is no Undersea Volanic influence on Global Warming.”

      Hmc0: “There is no Milankovitch Cycle influence on Global Warming.”

      Hcr0: “There is no Solar Magnetic / Cosmic Ray influence on Global Warming.”

      Hoc0: There is no ENSO influence on Global Warming.

      Hpc0: “There is no Cyanobacteriological influence on Global Warming, not to mention Climate Change” [boy, this is REALLY old business…]

      Hun0: There is no IPCC influence on Global Warming with all their world wide conferences. You don’t agree? Then why are there record low temperature everywhere they show up?

    • Recommendation #4 “Get Predictable, Ken” (see Rasey 5/19 23:23)
      Science is about replacing one useable set of rules about the world for another set that describes it better.

      As an example let’s look at the evaluation of General Relativity in the context of Global Warming. The Null Hypothesis Hgrl0 was “Gravity will not bend the path of light more than predicted by Newton’s laws

      Observe a 1919 eclipse. Repeat in 1922. Photographic plates confirmed that the solar gravity changes the apparent position of stars near the eclipse edge, passing near the sun. Null hypothesis falsified, right? Not so fast! General Relativity predicted that there would be a change in the stars position by 1.75 arcseconds, twice the deflection predicted by Newtonion laws. It took until 1960 before radio astronomy settled the matter.

      Suppose the stars positions moved by 2.5 arcseconds, or even 25 arcseconds, H0 might have been falsified, but H1 would NOT be replaced by it. The bending of light would not have been predicted by Gen. Relativity any better than before. If Light had been bent by 1.1 arcseconds, GR would hardly be better than Newton.

      But it is accepted that General Relativity got it right. It makes high precision predictions that can be tested by observation. Predictions better than competing theories.

      Which brings us to the sorry state of Global Warming predictions.

      Is Global Warming happening? Yep. Ask any geologist.
      Has it been warmer before? Yep.
      Has it been a lot colder? Yep.
      Is Sea Level rising? Yep. Ask any geologist.
      Has Sea Level been higher than today? Yep. Ask any geologist.
      And all of it without humans at the wheel.

      So today you come here with your new Human General Theory of Global Warming. Ten years ago you made predictions about how the Globe would be warmer, temperatures deviating from normal expectations, powered by human influence.

      What did you predict?
      What do we observe?
      What would an oceanographer, geologist, a solar astronomer, or the Farmer’s Almanac editor have predicted?
      Who’s predictions are closer to the current observations? Oh, yeah. Those observations have some systemic biases in them, too, don’t they.

      Better luck next time, Ken. Your Human General Theory of Global Warming isn’t close enough to observations to warrant replacing what we have. It overshot the mark.
      H0 is weak, but your H1 is weaker and less useful to a scientist.

  75. TimTheToolMan

    PaulM writes “The abstract of his paper says :
    “Given that global warming is unequivocal, the null hypothesis should be that all weather events are affected by global warming …”
    – nothing about human influence.”

    Global warming may well be unequivocal until it becomes global cooling at which time the proposed null hypothesis still stands except it now weather events being effected by the cooling.

    In essence isn’t the new null hypothesis a no brainer which cant be argued against and the whole point of the exercise from Trenberth’s point of view simply to push towards demonstable agreement from prominent skeptics such as Judith?

  76. Hansen 1988 vs Observation:

    Hansen’s Scenario A=> Exponential carbon emission growth rate of 1.5%

    Observed exponential carbon emission growth rate = 1.64%

    Where is the evidence for AGW?

  77. One difficulty which strikes me about this proposal is that the new “null hypothesis” becomes a remarkable vague and flexible thing. How big are the effects in this new null? What are their mechanisms? What specifically are new hypotheses to be measured against and compared against? Don’t know, don’t care, and you can’t pin me down on it. We just accept that something exists, with no clear idea of what it is, or what it looks like, or even how big it is. How convenient!

    And how chaotic to the process.

    I can’t see how this new null makes any statistical sense until it has at least been quantified, so researchers know what they are comparing their results against. Otherwise how on earth could the numbers be crunched?

    • A problem well stated is a problem half solved.
      Charles F Kettering

    • Trenberth has addressed some of those issues – see Comment 69380

      • Fred,
        Thanks for the longer excerpt. But Dr. T still relies on rgument by authority and more importantly, avoiding the real issue: are the changes worth worrying about?

      • Rob Starkey


        I believe you are correct. The single most important issue (imo) is the determination of whether any potential environmental changes are harmful to humanity overall, and whether the proposed “solutions” are worthy of the cost.

        The IPCC’s 4th assessment states:
        “With regard to changes in snow, ice and frozen ground (including permafrost), there is high confidence that natural systems are affected.” Throughout the report and assessment, there are multiple comments such as that with virtually zero information that can demonstrate that the potential changes are harmful to humanity overall.

        As an example, it would be easy to write a report pointing out all the creatures that would be killed or displaced as a result of building a dam that would both prevent floods and provide electricity. Before such a project is approved a through cost benefit analysis must be performed and evaluated.

        In the case of climate change, it seems that many of those who believe that it is a dire problem wish to jump to their conclusion and claim that their thoughts/position are based upon “wise scientific authority” and to disagree makes one subject to being a denier or an intellectual dolt.

      • Rob,
        Notice how the AGW believers avoid this issue.

  78. David Bailey

    “Given that global warming is unequivocal, the null hypothesis should be that all weather events are affected by global warming …”

    If the condition in that statement were really true, there would be no further need for much of the research going on – just as there is no need to check yet again that iron filings dissolve in dilute sulphuric acid!

    I like to start from the fundamental observation that only maths contains proofs. Some science can get close, but ultimately every scientific theory is an induction from a finite amount of data.

    Since any AGW effect is not going to suddenly switch off, there seems absolutely no reason to shift from a null hypothesis that CO2 is doing nothing to the climate, because all the data – including that which is already claimed to have produced an unequivocal result could continue to be used! More and more data should just entrench that unequivocal result.

    I’m afraid I see Trenberth’s statement as political, not scientific – even less logical!

  79. Eric Anderson

    Judith, I think it is important to point out in your essay that Trenberth is confusing the terminology to suit his own purposes. He does not mean AGW is a null hypothesis in the sense that it is something we should now attempt to disprove (he certainly won’t be shifting his work to do so). Rather, he simply claims (by attempting to co-opt null hypothesis terminology) that there is a consensus — that he and his colleagues have proven their case for AGW, and that this should be the working assumption for both science and policy going forward unless someone can affirmatively disprove AGW. This is an extremely pernicious position to take and he should be called on the carpet in no uncertain terms.

  80. People have done a good job of picking this thing apart. I am reminded of one of my favorite legal concepts: void for vagueness. See

  81. I would be remiss if I did not mention that warming, in the sense that Trenberth and the IPCC mean it, is far from unequivocal. In fact it is probably nonexistent. They are referring to the steady GHG-like warming shown between 1978 and 1998 in the Jones-type surface statistical computer models. The satellites show no such warming. The UAH temperature profile shows no warming whatsoever between 1978 and 1997. It also show no warming from 2001 to today, but the latter flat line is a little higher than the flat line before the 1998-2001 ENSO so there is in that sense some warming. But it is not GHG warming.

    • Nebuchadnezzar

      The IPCC are referring to the fact that over a range of periods ending around 2005, when the report was finalised, the world had warmed.

      This warming can be seen in measurements of troposphere temperatures measured by weather balloons and satellites, in measurements of ocean heat content, sea surface temperature (measured in situ and by satellites), air temperatures over the ocean, air temperature over land. The warming of the oceans and the melting of glaciers have raised sea levels.

      You can cut that warming up as you have done into shorter periods during which there is no significant warming, but even if you do that, you have concede as you did that over the full period there was warming.

      The ‘unequivocal’ statement refers to the fact that the world warmed. It doesn’t say what caused the warming, or what we might do about it.

  82. Readers looking for nuanced answers to what Trenberth is really trying to achieve here will note that Judith has filed this thread under “Scientific method”.

    So I would suggest that those still stumbling over Trenberth’s proposal should go back to the definition of the “scientific method”.

    Once one has read this, it’s easy to toss out Trenberth’s suggestion as rubbish.


    • You get things exactly backwards: what was maybe a couple of decades ago the alternative hypothesis is now the null hypothesis, on the strength of the accumulated evidence. You often criticize what you do not know or understand. AGW is well past the kind of denier claims you like to make and Trenberth is stating what is the case – based on the scientific method.

      • Sorry, Martha, I’m simply stating that we should not abandon the scientific method, as Trenberth is suggesting.

        The premise that AGW is a potential serious threat has yet to be validated by empirical data derived from actual physical observations or reproducible experimentation.

        As such, it remains an uncorroborated hypothesis.

        If the current “lack of warming” continues for another 5 or 10 years despite CO2 levels rising to record levels it will have become a falsified hypothesis.

        So it is a long, long way from being the “null hypothesis”

        If Trenberth is trying to pull a fast one to proposed as a “null hypothesis” the premise that humans may have had some impact on our planet’s climate on a local or regional (and hence, on average, a global) basis (a no brainer, as far as I can see) then this is a different story (but it does not mean much).

        I hope you can see my point here.


      • Martha, so are you claiming you know and understand AGW or that you believe that Trenberth does?

  83. I rephrase my basic views:

    From the point of view of science neither null hypothesis makes sense. Claiming that science just or principally about disproving null hypotheses is nonsense.

    From the point of decision making we have two related questions:

    1. Do we have enough evidence on detrimental AGW to be a reason to worry and to consider significant preventive measures? (Here I say “consider”, not “initiate”.)

    2. Do we have enough evidence on the non-existence of serious risks from AGW to conclude that no decisions on significant preventive measures are justified right now?

    Both of these have their related null hypotheses (stating that the evidence in question doesn’t exist), but they are not purely scientific, but contain also other considerations as an important component.

    Both questions are relevant at the same time unless the evidence on either one is very strong. Trenberth indicates that he considers the evidence on the first one to be strong, and that the emphasis should now be on the second one. Many on this site disagree, but even they are likely split on the question, whether there is better evidence for the second question. People, who answer negatively to both questions, are likely to have widely varying views on, what to do.

  84. Joe Lalonde

    If climate science can do it and NASA can do it, why can’t I?
    Generate models and use them rather than physical evidence.
    From the tilting of the axis with Earth quakes to the measurement of land rising are generated by models.
    These have generated mistakes which simple mathematics of backdating shows how bad the errors are by as much as millions of meters depending on how far back you go.

  85. trenberth’s quote is:

    Given that global warming is “unequivocal”, to quote the 2007 IPCC report, the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence.

    This may be obvious to everyone but me. However isn’t Trenberth conflating “human influence” with the IPCC consensus view. So Trenberth’s null hypothesis is that the IPCC consensus view gives valid predictions. That is the null hypothesis is that the currently observed rises in global temperature is the result of green house gases exclusively.

    This seesm to me to be quite a different null hypothesis than that that is given by a surface reading of Trenberth’s quote. One could quite possibly hold that the rise in global temperature has a human induced component and disagree with the IPCC scientific consensus. And by implications disagree with the policy decisions made from that consensus. The Pielke’s would seem to fall into this camp whose views would be excluded from Trenberth’s null hypothesis

    To me, Trenberth’s statement is more of a political than a scientific one. It seeks to privilege one scientific theory on human influence over another simply on the basis of consensus. Trentberth’s suggesting is a policy suggestion that would guide future research. This is a decision that is not resolvable by scientific means

  86. I note in the discussions above that AGW/CAGW proponents are quite vociferous in attacking any skeptic’s proposed “null hypothesis” for the consensus, ie. AGW/CAGW. So limiting the issue to AGW, I wonder what AGW proponents would describe as an appropriate null (or nulls) for their position? I suspect the political goals of those proponents might prevent them from agreeing to state their position in terms that are falsifiable, so I thought I might take a layman’s shot at it.

    The “consensus” is described on RealClimate as follows:

    “The main points that most would agree on as ‘the consensus’ are:

    1. The earth is getting warmer (0.6 +/- 0.2 oC in the past century; 0.17 oC/decade over the last 30 years);
    2. People are causing this;
    3. If GHG emissions continue, the warming will continue and indeed accelerate;
    4. (This will be a problem and we ought to do something about it).”

    It seems to me there should be 4 null hypotheses then, one for each element of the consensus.

    The null for point 1 would seem to be fairly straight forward, and the type of statistical null the AGW proponents argue for above: the earth has warmed 0.6 +/- 0.2 oC in the past century; and 0.17 oC/decade over the last 30 years.

    The problems arise with the remaining three points.

    For point 2, in order for there to be a statistically verifiable null, it seems that the degree of causation would have to be quantified. To what degree are “people causing” the warming? The RealClimate iteration – “people are causing this” – appears to attribute all of the recent warming to anthropogenic causes. So perhaps the null would be that: human CO2 emissions have caused warming of 0.17 oC/decade over the last 30 years.

    Point 3 is really problematic for the consensus. Given the framing of these points, I think a fair null for point 3 would be: Continued emissions of GHG at the same rate as the last 30 years will cause the warming to continue at the minimum rate of 0.17 oC/decade for X years. X being the time frame within which the absence of such warming would become statistically significant. (Some consensus types will might argue for a time period beyond the expected length of their careers so there will be no chance of falsification when it could adversely affect them, but perhaps there is some more logical, or scientific, means of establishing the time frame.)

    Finally, point 4 would have to be formulated to include the policy implications that are implicit in that point. (It is this point, really, that has caused all the kerfuffle.) In seems the null for this point would have to be something like: continued warming in the amount of 0.17 oC/decade over the next Y years will cause harm to people to an unacceptable degree. Not only would you have to quantify Y, but also the degree of damage that would be “unacceptable” for this to able to be falsifiable statistically.

    My feeble attempt at constructing null hypotheses for the consensus as outlined by RealClimate:
    1. The earth has warmed 0.6 +/- 0.2 oC in the past century; and 0.17 oC/decade over the last 30 years;
    2. Human CO2 emissions have caused warming of 0.17 oC/decade over the last 30 years;
    3. Continued emissions of GHG by humans at the same rate as the last 30 years will cause the warming to continue at the minimum rate of 0.17 oC/decade for X years;
    4. Continued warming in the amount of 0.17 oC/decade over the next Y years will cause harm to people to an unacceptable level.

    Now all we have to do is agree on X, Y and quantify “unacceptable level of harm” and we’re done. Wasn’t that easy?

    (No, I’m not a scientist; and no, I don’t think these should be taken seriously by anyone. But there are already 346 comments on this thread, and I had a half hour to kill so….)

    • Rob Starkey

      Gary- your point 4-
      “This will be a problem and we ought to do something about it.” is the most important and the one that can not be answered today by anyone.

      In order to really determine if higher levels of atmospheric CO2 are going to be a net problem for humanity we need to understand the impact on both temperature and rainfall at a regional or local level as a function of CO2. Today we do not have any reliable information/models that can tell us this information. What this means is simple terms (imo) is that all we have so far are varying degrees of relatively educated guesses and a group of people who seem to get frustrated when their guesses as to what will happen in the future are not accepted and implemented by the rest of the population.

      So today we can not accurately quantify the problem, but once we can we could then accurately perform cost benefit analysis that would evaluate various mitigation strategies.

      • An AGW believer like Steven Mosher will tell you that AGW theory doesn’t claim that AGW explains all the warming for a given period, but that AGW makes it That Much Warmer Than It Would Have Been.

        This means that whatever other factors cause the temperature to go down will still happen we just Won’t Go As Low As We Would Have.

        Make a null out of that. ;)


      • Bad Andrew

        It’s worse than that.

        Steven Mosher doesn’t specify “how much warmer” or “less cooler” it will be, as a result of human CO2.

        He doesn’t even really tell us specifically whether or not this will be a significant or a barely perceptible difference (because he frankly doesn’t know the answer to that).

        I personally would not count on all the SUVs and coal-fired power plants in the world to save us from a modern Little Ice Age, if that’s what Nature wants to throw at us.


    • GaryM

      Your approach for logically determining whether or not “something needs to be done” is basically OK, except we still have several unresolved steps in the process, as Rob Starkey points out.

      I think the easiest way to visualize this is to look at the attached logic flowchart of the “global warming decision process”.

      We have now just completed step 1 (global warming has been declared “unequivocal”) and are still trying to get conclusive answers to step 2 (i.e. has most of the observed warming been caused by natural or anthropogenic factors?). This question remains unanswered to date.

      Unfortunately there are some who would like to bypass the decision process and go directly to the “act now” step.

      Trenberth is one of these. Judith Curry has stated that she is not.


      • Yeah, uhhh, when I wrote “Wasn’t that easy” (particularly with respect to point 4), I was trying to be funny…apparently unsuccessfully.

        If I were to try to be serious, I guess I would say that it seems to me that scientific null hypotheses could be formulated for points 1 and 2 of the RealClimate defined consensus. A null for point 3 would be trickier, and I doubt the consensus and skeptical communities would ever agree on a time period. And as for point 4, which is the meat of the CAGW debate, I don’t think a statistical/scientific null can be formulated, because it is not merely a scientific term.

        In addition to the issues Rob raised regarding causation, point 4 of the consensus statement as articulated on RealClimate implies a cost benefit analysis, and requires an assessment of what level of harm/damage is unacceptable, and an assessment of the cost and likelihood of avoiding it.

        What this shows is that CAGW (and AGW to a certain degree) is political at its core, something skeptics sense intuitively, and the consensus tries to deny. The fact that CAGW can’t be reduced to scientific, falsifiable hypotheses, is just more evidence of that fact.

      • Gary M

        I would take issue with the statement that point 2 of the RealClimate “consensus” has been validated:

        Human CO2 emissions have caused warming of 0.17 oC/decade over the last 30 years

        It is anything but certain that “human CO2 emissions” have been the principal cause of this warming; this is pure conjecture, based on model simulations backed by theoretical deliberations and some highly doubtful interpretations of dicey paleoclimate data. Alternate explanations exist, which are just as plausible.

        This is “step 2” in the AGW decision process.

        Let’s not bypass this step simply because RealClimate tells us so.

        The “null hypothesis” so far could be stated, “it has warmed since the modern temperature record started in 1850, ergo: global warming is real”. Period.


      • Gary M

        BTW, point 3 of the RealClimate thesis was:

        Continued emissions of GHG by humans at the same rate as the last 30 years will cause the warming to continue at the minimum rate of 0.17 oC/decade for X years

        This premise (i.e. “warming to continue at the minimum rate of 0.170C/decade”) has already been falsified by the past decade’s lack of warming, so this one can be tossed out as falsified.

        Face it Gary. We’re back to premise 1, with all the rest still up in the air.


      • Manacker,

        I would take issue with the statement that “point 2 of the RealClimate ‘consensus’ has been validated” too. Not sure where you got the idea I wrote that. All I said was “If I were to try to be serious, I guess I would say that it seems to me that scientific null hypotheses could be formulated for points 1 and 2 of the RealClimate defined consensus.”

        All I have written about here is whether null hypotheses could be formulated for the 4 elements of the RealClimate consensus. Not whether they have been falsified of “validated.” Personally, I doubt that human CO2 emissions have been responsible for the recent warming, but that’s just one voter’s opinion.

        I also think that the lack of warming/cooling over the last 10+ years casts doubt on the whole CAGW hypothesis, but I do not know whether one can actual,ly say the theory has been falsified based on that period of time. Again, I am no scientist, but one of the reasons I am a skeptic is my belief that natural variability in the climate can be, and has been, substantial historically. It seems to me that, if there is an anthropogenic impact on recent warming (a big if for me), then that impact would be long term as long as emissions continue to rise. Is there some aspect of natural variability that could override such warming for a period of ten or more years? I don’t know enough to say.

        So I seriously doubt the consensus is correct and would not be willing to adopt changes to tax, energy and regulatory policy based on the claims of the CAGW proponents. But I don’t know enough to say that the theory has been falsified scientifically.

        In other words, I am not defending the “consensus,” I doubt that human CO2 emissions are causing global warming, I doubt that such warming would be bad for humans, and I doubt any catastrophic tipping points are imminent. I am an agnostic with little chance of being convinced of CAGW. But I am not really an atheist on the issue, because I don’t know that anyone understands the complex, chaotic climate well enough to say for sure either way.

        The consensus hasn’t “validated” CAGW, but skeptics may simply not be able to falsify it in the near term. Which is of course the whole point of the “burden of proof” and null hypothesis arguments in the political context.

      • GaryM

        Thanks for clearing that up.


  87. Judith: The scientific terms “statistically significant” and “not statistically significant” are analogous to the legal terms guilty and not guilty beyond a reasonable doubt. Lack of a statistically significant relationship between a climate phenomena and a hypothetical cause is similar to a verdict of not guilty in the sense that well-informed people understand neither proves innocence and/or lack of causation – only the absence of evidence that is conclusive by traditional standards (usually p<0.05 for scientists or "extremely likely" in IPCC-speak). In AR4, the IPCC found man "guilty" of causing at least half of 20th century warming. Due to his "criminal record", Trenberth is advocating that man should be considered guilty for all forms of climate change and unusual weather phenomena unless man can be proven to be "innocent beyond a reasonable doubt". This certainly appears to be the legal equivalent to reversing the null hypothesis, and "thereby placing the burden of proof on showing that there is no human influence". Other possible causes of climate change, particularly the sun, will continue to be judged by the traditional standards despite what we know about: the modest but statistically significant 20th-century solar forcing, the cooler young sun, and climate during the Dalton and Maunder minima. (In the eyes of some skeptics, the sun is guilty until proven innocent.) Despite their obvious limitations, climate models will still be considered reliable until the null hypothesis is invalidated. When Trenberth's scientific double standard becomes widely understood by the public, it is likely to cause significant damage to the credibility of climate science.

  88. Serendipitously, there is a major article on Falsifiability in the current issue of the American Economic Review (April 2011), by W. Olszewski and A. Sandroni. They show inter alia that “falsifiability imposes essentially no constraints when theories are produced by strategic experts” (pp.811-812). They must have had Trenberth in mind!

    Be that as it may, I have just had my paper “Econometrics and the Science of Climate Change” accepted (after peer review) by the Economic Society of Australia for its conference in Canberra in July (ANU, 11th – 14th). Its regression results for 3 strategic locations (Barrow in the Arctic, Mauna Loa near the Equator, and NYC/JFK) can interpreted both as confirming the normal nul that there is no anthropogenic role in climate change for non-condensing greenhouse gases, and falsifying the Trenberth nul that there is. What it does show is a major role for atmospheric water vapor, to which there is a significant direct human contribution from both the combustion of hydrocarbon fuels and the cooling needed by steam generation of power, but one that is totally disregarded by Trenberth and the IPCC.

    Interestingly, the paper “Climate Trends and Global food production since 1980” (Lobell, Schlenker, Costa-Roberts, in Sciencexpress, 5 May, Science 1204531) confirms my finding of the absence of climate change in the USA: “A notable exception to the [global] warming pattern is the United States, which produces c.40% of global maize and soybean and experienced a slight cooling over the period…the country with largest overall share of crop production (United States) showed no [adverse] effect due to the lack of significant climate trends”. This also is enough to falsify the Trenberth nul that there is climate change and it is caused by humans.

    My paper is available at or

  89. The Truth About Greenhouse Gases

    Men, it has been well said, think in herds; it will be seen that they go mad in herds, while they only recover their senses slowly, one by one.”

  90. What is the real question here? Isn’t who we should debate with, and about what?

    We could debate with Young Creationist as to the age of the earth. But it is probably not a good use of time. We could debate with parents that say vaccines cause autism, but, again, while we can “win” that debate on the science, “winning it” over and over does not seem to improve vaccination rates.

    Since the theory of AGW has won in the scientific realm, what is the value of proving it over and over? Psychologists tell us that arguing with people tends to strengthen, rather than weaken, their beliefs, regardless of the strength of the arguments presented.

    There really should be no need for a general null hypothesis. For specific issues, like cloud feedback, or sea level rise, or expected temperature increases, use the null hypothesis in the conventional way: whatever the bulk of the literature is saying at that point, or a normal distribution without an underlying trend, where appropriate.

    Those that want to have a general “Is AGW real?” or “Is cancer real?” or “Is the world older than 6,000 years?” are not really participating in a scientific discussion, so the concept of a “null hypothesis” is not really necessary.

    So how do you engage with such a person? However you like. You won’t convince them of anything, and because scientists know the difference between curious people arguing science and ideologies invoking “science,” there arguments will not influence the scientific debate.

    If you want to move the debate forward, you have to focus on the ideology — Christianism, liberatarism, conservatism or what have you — that is making the reality of the physical world so repulsive that its devotees are attacking the people trying to tell them about the physical world. No argument of rational argument, however forceful, will defeat that kind of programming. You have to either go over the heads of the ideologues, or find a way to make the facts seem less threatening to their belief system.

    • Robert

      If you want to move the debate forward, you have to focus on the ideology

      Not at all.

      The debate is because man made global warming is not supported by the data.

      1) Look at the extremely poor comparison between model forecast and observation.

      2) Look at an identical warming to the recent one 100 year ago.

      3) Look at the oscillation in the global mean temperature data.

      4) Look at an AGW advocate admitting they give data at their own peril.

      5) Look at the global cooling since 2002.

      If you want to move the debate forward, you have to focus on the DATA.

    • Peter Wilson

      Since the theory of AGW has won in the scientific realm

      Say what? One can only conclude that you have not been paying attention over the last few years. To suggest that AGW has “won” is akin to Al Gores proposition that “the debate is over”- both demonstrably false, and a good reason to disregard further arguments from the proposer.

      Which I shall do. Run along back to Grist or Climate Progress, your ludicrous assertions will earn you praise there. Here we debate the real science, not your warmist fantasies.

  91. Given that global warming is “unequivocal”, to quote the 2007 IPCC report, the null hypothesis should now be reversed, thereby placing the burden of proof on showing that there is no human influence.

    Global warming was “unequivocal” in the 1940s:

    Why was not the null hypothesis reversed then?

  92. Moderator: Can you delete the previous post, I think I made the mistake of editing in Word and then pasting it in, and must have carried some control characters that deleted a chunk of the content

    I’ve now had a chance to sit down and read the comments rather than just reacting to the odd one. I hope the following might be of some use.

    I think the thread does reinforce my earlier expressed view that a careful discussion of the issues related to the scientific method separate from the politics is important. I think this should be done first for two reasons.

    First because Dr T derives any authority he has from his place in the scientific community, and second because unpicking the methodological issues exposes the political.

    From a scientific method view point, as many have said, any hypothesis needs to be testable. For the sake of exposition let me frame up something that I’ll characterise as the AGW proposition (this is similar to Steven Mosher May 16, 2011 at 11:57 pm):

    “Delta ACO2 Granger caused more than 50% of delta SST over the 20th century” [H1:AGW]

    (Delta being the increase, ACO2 being man made CO2 in the atmosphere, Granger causality for testability)

    Note that this doesn’t say anything about the risks to humanity, but basically if this proposition doesn’t hold true than that becomes a bit academic. One can argue about other gases, “50%”, SST etc etc but for the purposes of exposition this will do.

    Now hypothesis testing under the scientific method would have us test how the null hypothesis associated with this proposition coincides with the observed data (NB: observed data not the output from GCMs). If the odds that the data doesn’t fit with the null then we say we reject the null at that level of confidence, and this gives us confidence in H1:AGW.

    I note in passing that the confidence level is not fixed and often chosen as a matter of convention. In this case it is quite legitimate to argue that because of the consequences of H1:AGW being true we should accept a low probability (perhaps 66% rather than the more typical 90 or 95%). I also note that being a Bayesian or otherwise just changes the way you calculate the probabilities.

    So what does H0:AGW look like? As many have observed H1:AGW is a complex proposition. It asserts among other things that delta CO2 Granger causes delta SST (and is >= 50% of the forcings), delta ACO2 >= 50% of delta CO2 and in fact is of a sufficient proportion to compensate for the balance of the forcings.

    So too therefore is H0:AGW complex. To rigorously falsify H0:AGW one needs to falsify delta CO2 doesn’t Granger cause delta STT, falsify that delta ACO2 is 50% of delta CO2 (if not higher), falsify other forcings don’t Granger cause more than 50% of delta SST etc.

    Note that these various sub H0:AGWs each need to be falsified, they are not necessarily independent, and their statistical distributions are likely to be complex, so it isn’t immediately obvious how to get 66/90/95% confidence in the rejection of the composite H0:AGW.

    Now a charitable interpretation of what Dr T is saying as he quotes IPCC is that H0:AGW has been rejected at the “very likely” level i.e. 90%. In as much as the bulk of the evidence relates to the output of GCMs, in a strictly scientific sense this H0:AGW hasn’t been rejected on the evidence presented in the IPCC reports. Dr T himself is saying that the next IPCC report will be less confident about its pronouncements (ironically because GCMs are now including more of the uncertainty).

    Let us now consider the sceptical position.

    If you are sceptical about the H1:AGW then all you are saying is “not proven”. A sceptic doesn’t consider that H0:AGW has been rejected with 90% confidence, that is all.

    From my own reading of the literature I’d have to say that this is a pretty reasonable place to be. Unlike Fred Moolten on the first comment on this post I do think the protagonists should be forced to state their hypotheses in testable terms; that there is no reason why this shouldn’t be possible (good science demands it); and once done the issues resolved at least to the point of agreeing “not proven” and a programme of research to help resolve the issues at hand.

    But Dr T is not of this mind. He wants to avoid that debate. Instead he wants to debate with those who perhaps wish to assert H1:~AGW (i.e. AGW is untrue). This could perhaps be characterised (again for the sake of exposition) as:

    “Delta ACO2 Granger caused less than 50% of delta SST over the 20th century”

    Note that this is quite different from sceptic’s position. This requires the corresponding H0:~AGW to be rejected at the (say) 90% confidence level. There is a lot of room for the sceptics between this position and H1:AGW (it obviously narrows as the confidence levels reduce, but so does what is being claimed on either side).

    One final point is that H0:~AGW is easier to falsify than H0:AGW. H0:~AGW can be reduced to a series of subsidiary H0s, but unlike H0:AGW falsifying one does for H0:~AGW. Simply falsify “delta CO2 Granger causes delta SST” or “delta ACO2 > 50% delta CO2”etc and H0:~AGW has gone.

    Just to reiterate, a careful analysis of the scientific methodological issues exposes most of the politics of Dr T’s pronouncements, and draws into question his competence in this domain, and hence his right to be taken seriously as a political voice on matters climate.

    • HAS thanks much for this

    • HAS

      We need more like this. (Though I admit being a most egregious offender in failing to adhere to such rigor very often.)

      Could you expand on the choice of “50%” in “..more than 50% of delta..” as opposed to any other figure, given the presumed increasing and cumulative nature of the model proposed?

      I could see replacing 50% with, say, 20% in all the above hypotheses and just accept a 60% middle ground where we accept the science is not ready to decide, if we’re really doing this for science, not policy. (Plus, it would be a ready analogy for the Italian Flag.)

      Given that the consequences argument that applies to accepting sigma-1 CI would apply equally here. (And some suggest, would apply in either direction.)

      Indeed, I would think knowing the various % of deltas for each CI would be more informative than a straight quest to satisfy the 50% accept/reject test.

      Could “Delta ACO2 Granger caused more than 5% of delta SST over the last 40 years” be shown at 95% confidence while “Delta ACO2 Granger caused less than 50% of delta SST over the last 40 years,” given the consequences, one might say the Trenberth conditions are met, no?

      • Sorry the 50% was just plucked out of the air for the purpose of exposition (and using it > and 10% of SST” with 1% confidence doesn’t tell us much either (although I’m not sure I’d necessarily sign up for even that :))

      • Ah ha it’s the use of the less than sign that causes the content to go. What I meant to say was:

        Sorry the 50% was just plucked out of the air for the purpose of exposition (and using it “more than” and “less than” respectively for H1:AGW and H1:~AGW an unnecessary confusion). These could be anything – in fact you could postulate a H1 as ACO2 as causing more than 60% and less than 65% if you so wished. If you widen the gap between H1:AGW and H1:~AGW then as you infer there is more room for the sceptic.

        Obviously the tighter the specification the more information generated (in a sense) by knocking over the null. Also there is a trade off between the tightness of the H1 and the confidence that can be achieved.

        This leads naturally I think to a discussion about the utility of the information generated (and this isn’t just utility in a policy sense, it includes utility for the advancement of scientific understanding). It isn’t very much use to disprove the H0:”No human influence on the climate” which Dr T sets up as the pre-2007 IPCC report null, because this tells us next to nothing. Similarly proving H1: “ACO2 caused more than 10% of SST” with 1% confidence doesn’t tell us much either (although I’m not sure I’d necessarily sign up for even that :))

      • HAS

        You got to the cool part of the reasoning.

        Doing it this way — estimating the utility of the information we have and the value of the information we lack, in such detail, makes for a specific bottom line on the value of investigations to provide that information.

        You’re never going to know if you ought get someone to pay for a full-on gridded, certified, validated, verified climate multimodal observation system on the loosey-goosey alarmism-rejectionism ballot – the data gathering structure minimally needed for the +/-AGW science. You need well-defined value costed-out for well-defined information to persuade those who care.

        Or, you could do it my way, but I’m not sure I can convince people to simplify the problem to its roots and just stop there as we have ample information already to see the reason for making it extremely expensive to emit CO2E, and extremely lucrative to those who own the right to breath.

        People like to know more than it takes to satisfy the conditions of a decision.

    • “I do think the protagonists should be forced to state their hypotheses in testable terms; that there is no reason why this shouldn’t be possible (good science demands it); and once done the issues resolved at least to the point of agreeing “not proven” and a programme of research to help resolve the issues at hand….But Dr T is not of this mind. “

      Trenberth does appear to do what you state – see Comment 69380. I think it would be unwise to underestimate him.

      • Peter Wilson


        The comment you quote does not even come close to stating any kind of testable hypothesis. Further, he makes the classic logical error of “begging the question” or assuming the proposition as part of the “proof” when he says Given that global warming is “unequivocal”, and is “very likely” due to human activities to quote the 2007 IPCC report, in addition to the obvious argument from authority.

        I wouldn’t overestimate him either.

      • Fred Moolten May 21, 2011 at 10:07 pm

        I had seen your comment, but rather than worrying about Dr T setting some measurable hypotheses, I’d welcome it. Somehow I don’t think it will happen because if you look at the type of science he does it is predominantly descriptive, not empirical.

        You can also see that he doesn’t have an instinctive feel for this stuff in the piece you quote.

        His stated concern is: “As a whole the community is making too many type II errors” (although how he knows the community is making too many type II errors rather begs the question when he can’t find an experiment to demonstrate that the null should be rejected).

        There is only one way to fix type II errors: better data and cleverer experiments (and be careful not put too much weight on the H0 on the basis that you can’t reject it – more of this below).

        What’s his prescription? Change from old H0 “Precipitation hasn’t increased” to new H0 is “Precipitation has increased by 5%”.

        But all he has done in moving from old H0 to new H0 is choose a Ho that makes it impossible to add weight to his desired H1: (“Precipitation has increased by 5%”). This for the rather obvious reason that his new H0 is his desired H1 – duh!.

        As I said, he has no feel for the subject, and getting him to start to write down a few experimental hypotheses and testing them would be a good start.

      • apologies, got caught in spam

      • “he has no feel for the subject”

        HAS – As someone with at least a bit of familiarity with the subject, I would say that approaching a discussion with Trenberth assuming he lacks a feel for the subject is to invite an unwelcome surprise. You are certainly right that he needs to specify the quantitation more than he has, and he may in his WIREs article. I don’t think the kind of metric discussed above, involving CO2 and SST in terms of a lead/lag
        relationship is practical, given the actual climate record.

        My own perspective has been that the upcoming pair of articles will be informative to the extent that they discuss what I expect will be Trenberth’s main focus – the evidence that convinces him the burden of proof should shift from whatever the current nebulous null is to one that presumes a quantitatively meaningful effect. I expect he will address hypothesis testing, but it will be the evidence rather than the statistical concepts that I’m guessing he will emphasize. To me, that would be the more meaningful type of discussion.

      • Fred,
        Have I missed your showing us the climate mitigation policies and programs that have worked?

      • Fred Moolten

        You wrote:

        My own perspective has been that the upcoming pair of articles will be informative to the extent that they discuss what I expect will be Trenberth’s main focus – the evidence that convinces him the burden of proof should shift from whatever the current nebulous null is to one that presumes a quantitatively meaningful effect. I expect he will address hypothesis testing, but it will be the evidence rather than the statistical concepts that I’m guessing he will emphasize.

        You apparently feel that Trenberth will attempt to support his proposal with evidence.

        My own perspective on this is that it would be a significant departure from earlier work by Trenberth if he really came with empirical evidence (i.e. data based on actual physical observations, rather than simply model simulations) to support his position.

        But let’s wait and see which of us is right.


      • Fred,
        So Trenberth is not going to defend his inversion of the null? He is just going to make appeals to authority?
        If he waves his arms fast enough, will it create a cooling breeze?

    • HAS, I enjoyed your piece and agree except for the “consequences of AGW” phrase.

      In this case it is quite legitimate to argue that because of the consequences of H1:AGW being true we should accept a low probability (perhaps 66% rather than the more typical 90 or 95%).

      In the context of your ACO2 Granger specification,
      AGW becomes defined as a causal connection between ACO2 and SST of significance The spec not speak to whether AGW may or may not be beneficial and welcomed. AGW may be just the thing we need to hold of the next Ice Age, for instance.

      There are a whole chain of hypotheses between
      “AGW” and “the world is coming to an end. Submit to Central Authority CAGW” In a sense, that is MacViolinist’s point (5/22 10:20pm) about connecting the p to the q.

      The H1:”AGW equals net harm to the biosphere” is a hypothesis that also needs to be tested and it’s H0 falsified. Other hypotheses must fall, uncluding those associated with precreiptions.

      Since there will be trillions of dollars involved and the loss of freedom of billions of people, I’d argue that HIGH levels of confidence, not Low, be demanded of the proponents. Prob 66% is little better than flipping a coin. The stakes of the game are far too high to flip a coin.

      Otherwise I enjoyed your piece. It is not enough to reject a strawman hypothesis. You must be able to replace the H0 with a testable, quantifiable H1 that is better than competing theories.

  93. MacViolinist

    One of the things that bothers me about the science at hand is that for the null/alternative hypothesis paradigm of research to work, there needs to be a clear connection between the two.

    It’s unsettling enough to me that a lot of “hypotheses” aren’t quantified, and therefore can’t really be tested. But it’s even more unsettling to me that the relationship of the null and alternative aren’t really binary relations.

    It always bothered me in my symbolic logic classes to accept the form “if p, then q” as true regardless of the content that p and q represent. In some cases it seems that climate science wants to push and pop from one level of meaning to another as often as is convenient.

    In a simple statistical model, you take a null hypothesis, “there is no significant difference between these two sets of numbers.” And that either turns out to be true, or it turns out that there is a significant difference. But as every textbook warns, not every significant difference is important, and not every lack of significance in unimportant.

    But at least in a statistical test the options are binary. Only one can be true–according to the selected test. In a symbolic logic situation where you have assigned semantic meanings to p and q, things can go quite far awry. You can say that if the grass is green, the moon is made of puppydog dreams. If it’s spring time in North Dakota, p is true and q should also be–even though it isn’t. If it’s summer in Texas, p is false and q should be false–which it is.

    I think what happens is that because it’s so difficult to quantify hypotheses in climate science, people make chains of small tests and assign them as a group to a symbolic p with an arbitrary q that they want to prove.

    Formally we have something that looks like it should be valid. But what we end up with is this:

    If (h1 + i1 + j1 + k1 + l1 . . . n1) [where h,i,j are all alternative hypotheses to various tests that, when combined, constitute a p] then, q.

    But when you substitute semantic meaning into the formula what you get is, “If the grass is sometimes not green, the temperature is sometimes significantly different, the CO2 levels are significantly higher, ocean levels are fluctuating over time, and we sometimes have higher instances of destructive weather events than others, then Humans are destroying the earth.”

    I don’t see any necessary logical connection between any of the smaller steps to the larger conclusion. The problem, as I see it, is not that any of the smaller notions being assigned to a symbolic p are really debatable. The question is the relationship between the p and the q.

    Flipping the null hypothesis, such as it is, really amounts to saying, “Disprove p so that you can falisfy q.” That’s really a very different problem, and one that lies outside the realm of null vs. alternative hypotheses. But no once really disagrees that much about p. The whole disagreement lies in the relationship between p and q.

    I realize this is tangential to your question, Dr. Curry. But I would say that focusing on a null hypothesis is going to be a challenge when faced with a philosphical assertion like that. Perhaps a fruitful area to investigate is what kind of hypothesis
    you could formulate that would combat such a formal system.

    • your statement is spot on:

      “I don’t see any necessary logical connection between any of the smaller steps to the larger conclusion. The problem, as I see it, is not that any of the smaller notions being assigned to a symbolic p are really debatable. The question is the relationship between the p and the q.”

      I have been indirectly raising this issue on some previous threads, see reasoning about climate uncertainty and waving the italian flac.

      • MacViolinist

        Now that you mention it, I remember your posts about uncertainty and the Italian flag. So perhaps my points are not useful to you.

        I might suggest an Aristotelian addendum to the scientefic process: hypotheses need to be related to each other as a part of a syllogistic paradigm.

        The point of an if-then statement is that it formally gets around the problem of the excluded middle. It assumes an appropriate middle.

        If we were forced to deal with the middle, we would have to account for the fundamental relationship betwen the p and the q. If we accepted a syllogism as the basis for our conclusions, we could then turn our attention to the truth of the terms in question.

        Another way of saying that is that we need use inductive reasoning to create the foundation of the premises we use in deductive reasoning.

        We need both types of reasoning to be successful scientists. We just have to make sure that we don’t confuse an assumption with a hypothesis, or confuse a hypothesis with a truth.

        And you should never encourage me with positive feedback.


      • MacV your posts are definitely useful to me! (+ feedback :))

  94. ferd berple

    The IPCC hypothesis is that human produced CO2 is the main driver of climate change (global warming). The null hypothesis would be that it is not the main driver.

    Under the rules of science, the burden is on science to prove CO2 “guilty”. CO2 remains innocent until proven guilty.

    T’s argument is that since we have “almost” proven Co2 “guilty” that should be good enough. The burden should now shift to “deniers” to prove CO2 “innocent”.

    The problem for “T” is that “almost” proven is not proven. If you accept that:
    “almost proven” = “proven”, then by substitution:

    “almost almost proven” = “proven”
    “almost almost almost proven” = “proven”

    and at the limit with:
    “not proven” = “proven”

    So, if you accept T’s argument, you end up with a logical contradiction. You end up with:
    “science” = “non science” = “nonsense”

    • fred,
      That is an excellent summary.
      A conclusion by corallary is that:
      Trenberth = scientist = nonsense.

    • Strictly, the only proof is to wait for it to happen. You can’t prove a future.

  95. I’m surprised that in this discussion of the “null hypothesis” there is no mention of the “unit root”.

    Was it only last year the VS thread was all over the climate sites?

    It seems to be before we can discuss statistical tests of any data set for the significance of any hypothesis, we must test the data itself for various characteristics. And if the data can’t be distinguished from noise — white, red, or some other flavor — then it’s unlikely that it tells us anything useful about any underlying hypothesis

    • Pouncer

      Yes. I recall the long thread with VS explaining the basics of econometrics (unit root, random walk, etc.) to a group of physicists and engineers, some of whom (like me) were having trouble following the discussing.

      VS had pretty much everyone running around in circles.

      The logic and references he cited pointed toward the global temperature record 1880-2009 being more of a random walk than a clear correlation between atmospheric CO2 and temperature.

      That’s more or less what I got out of it as a “lurker”.


      • It’s funny how these things pop up.

        Bishop Hill in his latest post links thorough to Doug Keenan’s piece a version of which was apparently recently published in the WSJ. This argues that global temperatures aren’t AR(1) because ARIMA(3,1,0) is superior, and from memory VS also argued using this model that the AR1 assumptions of the IPCC were rejected.

        This is directly relevant to this thread because Dr T uses IPCC’s analysis to state:

        “Given that global warming is “unequivocal”, and is “very likely” due to human activities to quote the 2007 IPCC report, the null hypothesis should now be reversed … “

  96. The IPCC may have “proved” that the planet is warmer and that man was at least partly responsible, but this doesn’t prove that ANYTHING else changed. Trenberth is saying that we have a choice of two hypotheses: 1) Man is responsible for a particular climate phenomena, making the null hypothesis that man isn’t responsible. 2) Man isn’t responsible for a particular climate phenomenon, making the null hypothesis that man is responsible. Before we can propose either hypothesis, we need to know that a particular climate phenomenon actually exists – that something different is actually happening. Only the traditional null hypothesis rules out the possibility that we are speculating about nothing.

    Evidence for anthropogenic global warming is not proof of anthropogenic “climate change”. Global warming (0.7 degC) and man’s estimated contribution (0.4-0.7 degC) is small compared with the natural variation in mean annual temperature AT ANY PARTICULAR LOCATION. The standard deviation in mean annual temperature at any particular location in the US is about 1 degC. The average location therefore has NOT actually experienced an increase in mean annual temperature clearly outside the range of normal variation for that location. Furthermore, given the regular seasonal variation in climate, variations in climate are experienced on a monthly or seasonal basis, not an annual basis. For example, an unusually cool April may delay planting crops or an unusually warm summer may stunt their growth. The standard deviation for the average location appears to be about 1.5 degC for seasonal temperature and 2.0 degC for monthly temperature. (All of these standard deviations were estimated by eyeballing city records at Mean global warming has been trivial compared with normal climate fluctuation.

    Another way of illustrating the trivial nature of the “climate change” produced by current global warming is to convert this warming into distance. In the continental US, the mean annual temperature increases about 1 degC for every 100 miles one moves south. Anthropogenic global warming has resulted in climate change equivalent to moving everything 40-70 miles (65-110 kilometers) to the south.

  97. Judith, for me there are really two opposing views of the climate:

    1. That the climate is mainly predictable and that its behaviour is the sum of known parameters with a small amount of (irrelevant) noise.
    2. That the climate is main unpredictable and that is behaviour is best modelled as a chaotic unpredictable system with small predictable perturbation.

    Those with a science background (and a poor one at that with no understanding of scientific methodology) tend toward the first view, those from an engineering or real world experience in the application of science tend toward the second view.

    Those that adopt the first view start with the idea that the climate is the result of known parameters so see the task as simply deciding how much each known parameter contributes to the apparent changes of the climate.

    Those that adopt the second view, start with the notion that the climate varies naturally, and then look for tests to see how much any particular change in the climate can be attributed to known perturbations.

    In essence the real null hypothesis of one view is the other. The null hypothesis that the climate can be modelled is predictable from known parameters is that it is not, and the null hypothesis that it is not predictable is that it is.

    So, there really is a lot of hypocrisy regarding the null hypothesis by sceptics. But where this is valid is where the “scientists” claim their assertions are “science” and have the certainty of real science when they do not subject these assertions to the rigorous tests of the scientific methodology.

    But I have to admit I have some sympathy with those honest individuals working in climate science because the null hypothesis is a rod to beat their backs with, but one which is not only difficult to apply, but where the relentless requirement of the rigorous scientific methodology would so limit the work that there would be very little any climate “scientist” could assert because so little planetary physics can be tested within a reasonable budget, within the ethics of not unduly affecting the planet and not within a reasonable time period.

    So, just as in the law we have two criminal procedures: criminal law where the requirement is “proof beyond reasonable doubt” and civil law “on the balance of probabilities”, I personally think it is time we had Science (proof by experiment, the null hypothesis – proof beyond all reasonable doubt) and “science” (soft science) … where assertions are made using rigorous assessment of the data and the application of known scientific principles, but assertions have to be made which cannot be subject to the full rigours of the scientific methodology.

    But perhaps I’m being too lenient on climate “science”. Perhaps the truth is that the appropriate solution for the subject which people like Mann and Hansen want to engage in is: “climate forecasting”. Perhaps the real division needs to be to create a subject which is the application of climate science – to create clear (melted ice) between climate science which is based on the scientific methodology and the null hypothesis and “climate prediction” where the “best” predictions are made based on the balance of evidence but there is no pretence that these predictions have or even can be tested (except by comparison to what happens … which I have to point out isn’t climate “science’s”/forecasters strong card!)

  98. Greg Jackson

    I’m not a scientist but couldn’t one accept Dr T’s null hypothesis and argue as follows:
    Untestable Null Hypothesis: Humans are dangerously influencing climate.
    Testable version of Null Hypothesis popular in recent years: If humans do not change current CO2 emission rates, the earth climate will experience a “hockey stick” increase in temperature.
    Experimental Test: Do nothing for 10 – 12 years.
    Experimental Result: No hockey stick.
    Conclusion: Reject Testable Null Hypothesis.
    Extension of conclusion to Untestable Hypothesis: Reject Untestable Null Hypothesis.
    Press Release: Science proves humans do not dangerously influence climate.

  99. Greg Jackson: The correct NUL hypothesis is that rising atmospheric CO2 does NOT cause significant global temperature increase. Trenberth does not understand that a nul must have the word NOT in it, amazingly that is what “nul”means. Now all he and his minions have to do is show the statistically significant coefficient from regression of changes in GMT with respect to changes in atmospheric CO22 which would if they can do it invalidate the NUL. They have not and cannot, because there is no such regression which passes all tests for spurious correlation (eg Durbin-Watson and collinearity). That is why no such regressions are reported in AR4 WG1.

  100. Willis Eschenbach

    HAS | May 21, 2011 at 7:19 pm

    <… From a scientific method view point, as many have said, any hypothesis needs to be testable. For the sake of exposition let me frame up something that I’ll characterise as the AGW proposition (this is similar to Steven Mosher May 16, 2011 at 11:57 pm):

    “Delta ACO2 Granger caused more than 50% of delta SST over the 20th century” [H1:AGW]

    (Delta being the increase, ACO2 being man made CO2 in the atmosphere, Granger causality for testability)/blockquote>

    I understand that that is made for the sake of exposition. However, there is a trap in it that you likely know but that readers should be aware of.

    It’s impossible in a general way to show that “A causes B”. However, a more restrictive test, referred to as “A Granger causes B”, is possible. The essence of the test is predictability. If you can do significantly better at predicting the future state of B by including A in your explanatory variables, then A is said to “Granger cause” B.

    Here’s the interesting part. There are not two, but three possibilities for Granger causality. The first two are obvious:

    A Granger causes B, or

    B Granger causes A.

    The third one is a stumper. It is that:

    A Granger causes B AND B Granger causes A

    When I analyze ∆CO2 vs ∆Global Temp, I get the third possibility. They both Granger cause the other.

    And that’s a whole separate subject, I just wanted folks to be clear that despite mathematical rigor and clarity that Granger causality itself isn’t all that clear, and because it’s not two-valued but three-valued, it’s not really suitable for a null hypothesis unless you include special clarifications … if that’s clear.

    I was looking yesterday at which of my writings on Watts Up With That had attracted the most page views. To my great surprise, it was my essay on Trenberth and the null hypothesis.

    However, your take is much more finely nuanced. Many thanks for your interesting contributions to the thread.


  101. Willis Eschenbach | October 25, 2011 at 3:06 pm |

    A Granger causes B AND B Granger causes A

    Haven’t you described Gallopin’ Granger Runaway there? That it hasn’t happened dis-confirms (falsifies) that option.

  102. (-)(-A) = A. Reversing the null hypothesis results in the hypothesis itself. It is the same as A=A, or verbally, human influence exists because human influence exists.

    Trenberth is demanding that scientists reject basic symbolic logic in order to promote a particular hypothesis. How long before the educated public’s cynicism turns to outright mockery?

  103. Great items from you, man. I’ve keep in mind your stuff prior to and you are just extremely excellent. I really like what you have got here, really like what you are stating and the best way by which you are saying it. You’re making it enjoyable and you continue to take care of to keep it smart. I can’t wait to learn far more from you. That is actually a wonderful site.