The Uncertainty Monster

by Judith Curry

Notions of uncertainty range from everyday usage in common parlance to specific definitions appearing in the philosophical and scientific literature.

Uncertainty can prevail in situations where a lot of information is available, and new information can either decrease or increase uncertainty. New knowledge of complex systems may reveal the presence of uncertainties that were previously unknown.

A sustained and systematic enquiry of how to understand and reason about uncertainty in climate science has not been undertaken by either climate researchers or philosophers. Such enquiry is paramount because of the challenges to climate science associated with the science-policy interface and its socioeconomic importance.

I first voiced my concerns about the way uncertainty is characterized in assessment reports in 2003, at a meeting of the NRC Climate Research Committee (CRC).  The CRC subsequently held a workshop on the topic of uncertainty and climate assessments (in 2004), but the emphasis was on communicating uncertainty to decision makers and role of uncertainty in decision making.  My concerns about this issue continued to grow, and expanded to include concerns about how we reason about the uncertainties in a complex system.  Last spring, I was fortunate to be invited to attend the Royal Society’s Workshop on Handling Uncertainty in Science.  This Workshop was seminal in catalyzing and crystallizing my thoughts on the subject.

My reflection on the topic of uncertainty has culminated in a manuscript entitled “Climate Science and the Uncertainty Monster”  (in collaboration with Peter Webster), which is currently winding its way through the publication review process.   Drawing on material from this paper, this post is the first in a series that explores ways to understand, characterize, and reason about uncertainty in climate science and assessments of climate science.  This post introduces the “monster”, describes a systematic taxonomy of types and levels of uncertainty, and discusses monster coping strategies at the science-policy interface.

Introducing the uncertainty monster

The “uncertainty monster” is a concept introduced by van der Sluijs (2005) in an analysis of the different ways that the scientific community responds to uncertainties that are difficult to cope with. A monster is understood as a phenomenon that at the same moment fits into two categories that were considered to be mutually excluding.   The “monster” is therefore the confusion and ambiguity associated with knowledge versus ignorance, objectivity versus subjectivity, facts versus values, prediction versus speculation, and science versus policy.  The uncertainty monster gives rise to discomfort and fear, particularly with regard to our reactions to things or situations we cannot understand or control, including the presentiment of radical unknown dangers.

Uncertainty lexicon

The lexicon of uncertainty terms that describe the nature and levels of uncertainty follows  Walker et al. (2003) and Petersen (2006).

Nature of uncertainty

The nature of uncertainty is expressed by the distinction between epistemic uncertainty and ontic uncertainy.   Epistemic uncertainty is associated with imperfections of knowledge, which may be reduced by further research and empirical investigation.  Ontic (often referred to as aleatory) uncertainty is associated with inherent variability or randomness. The distinction between these two types of uncertainty is useful in science because each entails different conclusions regarding the reducibility of uncertainty.  Ontic uncertainties are by definition irreducible, while epistemic uncertainties are in principle reducible by further research and empirical investigations.

Epistemic uncertainty of the state of the climate system includes uncertainty due to limitations of measurement devices, insufficient data, systematic error and the subjective judgments needed to assess its nature and magnitude, extrapolations and interpolations, and variability over time or space. Uncertainty in empirical quantities can also arise from disagreement among different experts about how to interpret the available evidence.  There can also be epistemic uncertainty about how a physical, chemical or biological process works.  Epistemic uncertainties in global climate models include missing or inadequately treated physical processes, uncertainty in the numerical value of physical parameters, discretization and algorithmic approximations, and uncertainty in the specification of external forcing.

Ontic uncertainty in climate science derives from the complexity of the climate system and indeterminacy of human systems. Natural internal variability of the nonlinear climate system contributes to ontic uncertainty in climate simulations.  The climate system is stochastically uncertain because of its chaotic nature, i.e. small differences in the initial conditions of a global climate model can yield very different results. Scenarios of global greenhouse gas emissions are inherently uncertain because they depend on human behaviour, e.g. the uncertainty of the future fertility rate and future economic development.  Initial condition uncertainty is partly epistemic (inadequate and incomplete observations) and partly ontic (chaos).

Level of uncertainty

A spectrum of knowledge and uncertainty exists, ranging from complete deterministic understanding to total ignorance. Total ignorance implies a deep level of uncertainty, to the extent that we do not even know that we do not know. Walker et al. (2003) characterizes the levels of uncertainty as a progression between deterministic understanding and total ignorance:  statistical uncertainty, scenario uncertainty, and recognized ignorance.

Statistical uncertainty can be described adequately in statistical terms. An example of statistical uncertainty is measurement uncertainty, which can be due to sampling error or inaccuracy or imprecision in measurements.

Scenario uncertainty implies that it is not possible to formulate the probability of occurrence of one particular outcome. A scenario is a plausible but unverifiable description of how the system and/or its driving forces may develop in the future.  Hence the use of scenarios is associated with greater uncertainty (more ignorance) than statistical uncertainty.  Scenarios may be regarded as a range of discrete possibilities, often with no a priori allocation of likelihood.  An example of scenario uncertainty of relevance to climate science is associated with future greenhouse gas emission scenarios used to force global climate models.

Recognised ignorance refers to fundamental uncertainty in the mechanisms being studied and a weak scientific basis for developing scenarios. Reducible ignorance may be resolved by conducting further research, whereas irreducible ignorance implies that research cannot improve knowledge.  An example of irreducible ignorance is what happened prior to the big bang or what is happening beyond the cone of observations defined by the speed of light.  Border with ignorance idenotes knowledge of the presence or possibility of ignorance.

Levels of uncertainty cannot be described completely without reference to Donald Rumsfeld’s infamous statement:

“[A]s we know, there are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are some things we do not know. But there are also unknown unknowns — the ones we don’t know we don’t know. And if one looks throughout the history of our country and other free countries, it is the latter category that tend to be the difficult ones.”

With respect to the Walker et al. classification, a known known encompasses both deterministic certainty and statistical uncertainty. The known unknowns encompass scenario uncertainty and also situations where there is some sense of the shape of the probability distribution but knowledge of key parameters is lacking.  The unknown unknown corresponds to ignorance, including border with ignorance where the unknowns are perhaps suspected.  Unknown unknowns may be the targets of frontier research or philosophical speculations, or may correspond to future circumstances or outcomes that are impossible to predict or to know when or where to look for them.

Monster coping strategies

An adaptation of van der Sluijs’ strategies of coping with the uncertainty monster at the science-policy interface is described below.

Monster hiding. Uncertainty hiding or the “never admit error” strategy can be motivated by a political agenda or because of fear that uncertain science will be judged as poor science by the outside world.  Apart from the ethical issues of monster hiding, the monster may be too big to hide and uncertainty hiding enrages the monster.

Monster exorcism. The uncertainty monster exorcist focuses on reducing the uncertainty through advocating for more research. A growing sense of the infeasibility of reducing uncertainties in global climate modeling emerged in the 1990’s, in response to the continued emergence of unforeseen complexities and sources of uncertainties. Van der Sluijs states that:  “monster-theory predicts that [reducing uncertainty] will prove to be vain in the long run: for each head of the uncertainty monster that science chops off, several new monster heads tend to pop up due to unforeseen complexities,” analogous to the Hydra beast of Greek mythology.

Monster adaptation. Monster adapters attempt to transform the monster by subjectively quantifying and simplifying the assessment of uncertainty. Monster adaptation was formalized in the IPCC AR3 and AR4 by guidelines for characterizing uncertainty in a consensus approach consisting of expert judgment in the context of a subjective Bayesian approach (Moss and Schneider 2000).

Monster detection. The first type of uncertainty detective is scientists that are challenging existing theses and working to extend knowledge frontiers.  A second type is the watchdog auditor, whose main concern is accountability, quality control and transparency of the science. A third type is the merchant of doubt, who distorts and magnifies uncertainties as an excuse for inaction for financial or ideological reasons.

Monster assimilation.  Monster assimilation is about learning to live with the monster and giving uncertainty an explicit place in the contemplation and management of environmental risks.  Assessment and communication of uncertainty and ignorance, along with extended peer communities, are essential in monster adaptation. The challenge to monster assimilation is the ever-changing nature of the monster and the birth of new monsters.

The climate uncertainty monster is too big to hide, exorcise or adapt.  In future posts I will present arguments for ascendancy of the monster detection and assimilation approaches.  Van der Sluijs argues that the climate assessment process should be open to both technical skeptics and the doubt merchants in terms of monster detection, “and that the unpleasant way in which the game is played and the mixture of valid and ungrounded criticisms that it produces is the price that has to be paid for the key advantage for quality control of the identification of weak spots in the knowledge base.”

The challenge is to open the scientific debate to a broader range of issues and a plurality of viewpoints and for politicians to justify policy choices in a context of an inherently uncertain knowledge base.

David Spiegelhalter provided the following wise words at the recent Workshop on Handling Uncertainty in Science, that can help tame the monster:

  • We should try and quantify uncertainty where possible
  • All useful uncertainty statements require judgment and are contingent
  • We need clear language to honestly communicate deeper uncertainties with due humility and without fear
  • For public confidence, trust is more important than certainty

Climate science, and particularly assessments of climate science such as the IPCC, needs to do a much better job of characterizing and reasoning about uncertainty.  The events of the past year that have challenged the credibility of climate science are symptoms of an enraged uncertainty monster.

235 responses to “The Uncertainty Monster

  1. It is not so much the uncertainty of science but the input that may show missed areas in that research field.

    Answers to nagging questions are just as easily right under your nose and looking at it everyday.

    Take a simple yet complex area like water. We see it and use it. IT IS COMPRESSED GASES. It is a state of stored energy through rotation generating pressure and combining it into a liquid. Trace elements attached form another function that interacts with the magnetic field.

  2. Judith, are there any of the IPCC’s SRES that can be numerically classified as more likely than the others? Could they be put in coherent segments ranked by likelihood?

    • Tom, the whole issue of scenarios is an interesting one. I will write much more on this in a subsequent post, I will argue that the scenario range in the SRES is far too narrow, and agree with Betz (in a paper that is unfortunately not available online) on the idea of modal falsification as the appropriate framework for scenario analysis.

  3. Not sure that the name “Uncertainty Monster” is helpful here.
    Although it is quite narrowly defined in this scientific context, it carries so many connotations with it for a lay reader that the specific elements that are evaluated here never reach the appropriate priority.

    Humility sufficient to admit that the conclusions that are reached may be very much mistaken would help improve the credibility of science research.
    Restoring science to its well accepted role as an adviser rather than trying to remake it as an advocate/decider/enforcer would also ease some of the current concern.

  4. Thank you, Dr. Curry, for having the courage to question the way uncertainty has been characterized in climate assessment reports.

    If your concerns had been seriously considered in 2003, a great deal of the damage to those championing CO2-induced global warming might have been avoided.

    Efforts to downplay or avoid mentioning uncertainty is one of three common indications that politicians and scientists have joined forces to promote an idea.

    Another sign of bureaucratic efforts to downplay experimental uncertainty is listing a large number of scientists as co-authors.

    A third sign of collusion in science is collective inability to see experimental data that disprove a cherished model, e.g., the correlation of primordial He with excess Xe-136 that was first observed in the Allende meteorite [“Host phase of a strange xenon component in Allende”, Science 190 (1975) 1251-1262]

    With kind regards,
    Oliver K. Manuel

  5. Excellent post. When we were young we all feared “the monster under the bed”. Many have not learned to live with monster.
    In the real world (such as gold mining) one learns to live with the error in the estimate exceeding the estimate. (0.06 oz/t Au+/- 0.07 oz/t Au) This is not unusual where the sample value distribution is lognormal.

  6. Judith,

    Thank you for a thought-provoking paper. I have nothing to add at the moment, other than to say, as someone who could call himself a ‘political scientist’, that virtually all government decisions are taken in the context of uncertainty. We rarely have all the data we need when we need to make a decision, and I cannot think of any decision of importance on areas that concerned me, when I was part of government, that did not have unintended and unforeseen consequences.

    To that degree I am sympathetic to those who feel that we cannot wait and wait to be sure about AGW. But in their case I would insist that they deal with the inconvenient questions that I and others would offer them, before governments act. They haven’t, and governments have acted, and the results have been deeply unimpressive at many levels.

    • Agree. Governments rarely ‘act’. Normal government movement, of any kind, is a ‘reaction’ to popular expectations (with the exception of China and a few others;-). The political ‘reactions’ that have been passed and implemented to date by major powers, are precisely ‘political bandaids’ that have really achieved nothing whatsoever; which, in actuality, was precisely what they were designed to do.

    • Don – I’m an Aussie too, so I agree that our Govts have not dealt with the uncertainty issue at all (rather the reverse, wherein all questioners are simply insulted out of hand)

      But the proposed “Parliamentary Committee” to deal with this issue is already compromised before starting to the point where dissent is simply not permitted

      I can see nothing but unintended consequences from this

      • Yes, that is alas the case. My suspicion, however, is that there will be a lot of talk in Australia, but not much action. The great problem for politicians who get into these binds, is that they create expectations that they then have to manage. While there is growing scepticism in Australia about the human contribution to whatever global warming is occurring, as elsewhere, there are very many people for whom this is all part of some kind of ethical or moral obligation on us ‘to save the planet’ or ‘combat climate change’. I doubt that their numbers will unless there is a renewal of warming, but they are unlikely to change their minds about it all. And they vote. And the either vote Green or they are sympathetic to the Green view. And the politicians of both sides who got us all into this mess, now can only deal with it by talking and posturing, but not by doing anything of consequence — as others have said on this and the other posts.

        Let all this be a warning to other scientists: ‘be careful what you wish for’.

  7. “A third type is the merchant of doubt, who distorts and magnifies uncertainties as an excuse for inaction for financial or ideological reasons.”
    What of the case where the doubt is truly great and the financial cost of mitigation high. The actor in this case does not have to distort of enhance the degree of doubt as it is great enough to delay mitigation due to the cost. In other words, why is it “the merchant of doubt” is described as distorting the truth about the uncertainty. Or maybe there is another category for what I’ve described?

    • intrepid_wanders

      I would bet that would be “The Honest Broker” (shill please Roger ;) )

      Ideally, this role should be played by an independent academia type(s) that is well versed in the technicalities. With the monies involved with anything new (expecially regulatory) you naturally have “corruption” on both sides of the “uncertainty”.

      It naturally manifests that the public will want to STOP everything and decide “What the heck IS going on?”. Even if a portion of the public is undecided and leaning towards action, the “Merchants of Doubt” on BOTH sides pressure harder and call you “ignorant”. Fights insue, nothing gets done, the MSM is no help. You just want to stop thinking about it.

      This is where:

      1. We should try and quantify uncertainty where possible
      2. All useful uncertainty statements require judgment and are contingent
      3. We need clear language to honestly communicate deeper uncertainties with due humility and without fear
      4. For public confidence, trust is more important than certainty

      …comes into play, working VERY HARD to become an ideal (The Honest Broker). This means fiscally responsible (Auditors needed), data and metrology responsible (QA and Auditors needed too) and archiving of data for future scholars, curious and auditors.

      I find Judith is definitely on a constructive path. Keep up the work, plenty of work to be done.

      • Given that it seems inevitable that science, where it might influence public policy, will be wed to scientists on the payroll and propagandists, I recommend a look at the legal system as a model for resolution of contending views by parties with a vested interest.

        In the legal system, hireling advocates are at the heart of the system and the best efforts of those hired-guns, through an open and free-wheeling dialectic, provides a pretty reliable method by which the truth is found. Of course, such a system would require a scientific judiciary that is independent and disinterested. And, naturally, the whole system should be open to the public.

        Nothing’s perfect, but when it appears that the science in a certain area has become corrupt and the peer review process either corrupt, as well, or dysfunctional, then appeal to a “science court” might make sense.

      • More lawyers? No, please. Before we do that let’s toss a coin. (But neither will work.)

        The ‘science’ must be resolved by scientists. Everything else is up to the majority and is a political process. If the decision in country ‘x’ is to wait and see, so be it. If the decision in another is to do something ‘green’, so be it. When you go from science to politics, you’re in a different game and kettle of fish, and there is no ‘right’ or ‘wrong’ answer. One of the big problems on the current ‘global climate question’ is the idiotic idea that there is one ‘global political’ answer, there definitely IS NOT.

      • Pascvaks:

        For what it’s worth, I share your horror of lawyers (I’m not one myself, thankfully). Unfortunately, it looks like the lawyers are already in the wire, based on various challenges to EPA’s carbon regulations.

        Better that legal challenges to science be settled in a venue that ensures the judge/jury has the genuine expertise to sort out the competing claims, I recommend. Hence my suggestion of a specialized “science court”–not necessarily a part of the formal judicial system, but one which employs courtroom like techniques of investigation. Such a “court” could even be an administrative institution of the science establishment that supplements the peer review process with scientists as the “lawyers.” In that regard, I think there’s a place for a vigorous, adversarial cross-examination of competing views (isn’t that the “thing” missing from the various climategate inquiries?) in instances where there appears to be a breakdown in conventional quality control mechanisms of a scientific field.

        I mean, what else do you do when there is a corruption of the scientific establishment, to include the peer review process. Start a blog? (Actually, I think that’s a good idea, too.)

      • I’m personally in favor of letting science guys work this out all by themselves. If they end up committing suicide, so be it. Say “La V”. But, as soon as you try to ‘reasonably’ let the legal system (with any new tweaks you want add) get their greedy little hands on anything like this fiasco, you –and not the scientists– are the one committing suicide. Lawyers and the courts are the last step before chaos and anarchy and I don’t see them as capable of solving ANYTHING regarding Global Climate or anything else “Global“. They’ve nearly pushed this country down the sewer as it is. (I’m old and skeptical and grumpy;-(

      • Nice thing about this blog is how many fellow old, skeptical grumps you meet.

      • To advance the openness and auditing function, why not have the Library of Congress publish all papers produced with any amount of Federal money thirty days after being published in a journal. Also, require any study using Federal dollars to disclose all data, metadata, and computer code and have it made public also so the audit can be thorough and the openness complete.

      • Jim, I like your idea.

  8. Very interesting. I like this concept very much.
    If I may be so bold as to offer a category you may have over looked:
    The Monster Promoter: Someone who magnifies and manipulates in the name of science the perceptions of the science or policy implications for their own fame glory or treasure or for their own cause’s victory.

    • In the context of this post, the Monster Promoter would be those who magnify and manipulate “uncertainty” for their own fame glory or treasure or for their own cause’s victory.

      • I was thinking of that same role. I would call that actor the “Merchant of Fear,” one who minimizes the uncertainly of global warming while greatly exaggerating the purported bad consequences for financial or political gain.

      • Hummmm… sounds plausible.

        Off hand I can’t think of anyone who has taken advantage of the situation and tried to make a name for themself,
        or profited in such a terrible way. (sarc off)

      • Well to balance the “merchant of doubt” I would instead use the term “merchant of certainty”: the merchant of certainty, who distorts and minimizes uncertainties as an excuse for action for financial or ideological reasons.

        This is one of the more irritating things for me as a denier not because I disagree with knowledgeable climate scientists but because i disagree with necessity for immediate policy action. But that’s a side issue.

      • Ted, I have actually used the term “uncertainty denier” on some occasions :)

      • Now *that* is an interesting turn of phrase.

    • Ron & Jim,
      Follow the money and power. Who has become rich and powerful in this issue?
      Hint: It ain’t the skeptics.
      And frankly, in the context of what Dr. Curry has written, the Monster Promoters are in fact using the doubt and uncertainty regarding AGW to do exactly what I suggest. If the science actually supported the extreme policies that have been or are being demanded to be imposed world wide, we would not see the large number of climate science meetings that involve marketing, re-branding, ‘communicating’, etc.

  9. Thank you, Dr. Curry, for introducing the discussion of uncertainty and questioning the way it has been used in climate science.

    I have been both an operational and research meteorologist for more years than I care to remember. Weather forecasting and research is based, in large part, on uncertainty. We know we don’t know all of the physics and non-linear interactions involved. We know the weather models are approximations to the real world, sometimes they are very good and sometimes they have serious failings. When we issue forecasts or publish research papers we try to convey our level of uncertainty. In this way the user/reader can, hopefully, make a more informed decision.

    I believe it is largely for this reason that many atmospheric scientists, myself included, cannot understand how climate scientists can portray an illusion of almost complete certainty in their models and projections. Surely they cannot believe their models are as infallible as they seem to imply when questioned.

    Weather forecasting is a very good teacher for anyone who thinks they have all the answers when it comes to nature. It induces a great amount of humility as you get rapid verification of your forecasting, sometimes within minutes. You soon learn that there are thing we know, things we think we know, and things we know we don’t know.

    Thank you again for opening up this interesting discussion.

  10. Uncertainty monster is a very interesting idea. It should be suitable to represent the uncontrollable property of the uncertainness. But, uncertain characteristics of Mother Nature might be better represented by “God (or Goddess) of uncertainty.” You can probably find a suitable name, for instance, in Greek mythology. This is becausethey say that the social condition of the Greek era is similar to the present, and hence, the risks of the present time can be named after Greek mythology, such as “Damocles type” and “Medusa type.” I call him/her Goddess U here temporarily. She is the mistress of the uncertainty monsters.

    Difference between the monster and the god/goddess is that human may be able to conquer the monster in future, but not the god. Isn’t it a good idea that Goddess U will be angry when humans neglect her or her men? Goddess U requires humans to respect her.

    This may have happened in the “events of the past year” (according to Judith) because a climatologist behaved as if she did not exist; that is, the climatologist has hidden the uncertain portion of data.

  11. Judith:

    Natural internal variability of the nonlinear climate system contributes to ontic uncertainty in climate simulations. The climate system is stochastically uncertain because of its chaotic nature, i.e. small differences in the initial conditions of a global climate model can yield very different results.

    I would disagree quite strongly with this block.

    First, I have seen no evidence of an influence from ‘internal variability’ on any timescale longer than a decade (i.e. the limit of ENSO effects). Unless you can back this up with published evidence, it should be withdrawn.

    Second, models of climate do not generally show the kind of extreme sensitivity to initial conditions that you would expect in a chaotic system – at least, not on the decadal timescales that matter. I think you may be confusing weather and climate here.

    Finally, whenever I read the IPCC reports I have to note the obvious and deliberate statements of uncertainty. The ‘uncertainty monster’ is hardly hiding in the closet; but I would add this: It works both ways. We could also by systematically underestimating the impact of adding carbon dioxide to the atmosphere.

    • Leonard Weinstein

      You seem to be ignoring the 60 year cycle of ups and downs clearly shown superimposed on the gradual rise following the little ice age to present to the period of recent interest. Also the several hundred to over 1000 years cycles shown by ice core data and sea floor cores show cycles of significantly larger variation in temperature, including greater highs than present. The lack of models showing large variability may be the problem in the models, not the lack of variability in the real world.

      • Fair enough, I’m ignoring them because I have not seen the references – can you supply them?

    • First, there is no need for Judith to withdraw anything. The honest truth is we simply don’t know what caused most of the climate fluctuations in the past. They may be related to some kind of ‘forcing’ or they may just be natural fluctuations of a complicated nonlinear systems.

      Second, crude climate models do not show chaos because they are too simple and overdamped. More sophisticated models do show chaotic behaviour, but not enough, because they all have too much damping (either ‘hyperviscosity’ or a value of viscosity orders of magnitude too large) and they have to omit many dynamical processes (such as convection).

      Finally, the uncertainty estimates in the IPCC reports have no real meaning – they are the subjective opinions of the carefully selected pro-AGW scientists that make up the IPCC. If you asked scientists excluded or forced out of the IPCC process (Lindzen, Pielke, Landsea, or even our own JC) you would get different answers.

      A relevant quote from statisticians McShane and Wyner:
      “Climate scientists have greatly underestimated the uncertainty of proxybased reconstructions and hence have been overconfident in their models.”

    • Andrew, lookup these terms… though some of the names speak for themselves
      Atlantic multidecadal oscillation (AMO) – 50 to 90 year cycle
      Pacific decadal oscillation (PDO) – 40 to 60 year cycle
      interdecadal Pacific oscillation (IPO or ID) – 15 to 30 year cycle
      Glacial and interglacial periods – thousands of years cycle

      • I am perfectly aware of these.

        However, none of them have any effect, as far as I am aware, on global temperature – with the obvious exception of the glacial-interglacial cycle. If you have evidence to the contrary then please show me.

        The glacial/interglacial cycle is not internally generated either; it appears to be driven by orbital changes, and acts on a much longer timescale than current climate changes.

        So.. I see no evidence that climate is chaotic, and I see no evidence of internally generated variability on a timescale longer than a decade. If people are going to assert either of these points, then they should be prepared to offer evidence, not assertions.

      • Andrew,

        If you just eyeball the glacial-interglacial chart, it isn’t a sine wave. The math would have to be done and may have been done, but just looking at it, it appears to be a chaotic system perturbed by a, possibly external, cyclical influence. The cycles are similar, but not identical anywhere.

      • Maybe this is what you guys should discuss

  12. Would it be fair to say, if the true uncertainties in the climate models were enunciated clearly, Kyoto would not have happened and Copenhagen would have been a side show attended by a handfull of people?
    “Come back when you are more certain” may have been the cry.
    But ofcourse, that wouldn’t have suited those who want action now and action aplenty.

    • Its more complicated than that. Policies were being enunciated back in 1992 (the UNFCCC) before there was any kind of uncertainty. Uncertainty should not preclude the consideration of a risk, particularly when the consequences of the risk are large. But the kinds of policies that were developed (Kyoto, Copenhagen) were not robust or politically viable.

      • …”(Kyoto, Copenhagen) were not … politically viable.”

        This point, obviously, has nothing whatsoever to do with ‘science’. AND, it is THE THORN in many an argument. The science is not settled (despite what some may claim) and toward a better and more complete understanding of ‘the science’ we should move. The ‘politics’ is no where near settled (and never will be in my humble opinion), and regarding this matter science should continue to advise and enlighten the public and the politicians but should not engage in partisan politics of any kind. I know this last point will anger many. But, as the old saying goes, ‘you can’t have your cake and eat it too’. The scientist who is not into politics remains a scientist, the scientist who jumps into politics is a political scientist. (PUN intended.)

      • Has to be said.. is there *any* policy that involves constraints on CO2 emissions that the USA and/or BRIC would actually accept?

      • Judith “before there was any kind of uncertainty”? I can not remember a time when there was “no kind of uncertainty”. I can, of course, remember a couple of decades when those expressing it, whether scientifically or polemically, were effectively silenced. I hope that’s what you meant.

      • Judith – sorry, that should have been “I hope that’s NOT what you meant”

      • Andrew Dodds says:
        September 23, 2010 at 10:01 am
        ‘Has to be said.. is there *any* policy that involves constraints on CO2 emissions that the USA and/or BRIC would actually accept?’

        No, there isn’t. And it’s past time that more commentators got their heads around this inconvenient truth. Perhaps then fewer of us would invest so much time in self-important but futile debate about something that we cannot change.

  13. Judith: “The climate uncertainty monster is too big to hide, exorcise or adapt. In future posts I will present arguments for ascendancy of the monster detection and adaptation approaches.”
    I think you mean “detection and assimilation” at the end there.

    This post is all very well as abstract philosophy, but I think you’re misapplying it to climate science. First point is that if you read the IPCC reports (and indeed just about anything else published in mainstream climate science), what you call “monster assimilation” just leaps out from every page. I find it hard to reconcile my conversations with climate modelers, and my readings of their papers with your assertion that the field has failed to assimilate the uncertainty. The problem as I see it isn’t how uncertainty is handled within the expert community of climate science, it’s with the communication of this uncertainty with lay audiences, and the fact that there are political forces at work who wish to distort the uncertainties to advocate various ideologically driven policies.

    So the question that you never get to is: “what’s an appropriate course of action, given the seriousness of the threat of climate change, and the inherent uncertainties in understanding it?”

    And the very serious problem that you haven’t acknowledged is that the discussion of uncertainty cuts both ways. Setting aside your quibbles over how they represent uncertainty, the IPCC assessments represent the current “best guess” of the climate science community for how earth systems will respond to each of a number of different future emissions scenarios. If their uncertainty is underplayed, then the logical conclusion is that they may be wrong in either direction – under or over-estimating the impacts of climate change.

    The problem, which is never acknowledged in the public discourse, is that if the IPCC assessments are a whole lot less certain than the summaries make out, then the risk of impacts far worse than the IPCC reports describe go up, including the likelihood that climate sensitivity is higher, the likelihood of reaching climate tipping points earlier than expected, and the likelihood that we’re already committed to more climate change than we thought. If we draw the probability density function, and then change nothing but our assessment of uncertainty, then greater uncertainty doesn’t affect the mean, but it flattens the curve, with a much bigger long tail in both directions. Good risk management practices say that in the face of higher probability of risk of catastrophic outcomes, you’d better damn well get your risk mitigation strategy in place early.

    The one thing we know for certain is that the climate system does respond to forcings in the form of higher concentrations of GHGs. Denialists seem to assume that the relatively stable climate we’ve seen in the past is a given, and uncertainty just means it won’t change much in the future. Climate scientists in contrast take the sum of evidence we have about responses of the earth system to generate best guesses about climate sensitivity and other indicators, and take it that uncertainty means they might be wrong in either direction. One of these positions is logically untenable. The thing that surprises me is that you, Judith, seem to lean towards the former rather than the latter in your assessments. If you really think the major problem with the IPCC process is under-estimation of uncertainty, then you ought to be advocating a more serious risk mitigation strategy.

    • Steve, this post is the first of about a dozen addressing uncertainty. Yes, we could be wrong in either direction. And also, I will have an entire post devoted to your last sentence. I am not advocating any kind of a risk strategy. I think the risk has been mischaracterized, too much focus on the mean (trying to figure out of CO2 sensitivity is 3.0C or 2.5C), and missing out what could happen on both ends (natural variability and catastrophic warming).

      • Judith: “I think the risk has been mischaracterized, too much focus on the mean (trying to figure out of CO2 sensitivity is 3.0C or 2.5C), and missing out what could happen on both ends (natural variability and catastrophic warming).”

        Surely you’ll acknowledge then that the main driver that has people like Jim Hansen and Stephen Schneider speak out publicly to push for strong, urgent action on emissions reduction is precisely their understanding of the uncertainties involved, and their realization that the probability of civilization threatening effects on the upper ends of catastrophic warming are far higher than people generally think, and certainly higher than the IPCC reports would lead us to believe? Hansen in particular has struggled with reconciling his accustomed role as a neutral scientist with his realization that people who understand the uncertainties need to speak out about the magnitude of the risks involved.

      • Scientists have an important tole to play in articulating the risk (which includes discussion of the uncertainties). Once scientists cross the line into advocacy (which Jim Hansen definitely has done), I personally don’t think this helps. With regards to the probability of catastrophic impact, the uncertainties are to large to develop any kind of sensible probability distribution; but one can articulate possibilities that are worse than the IPCC scenarios. I will be discussing the concept of of a possibility distribution and modal falsification method for scenario development in a future post.

      • Steve, one other point here. The characterization of advocates such as Hansen as green enviro extremists just isn’t correct. Their motivation seems to be more believing their computer model projections, and not just wanting to sit by and watch their predictions of calamity be realized (I saw a quote like this from Mario Molina, but can’t find it).

      • Judith,

        I find the idea that there is something wrong with scientists advocating policy to be quite odd.

        Coming from the medical field, I can’t imagine why medical researchers, with the best understanding of a particular health issue would be anything but well placed to comment on policy related to that issue.

        I think the view that scientists can have no role outside of the actual scientific process is mistaken on many levels.

      • Michael, being involved in the policy process and informing the policy process is very different from advocating for a specific policy. See Roger Pielke Jr’s Honest Broker.

    • Steve one more point. Yes the IPCC lists lots of uncertainties. Then they use words like “very likely” for their main conclusions. For example in the 20th century attribution, the confidence on the external forcings is mostly listed as low or even very low (except for greenhouse gas forcing), but the attribution to AGW is “very likely.” I will deconstruct this one in detail later in the series.

      • I wish you would. With special reference to the TAR and AR4 attribution of TSI as the dominant cause of the ~1910 – ~1940 warming.

        Once one recognises that the IPCC used obsolete TSI reconstructions to arrive at an incorrect attribution, one realises that we do not in fact know what caused the warming during those decades.

        Decades when atmospheric CO2 was below 300 ppmv and not capable of temperature forcing on the scale that actually occurred.

        One can then ask those who dismiss internal variability to suggest a possible alternative natural cause for that warming and explain why there can be no possible (but unknown) natural factor behind much of the recent warming.

        Moving on, one might ask for compelling empirical, not modeled, evidence for the supposed aerosol forcing that cooled the post-war decades until the Pacific Climate Shift in 1976.

        When none materialises, one might have to admit that the IPCC consensus on the attribution of climate change pre-1976 is without merit.

        From there, it is not much of a stretch to wonder if its assessment of anthropogenic cause for recent warming is similarly unreliable.

    • Leonard Weinstein

      Your argument is basically the precautionary principal. Unfortunately, it is not reasonable here. The probability that we are heading for a new little ice age and soon will be heading for a major ice age (we are near the end of the holocene if the last several interglacial periods are representative) is possibly greater than that we are headed for a hot problem. If that is true, doing something to stop warming would be a major error and have just the opposite effect as required, and at the same time negatively impact progress.

    • “Denialists seem to assume that the relatively stable climate we’ve seen in the past is a given”

      When has the climate been relatively stable? Even in the context of the Holocene, climate has been all over the place.

  14. Michael Larkin

    Find as many people as you can who understand climate science, but who are agnostic about AGW. They must genuinely not know whether the hypothesis is true or not, though it’s quite likely they might have a leaning one way or the other (which is my own stance).

    Charge them with looking at the current state of the science, identifying the key issues and commenting on the known knowns and the known unknowns. There may be some scope for commenting on suspected uknowns, but not, by definition, on unknown unknowns except to acknowledge they could well exist, maybe drawing parallels with historical examples where this is now known to be the case.

    If I could be confident such a group were truly agnostic, I would have a lot of trust in what they would say even if it gainsaid my current sceptical leaning. But they must only address the science, and eschew all policy considerations. It should be apparent why I have no faith at all in the IPCC process, not least because it starts with a predefined agenda, glosses over the null hypothesis, and filters out or downplays dissenting opinion. It pollutes the science with politics.

    I guess what I’m trying to say is that I want to see honesty and integrity in action. Along with humility, which has already been commented on, this is what one needs to truly deal with monsters. You can’t dispense with such basic qualities. In the end, whatever fancy rubric one uses, whatever philosophical approaches are taken, it will all come to naught without them.

    This is, of course, not going to happen. Some day, the truth, whatever it might be, will out, but I fear it will be despite, and not because, some elite group cares to take a principled (in the ethical sense) approach to addressing the matter. Despite the impossibility of hiding the monster, the temptation to keep on trying to hide it is too tempting. Human egos are involved, and next to mountains, they are the most stubborn things on earth. I know whereof I speak, because I too have an ego and know what a bugger it is to deal with.

    • That experiment has been done over the past decade or two and I am one of the participants. Some of my current work relates to climate but is not directly about climate change. I was not originally trained in climate science and came into the field from outside. I know many of the major players personally and have opinions of their skill and integrity. For the most part, I am favorably impressed.

      I know many scientists in a similar situation as myself. Almost all of us started out much more skeptical than we are today. As the years progressed, as new data came in, as models improved, almost all of us became more convinced that the risk is real and serious.

      I think the uncertainty has been downplayed for public consumption. But as others have pointed out, increasing the uncertainty increases the risk.

      • The consequences of space aliens annihilating the Earth are lethal and very bad, for the Earthlings anyway. But the uncertainty of this happening is huge. So in your opinion, since the uncertainty is huge, we should spend a huge quantity of resources to mitigate this disaster? I’m just trying to understand what you are suggesting.

  15. Judith,

    Here is an unknown senario:

    Is it extremely dangerous to be replacing water for oil by pumping it into wells?
    2 parts hydrogen and 1 part oxygen compressed.
    What happens if magma breaks open into that chamber?
    We have certainly set up many chambers like this in a need for oil.

    • Same as magma hitting any aquifer.

      The results depend on the quantity of magma, quantity of water, magma composition and depth. At deep depths the water will mix with the magma, at shallow depths it may be involved in explosive eruptions.

      May I suggest reading some basic textbooks in geology (and physics, and chemistry)?

    • Joe,

      I am not aware of water injection as a means of increasing oil production into anything other than a water drive reservoir (which means that water occurs naturally underneath the oil). Lots of things I do not know, but are you aware of an exception?

      • Water, which frequently occurs with oil as brine, is separated from the oil and re-injected into the formation. This can be supplemented with water from the sea or rivers in the latter stages of water flood production. CO2! is also injected into oil-bearing formations and used to dissolve/flush out oil. Steam is used and there is even a method called fire flood, where air is injected and the petroleum set afire. This melts petroleum solids and reduces the viscosity of the liquids.

  16. Steve Easterbrook and Andrew Dodds: Excellent points.

    The main issues are:
    – Uncertainty cuts both ways (but the tail is generally fatter at the high impact end, eg for climate sensitivity and impacts)
    – Uncertainty has been hidden
    – (the role of) uncertainty has been misunderstood by the public, in part due to the very different languages (science-speak vs normal speak)
    – Strong voices that try to (ab-)use uncertainty for political purposes (e.g. confuse it with ignorance; merchants of doubt; but also those who underplay uncertainty as if all details are certain)

    I’m reminded of your Feynman quote in your post “doubt”: some things are better known than others, and it’s the general picture of what we know fairly well that’s most policy relevant. Bickering about the details is A) scientifically interesting but B) politically a great strategy to postpone meaningful action.

    • Whoops, I meant to write:

      – uncertainty has *not* been hidden.

      (no, it wasn’t a Freudian slip)

      • Bart:

        The main issues are:
        – Uncertainty cuts both ways (but the tail is generally fatter at the high impact end, eg for climate sensitivity and impacts)

        how certain are you the tail is fatter?

        And to your point on no uncertainty being “hidden?”

        you sure about that? Knowing that would imply that you know about all uncertainty and that none is hidden. Is their uncertainty that is “unexposed”.. I’m thinking you take issue with the idea that somebody knows about an uncertainty and is hiding it. So, you trust that no one is doing this. You don’t know that no one is

    • Bart, the objective of this series on uncertainty is to defuse uncertainty as a political weapon in the policy wars. This requires a much better characterization of uncertainty than has been done by the IPCC, and more sophisticated ways of reasoning about uncertainty than the consensus approach described by Moss and Schneider. And most importantly, it requires robust decision making strategies that can accommodate deep uncertainty, combined with a range of policy options. The linear model does’t work: reduce the uncertainties, political support lines up with science, and the optimal policy derived from the science is implemented. We’ve already seen that it doesn’t work. Too many of the uncertainties are irreducible, so this approach even if it was politically palatable wouldn’t produce good policy. The UNFCCC emissions target/CO2 stabilization approach is a brittle non-robust strategy: it may turn out to be inadequate on the one hand, or it may turn out to be overkill with unnecessary expense on the other hand.

  17. I think there should be more efforts such as Dr. Spencer’s or Dr. Lindzen’s where an attempt is made to delineate a testable hypothesis concerning the computer models or predictions from other sources. Then test the hypothesis. This sort of effort can greatly influence the uncertainty.

  18. Andrew Dodds

    Newtons theory in physics is incorrect as it uses formulas and theories that motion is forever and that other factors slow objects down. This has spawed many a string theory and point to point theories that do NOT include rotation and solar movement.

    Actual evidence shows that planets and suns INFUSED with energy and the natural use of energy slows these down. Suns expand through time with the slow relaxing and release of energy. This changes energy on planets and allows evolution.

    We seem to forget that water in its natural form is a two gases combined. So through rotation and pressure build-up, water is pressurized gases with stored energy of trying to go back to being a gas.
    Life on this planet and the evaporation cycle are not in natural states but in a pressurized state.
    What makes water more interesting is it has picked up elements that we drink and our bodies need to function such as iron. We NEED the energy water carries to fuction.
    Electro-magnetics makes a more complex system as the planets energy through centrifugal force gives us more complex life. If not, the pressure and electro-magnetics would keep us pools of chemicals, so centrifiugal force is nedded as well.

    • Oh dear.

      Besides the fact that this is a load of rubbish, what has it to do with the subject raised by Judith?

  19. If any long term modeler, whether it be climate, sunspots, earthquakes, weather, etc. is truly honest about the uncertainty in their “prediction” will be greater than the difference between and the current norm – namely our predictive ability is nearly zero.

  20. Judith,

    I like this a lot, but I’m dubious since what you call coping strategies are a mixture of actual strategy and motivation. In particular, the description of the “merchants of doubt” strategy, even in your version, reads to me too much like a conspiracy theory and “we’re at war with the evil empire”, even though I know you don’t believe that’s a useful way of thinking about it. This is the only strategy that is described as being driven by “financial or ideological reasons”. I believe all of them could be driven by such reasons. It looks like an over-simplification of the complexities of motives and incentives.

  21. The adult response to uncertainty is to deal with it. That is why people buy insurance. The adult question is how much insurance, which is where the possible threats and the amount of uncertainty are weighed.

    In this discussion, if we are to be adult about it, as Steve Easterbrook says, the point is that the uncertainty is asymmetric. We KNOW that there is a minimal response of the system on the favorable side. That is strongly limited (~1K/doubling, but of course, that excludes the negative effects of ocean acidification). The limits on the unfavorable side are nowhere so strong.

    • Insurance is the best response to uncertainty that can be measured in an agreed on manner.
      AGW is not close to that.
      Another thing to consider is why the large number of climate scientists who are making profound policy demands about things like biological systems, food distribution, and other areas totally outside of climate science. Imagine how insurance actuaries would be considered if they started talking about what kinds of medicines or surgeries are appropriate for preventing situation X.

      • Imagine if medical doctors would start diagnosing somebody’s medical condition.

        Oh, right.

      • I’ve asked previously under a different topic about this – who are these climate scientists who are making ‘profound policy demands’?

        And what are the demands?

      • …”Another thing to consider is why the large number of climate scientists who are making profound policy demands about things like biological systems, food distribution, and other areas totally outside of climate science.”…

        These are NOT ‘Climate Scientists’ as you say, these are ‘Climate Political Scientists’. I believe you’ll find that there is no academic criteria to graduate as, or become by self-proclamation, the latter. They usually assume the title after conversion I understand. I believe the confusion is primarily the result of those with valid advanced degrees speaking out of their area of expertise and advocating political and economic solutions to perceived climate problems.

      • Imagine if x-ray technicians started prescribing medicine.
        Climate scientists are not biologists. They are not civil engineers. They are not agrarian experts.
        Yet ‘monster promoters’ regularly make wild claims about biological and agrarian and civil engineering implications of their work on climate.
        Climate scientists are not the world’s doctors.

      • Andrew,
        Good question.
        Have you read anything by Hansen, for starters?

    • There is no proof that the uncertainty is symmetric. That is purely an intuitive way of analysis. In fact, there are multiple factors for which the uncertainty has been assumed to point in the direction of alarm (cloud albedo, aerosols, water vapor, hurricane effects, ice shelf collapse, sea level rise, etc etc). If we go back and reexamine the uncertainties of all of these and take a more likely (vs. the extreme) value in each case, or sample from a random distribution, the chance of runaway anything goes DOWN not up. A specific example is in some of Lindzen’s recent work, as also in the assumption that any environmental change will be bad for ecosystems without even doing an analysis.

      • Dr. Loehle
        Your 2000 yr temperature reconstruction has an aspect to it, which may (possibly) reduce degree of uncertainty discussed.

      • Uncertainty is certainly asymmetric because, if nothing else outcomes are more strongly bounded with less perturbation of the system. That Craig, is not speculation.

        Runaway anything is a strawman is one of the usual handwavers but it is amusing how you dismiss it while down below others are pontificating about how climate is chaotic (it is not, weather is). Please get your group together and choose one or the other.

  22. Uncertainty messages get ‘squashed’ ie, the media response to climategate/treatment of sceptical scientists

    Richard Black BBC has just written this….
    (Romm has had a go about a BBC article on Artic Ice, not consensus enough?)

    BBC: ‘Warmist’ attack smacks of ‘sceptical’ intolerance
    “I am wondering, therefore, whether it does presage the start of something – whether it is now going to be routine for those of us who attempt to report on climate change objectively to be on the receiving end of barrages of critical mail, stimulated by bloggers with a definable agenda, whenever we write something that does not tally with their agenda.

    What about scientists? If researchers publish papers on climate change that do not include cataclysmic warnings of where the world is heading, will they receive the same treatment?”

    ie the media response to climategate/treatment of sceptical scientists

    I make no apoligies for my involvement in the comments there….
    (it is my BBC as well)

  23. Let me play devil’s advocated; not entirely tongue in cheek. There is no uncertainly about AGW; it is a hoax. There are three crucial numbers for a doubling of CO2; the change in radiative forcing – 3.7 Wm-2 -; the change in global temperatures without feedback – 1.2 C -; the multiplying effect of feedbacks – between 4 and 6 -.

    The first and third numbers are based on the output of non -validated computer models, and are completely meaningless. The second number makes a wrong inherent assumption that all the radiation that escapes into space comes from the surface of the earth.

    As I say, AGW is a hoax. You, Judith, have agreed that, at least, it is only a hypothesis.

    • It’s time to kill off this “unvalidated computer model” meme once and for all. It’s very easy to repeat when you actually know nothing about what these models are and how they are tested. Go and learn something:

      • Steve, the whole issue of complex system models, particularly of the nonlinear dynamical variety such as global climate models, presents some substantial epistemic challenges that have not been adequately explored. I agree that the conventional verification/validation approach used in closed system models for engineering applications doesn’t make sense, but some sort of V&V approach suitable for complex system models should be pondered.

      • Steve: I have examined lots of computer model output. They do pretty well in some cases and not so well in others. Some models freeze up the entire arctic (no open water at all) and they vary in the jet stream paths and many other outputs. Do they generally look like Earth, yeah, sort of. But maybe I believe them enough to base world econ policy and maybe I don’t. I don’t think your umbrage at someone criticizing the models is justified–they aren’t that good.

      • Dr. Loehle, the commenter wasn’t merely criticizing, he was asserting that scientists like Prof. Easterbrook are perpetrating a “hoax.”

        I’d be rather proud of myself if I were able to restrain my response to “umbrage” if someone made that accusation about me…

      • Thanks steve, nice link

      • Ok, let’s try to learn. Your concerns are (almost random pick):

        – that the model will compile and run without crashing;
        – also include the built-in tests for conservation of mass and energy;
        – that the model simulates a realistic, stable climate,;
        – matches the trends seen in observational data when subjected to historically accurate forcings ;
        – literature on the philosophical status of model validation in computational sciences (see for example, Oreskes et al ;
        – reduce 3D equations of motion to 2D horizontal flow;
        – This allows for testing of numerical convergence ;
        – Statistical tests are then applied to check for realistic mean climate and variability;
        – Multi-model ensembles, Perturbed physics ensembles;
        – Varied initial conditions within a single model…

        Is this a kind of a joke? Do you have any foggiest idea what kind of topological complexity you are dealing with when you are talking about “perturbed physics” and “perturbed initial conditions”? How about “perturbed boundary conditions” other than your model does not blow up? Aside that half of your validation criteria are trivial and the other half is boldly wrong, you don’t have any adequate and reliable data to compare with. Do you have any real criteria? Something even as simple as “rain happens only when clouds are there”, or “weather formation cannot fade away on area of 100,000 km2 in 15 minutes”, as one can observe in say, “”?
        No, this must be a joke…

      • You’re bullying.

      • See, now you’ve gone and done it.

        Between reading what snippets I can of, for example, and like books, and the link you’ve provided, I now am pretty much obligated to shut up here except to ask questions pertinent to my learning, and devote myself to critical enquiring by actual study.

        You’re terrible people, making me read hard and think for myself with rigor and discipline in my free time.

      • Bart R,

        Please reconsider. You’re one of the reason I’m still reading.

    • Steve, I suspect you and I are using the word “validate” to mean two entirely different things. By validate, I mean the model must quantitatively, and within stated errors, successfully predict the future on many occasions. Once is not enough; it could be coincidence. None of the models used in climate science have ever been subjected to this sort of rigorous test. You cannot validate radiative transfer models to estimate the change in radiative forcing, since you can never actually measure radiative forcing. Feedbacks are purely theoretical, and can never be actually measured. So it is impossible to validate these models.

      Other models have other problems. I remember Gavin Schmidt claiming that the “sweet spot”, for climate models was 30 years in the future. On my definition, one could not validate such models for at least 30 years.

      • Now that’s very convenient: we should wait 30 years until the models could even be validated. Any chance that that coincides with your political desire of contuing business as usual?

        Don’t use science as a dress if you’re really talking politics, please.

      • Jim, or perhaps you and I are using the word “model” to mean two entirely different things. The models I’m familiar with (GCMs, and ESMs) aren’t built to predict the future, they are built to understand the processes that drive climate change. The validation they are subjected to is incredibly thorough and appropriate to the task.

        Climate modellers are frequently asked to produce long term predictions of the future (e.g. for the IPCC assessments), but are generally reluctant to do so, as they know this isn’t what the models are for. There was extensive discussion of this at the last AGU meeting, where, contrary to Judith’s claim, uncertainty in all its forms was extensively discussed and analyzed:

      • Steve Easterbrooke writes “Jim, or perhaps you and I are using the word “model” to mean two entirely different things”

        I specifically referred to the models which estimate radiative forcing, and the feedbacks. So, radiative transfer models are used to predict the change in radiative forcing for a doubling of CO2. And the GCMs are used to predict how much more than 1.2 C global temperatures will rise due to feedbacks. Neither of these estimates can be measured, so I fail to see how they can ever be validated. How do you validate a model, if the output can never be measured?

      • We have given the ‘monster promoters’ of AGW > 20 years to show us the apocalypse.
        So skeptics asking for a few years to finally settle if there is anything to it at all seems reasonable.

      • How do you validate a model, if the output can never be measured?

        I dunno, see if “it matches the trends seen in observational data when subjected to historically accurate forcings?” Maybe? You think?

      • pda writes “How do you validate a model, if the output can never be measured?

        I dunno, see if “it matches the trends seen in observational data when subjected to historically accurate forcings?” Maybe? You think?”

        And how does this work if you are using radiative transfer models to estimate the change in radiative forcing for a doubling of CO2? Or GCMs to estimate how much the global temperature will rise more than the 1.2 C estimated for a doubling of CO2 without feedback, when the feedbacks actually occur?

      • And how does this work if you are using radiative transfer models to estimate the change in radiative forcing for a doubling of CO2?

        Um, the same way?

        Or GCMs to estimate how much the global temperature will rise more than the 1.2 C estimated for a doubling of CO2 without feedback, when the feedbacks actually occur?

        The phrasing here is difficult to parse. Are you saying it’s impossible to model the effect of CO2 forcing isolated from other forcings? Or that feedbacks can’t be modeled?

        There are folks here that could answer those questions for you, but if you think it’s all a “hoax” I guess it doesn’t really matter.

      • PDA, thanks for those.

    • > You, Judith, have agreed that, at least, it is only a hypothesis.

      It’s not certain. So it’s uncertain. So it contains uncertainties.

      Uncertainty. Some dots. Hoax.

      What kind of knowledge is only based on certainties, again?

  24. Steve Fitzpatrick


    It is constructive to examine scientific uncertainty and the different components that go into it. But I think it is important to not lose sight of the impact of what I would call the public’s perceived ‘total uncertainty’ on the extent of contentious political debate related to global warming. ‘Total uncertainty’ is a combination of scientific uncertainty, and ‘political uncertainty’. I suggest that political uncertainty is at least as important as scientific uncertainty.

    I think it is fair to say that an overwhelming majority of people would be willing to make substantial personal sacrifices to avoid serious calamity due to global warming. If calamity resides in the high warming tail of the probability distribution, and minor impacts reside in the low warming tail, then people will honestly disagree about what represents suitable public policy; but in the end a consensus will be reached. This kind of normal political process has been side-tacked by the low credibility many people give to the stated scientific consensus; this is the ‘political uncertainty’ component.

    Many who have read the private email communications between several well known climate scientists (most of them contributors to the IPCC process) have concluded, rightly or wrongly, that these scientists have obvious and strong personal/political views related to global warming, and that these scientists believe pretty uniformly there is need for immediate and drastic public action. These scientists acted behind-the-scene in ways that many people see as political advocacy influencing ‘the science’ as well as the IPCC process. They are perceived by many as having had their thumbs on the climate science scale all along. The apparent tendency of the IPCC to overstate certainty seems to me symptomatic of the influence of strongly held policy views.

    The acceptance by the public of stated scientific projections and associated uncertainties, as well as a reasoned public debate about appropriate public action, both depend on convincing the public that advocacy does not play a role in the process of climate science. In light of the history of climate science over the past 15 years, reducing political uncertainty appears to me to be a very difficult task.

    • Steve, there’s an important political/policy debate that needs to be had on this issue, than involves values, economics, environmental justice, etc. Instead, the science has become a proxy for what should be political/policy debates. Not enough attention has been given to formulating robust policy strategies that are palatable across the range of cultures, locales, etc. The top down UN style approach won’t work.

    • “I think it is fair to say that an overwhelming majority of people would be willing to make substantial sacrifices to avoid serious calamity due to global warming.”

      No doubt the above view is correct. However look at some of the “high carbon lifestyle” exceptions to that observation: Al Gore, Senator Kerry, Prince Charles, Tom Friedman, etc.

      The exceptions are, curiously, the public leadership of the CAGW advocates. Yet, despite their leadership positions, these individuals offer no more than token gestures when it comes to moderating their own extravagant, high-carbon lifestyle. In the search to reduce our uncertainty, one effective technique, I would recommend, compares words with deeds. Using that technique, we can immediately see that the well-informed “big boys” really think the whole CAGW business is a crock on its merits (but probably a good peg for their hustles, no doubt).

      • > In the search to reduce our uncertainty, one effective technique, I would recommend, compares words with deeds.

        Not doubt this technique is effective. Yet, it still is a fallacy.

  25. Judith,

    Your first of many (I hope) uncertainty posts is thought provoking.

    My initial comment is whether there is a false dichotomy presented by the categorizing of the two fundamental uncertainties as either epistemic or ontic (aleatory). I think there may be a problem with the ontic uncertainty concept.

    If by ontic uncertainty it is held that at the present state of statistics/technology/etc they are irreducible, then this category of uncertainty is a pointer to more needed development of science capability to solve the problem of dealing with it. It is just a problem to be solved. Then ontic is unrealized epistemic due to the state of man’s knowledge at this time.

    If by ontic uncertainty it is held that it is an impenetrable barrier that, epistemologically & metaphysically, man’s mind cannot go beyond, then there is a fundamental philosophic problem of how anyone can know enough to claim man cannot know it. This is the well worn-out Kantian/Platonic type “certain things cannot be known” stuff. It is not on the path to that led to western science as we know it today. It is on the path that leads to mysticism.

    So, the ontic categorization as it stands in your post appears problematical.

    If it is not essential to your argument, it will be a philosophical legacy that you may not wish to carry forward.


    • Tomas Milanovic

      If by ontic uncertainty it is held that it is an impenetrable barrier that, epistemologically & metaphysically, man’s mind cannot go beyond, then there is a fundamental philosophic problem of how anyone can know enough to claim man cannot know it.

      It may be a fundamental problem in philosophy but it has been solved long time ago in mathematics and physics.
      If I demonstrate that orbits of a chaotic system diverge exponentially, I will establish an absolute barrier for man’s mind to know the future evolution of the system.
      This demonstration exists since 50 years and it is indeed enough to claim that man will never be able to predict (e.g “know”) chaotic orbits.
      One could also mention Gödel’s theorems.

      Similarily the knowledge of quantum mechanics is enough to demonstrate the Heisenberg uncertainty relations that are also an absolute barrier to knowledge/measure. As this example comes from physics , on top of mathematical and logical consistency, there is experimental evidence that these relations are true.

      • Tomas Milanovic ,

        Thanks for your dialog on this matter.

        you said “. . [edit] . . . Heisenberg uncertainty relations that are also an absolute barrier to knowledge/measure . . . [edit] . . .” I say except that it does not categorically preclude man from ever knowing, it implies with the current state of knowledge and tools of observation (technology) that we should not be able to know. It does not allow you to say that area of knowledge is dead-ended.

        You said “If I demonstrate that orbits of a chaotic system diverge exponentially, I will establish an absolute barrier for man’s mind to know the future evolution of the system.” I say except that you are basing your statement on an assessment of man’s current capabilities. That does not logically allow you to claim all future knowledge developments are impossible.


      • Tomas Milanovic

        I say except that you are basing your statement on an assessment of man’s current capabilities. That does not logically allow you to claim all future knowledge developments are impossible.

        This is not how mathematics work.
        A mathematical result does not depend on subjective “current” of “future” capabilities.
        A theorem once demonstrated stays demonstrated forever unless you made a mistake.
        As the verification of validity of a theorem must be (per definition) a time finite process, at the end of the process the theorem is then validated forever.
        The content of the theorem is not important and there is no qualitative difference between a theorem demonstrating an impossibility of something or an existence of something.

        In my example once you have the unpredictability theorem for chaotic orbits (and you have it), then the theorem stays true forever independently of any future additional knowledge.
        This is the consequence of the consistence of mathematics.
        Which is itself also object of Gödel’s theorems.

  26. Going back to David Spiegelhalter’s wise words, it’s unfortunate that his first instruction is: “We should try and quantify uncertainty where possible”. This serves to reinforce the common illusion that all uncertainty can and should be quantified. That is responsible for much of the confusion surrounding its management. Silvio Funtowicz and I devised the NUSAP system for bridging the gap between the uncertainties that can easily be quantified, and those for which quantification is impossible.

    • Jerome, thanks for stopping by. Speigelhalter qualifies his statement with “where possible.” Your excellent and very useful NUSAP scheme (and other similar schemes) will be the subject of a future post.

  27. An excellent article. I have spent most of my career in risk management. The application of financial approaches to the management of water for the production of electricity, to both maximise profits and ensure supply. In this we have grappled with what we know, we don’t know and the very natural variability (rainfall, inflows, floods, temperature and ultimately demand for electricity) that is discussed in this article.

    Steve Easterbrook says, ” So the question that you never get to is: “what’s an appropriate course of action, given the seriousness of the threat of climate change, and the inherent uncertainties in understanding it?” “.

    This is an interesting statement. Seriousness implies many aspects. Serious in the magnitude, serious in frequency, serious in teh consequence etc. More importantly there is the threat of climate change and inherent uncertainties in understanding it. Serious suggests that some analysis has been made and it is agreed that the combination of risk (or probability) and consequence is so great that immediate action is required. But what actions?

    If there are inherent uncertainties, then the scope and scale of the threat is likely to itself uncertain. Earthquakes are an interesting example (as is Judith’s previous article on the floods in Pakistan). Earthquakes are inherently uncertain, we have some understanding clearly about the mechanism by which they occur and the potential range of consequences. We are currently unable to determine when they will occur or how big they will be. This does not stop us from dealing with the risks in a sensible (qualitative statement) way. We have building codes to limit physical damage, we have insurance (hopefully), we have remediation or retro fitting to improve older buidlings etc. As much as some people might wish, we can’t actually get people not to live on or near fault lines (California, Japan, New Zealand etc) or on flood plains.

    As with the above we should be asking what a sensible set of actions is to reduce the tail events, the more frequent events etc with respect to the “potential” for climate change. I have deliberately used the word potential, because we are dealing with an uncertain future and we do not have all the information. This approach implies that the in developing the strategy to deal with the risk will take account both its probability and consequence, as best we can assess them.

    The range of events we may expect we need to deal with, probably falls within the distribution of events we would (eventually) have to deal with anyway. Again the flood article is good in this regard. The flood may not have been exceptional in the quantity of water that fell (it may have been on the edge of the historic distribution, but the distribution itself may be limited by the time collection period of the data), but the aggregate effect of other changes, land use, deforestation, increasing population density amy have meant that the consequence was greater.

    This would argue for many of the actions indicated in the flood article, as a start. As proposed in that article, the changes would be beneficial irrespective of the actual cause of the rain.

    The presentation of potential effects of climate change has been too driven by emotion, politics and agendas to date, to allow a sensible analysis, in a manner that allows a significant number of people to understand the costs and consequences (and benefits).

    Just as it is unrealisic to expect a government to “fix” the economy (whatever that means) it is unrealistic to expect them to “fix” the environment; what is the target? The environment with no people, as it was in 1900 etc? What is it we are atually trying fix or prevent, the cause, the effect, the consequence? Are we trying to remedy, mitigate or avoid?

  28. Tomas Milanovic

    Natural internal variability of the nonlinear climate system contributes to ontic uncertainty in climate simulations. The climate system is stochastically uncertain because of its chaotic nature, i.e. small differences in the initial conditions of a global climate model can yield very different results.

    I fully agree with this statement and consider that this point is one of the largest if not the largest issue in the discussion about uncertainty.
    There are several notions that are so taken for granted that people don’t even discuss them. I will list several . All of them are intimately related to your statement .

    1) If I run a climate model 100 times and find a state A of the system 20 times, then the probability that the real system will be in the state A is 20 %.
    What is the justification of this belief?

    2) If I run another model on another computer with another numerical scheme and find the state A in 5 runs out of 100 then the probability that the real system will be in the state A is 12.5 % (20 + 5 /2).
    What is the justification of this belief?

    3) Weather is chaotic, nobody disputes that.
    The “climate” is exactly the same system, obeying to the same laws and described by the same equations like weather.
    The only difference being that the variables of the system “climate” are space and time averages instead of the instantaneous values.
    In adition for practical purposes the weather time scale is defined in days so that many slow variables are considered constant what spares computing time.
    However it is clear that if the system is chaotic with these constant coefficients , it will be chaotic with variable coefficients on longer time scales too.
    This last statement is equivalent to what R.Pielke Senior wrote some years ago when he stated the the “climate” is NOT simpler than weather but on the contrary more complex.
    Yet is believed that space and/or time averages of a chaotic system are no more chaotic and can be deterministically predicted.
    What is the justification of this belief?

    Climate models don’t solve equations governing the dynamics of the system . They can’t do that and will never be able to do that.
    What they do is that they simulate the system by constraining it to obey the conservation laws. Because of the very low resolution this can’t work exactly so there are adjustements.
    However to obtain a state B of the system when starting from the state A, obeying the conservation laws is a necessary condition but not a sufficient one.
    The dynamically allowed states of a chaotic system are a very small subset of the states that are allowed by conservation laws- this is why attractors exist.
    It is believed that ANY computed state of the system must be a dynamically allowed state so that it can be a REAL state of the system (with unknown probability).
    What is the justification of this belief?

    Space averaging destroys the spatial correlations. Yet the climate science uses often space averaged variables, sometimes averaged even over the whole Earth.
    It is believed that there is still some meaningful signal in such space averaged variables despite strong and obvious spatial correlations (air&ocean is much more about waves propagating in space than about statical equilibriums and standing waves).
    What is the justification of this belief?

    There is more along the same line but even only these 5 points which impact seriously the uncertainty evaluation are rarely discussed in depth, and certainly not in IPCC reports.

    • Tomas, thank you for this clear statement of the issue, it is definitely the elephant in the room.

    • Tomas: “…even only these 5 points which impact seriously the uncertainty evaluation are rarely discussed in depth, and certainly not in IPCC reports.”

      Unless you actually read them, where you will find such questions discussed in depth. Try chapter 8 of WG1:

      • Just searched chapter 8 and there is no mention at all of whether the models are chaotic, let alone ‘in depth’. So Tomas was right.

      • Tomas Milanovic

        First I have read IPCC and take exception for you implying that I didn’t.
        If I said that these points are not discussed in the IPCC reports then it means that I didn’t found them there. Neither in chapter 8 nor elsewhere.
        Like PaulM writes, your link answers nothing.
        What comes closest is:
        Models, both AOGCMs and less complex models, have produced examples of large abrupt climate change (e.g., Hall and Stouffer 2001; Goosse et al., 2002) without any changes in forcing. Typically, these events are associated with changes in the ocean circulation, mainly in the North Atlantic. An abrupt event can last for several years to a few centuries. They bear some similarities with the conditions observed during a relatively cold period in the recent past in the Arctic (Goosse et al., 2003)

        Unfortunately, the probability of such an event is difficult to estimate as it requires a very long experiment and is certainly dependent on the mean state simulated by the model. Furthermore, comparison with observations is nearly impossible since it would require a very long period with constant forcing which does not exist in nature. Nevertheless, if an event such as the one of those mentioned above were to occur in the future, it would make the detection and attribution of climate changes very difficult.

        This is hardly an in depth discussion.

    • fizzy water solution

      The issue of predictability and chaos in climate systems has been the subject of many studies within the past two decades. The assertions of Tomas Milanovic regarding this issue reflect an unfamiliarity with the extensive literature on the subject. Examine some of the many papers found

      • Tomas Milanovic

        Thank you Fizzy to have shown that a Google search can find papers related to specified key words even if it doesn’t answer any questions.
        Obviously I am well aware that there has been much work done on chaotic systems since Poincare 100 years ago.
        However neither I nor anybody else have read all published papers and not all of them are interesting or important.
        Giving just a Google link to thousands of papers , many of them having less than 5 citations and some of them contradicting others is not very helpful for this discussion.

    • 1 and 2 – Without context I have no idea what you are on about here.

      If I roll a die and it comes up 6 1 time out of six.. then I can predict that 1 in 6 die throws will be a six. I really don’t understand why that needs to be justified.

      3 – Climate is not weather.

      Again, to illustrate, I can predict that the average of a set of die rolls will be 3.5; and do it trivially. Predicting what an individual die roll will show is not possible. Clearly, the average of a set of trials behaved differently from the individual trials.

      4 depends on 3; climate is not chaotic, you have to demonstrate this first, not assume it. That means real evidence from the real world.

      5 – You’ll have to clarify, this point seems incoherent to me.

      • Between each roll of your die is a reset. Your second roll is not influenced by the outcome of your first roll. Your third roll is not influenced by your second, which was not influenced by the first.

        If you drop a ball from 5ft, six times, and each time it bounces and rolls 0~2ft, you can predict with reasonable confidence that on the sixth time, the ball will come to rest within 2ft of your start position.

        But what is your confidence in the ~2ft prediction after the sixth drop if each time you drop the ball from 5ft above its adjusted position (above the position it rolled to)?

        Now contemplate uncertainties if the ball is not perfectly round, and the surface you’re dropping it onto is not perfectly flat.

      • Urgh.. “on to”, not “onto”. I do that alot. Err.. a lot. :(

      • Tomas Milanovic

        1 and 2 – Without context I have no idea what you are on about here.

        If I roll a die and it comes up 6 1 time out of six.. then I can predict that 1 in 6 die throws will be a six. I really don’t understand why that needs to be justified.

        I am not surprised that you have no idea and it actually proves one of my points.
        You don’t even realize that in your die example you make this statement because you implicitely suppose ergodicity even if you may ignore what ergodicity means.
        You are lucky with the example because the die actually is ergodic. You could have also used a roulette example which is ergodic. Not everything is.
        In particular the weather is not and nobody knows what the climate is.
        I mentionned 2) precisely because this would be a proof that either the system is not ergodic or that both models are wrong.
        There is no reason that an invariant probability distribution if it exists would be the arithmetical average of 2 wrong guesses.

        3 – Climate is not weather.

        Again, to illustrate, I can predict that the average of a set of die rolls will be 3.5; and do it trivially. Predicting what an individual die roll will show is not possible. Clearly, the average of a set of trials behaved differently from the individual trials.

        Come on . I am trying to have a serious discussion here. See the comment above concerning 1) and 2).

        4 depends on 3; climate is not chaotic, you have to demonstrate this first, not assume it. That means real evidence from the real world.

        Indeed. As your 3 fails, your 4 fails too. The situation however between you and me is not symmetrical. We do know that the weather in the real world IS chaotic. Now you prove that the averages are not.
        The proof that the averages of chaotic trajectories are chaotic exists already btw even if only for temporal chaos and not spatio temporal chaos. It is actually often a simple exercice asked for Lorenzian systems.

        5 – You’ll have to clarify, this point seems incoherent to me.
        Well then you must specify what seems incoherent to you because it seems quite clear to me.
        What part of spatial averaging of a spatio-temporal solution of continuous equations is unclear to you?

      • Tomas –

        I would note that you haven’t actually given examples, or references. What is ‘state A’ in the context of a climate model? What is ‘State B’? This being the real world, abstract models do not satisfy.

        You contend that the climate is chaotic. Prove it. There is a lot of data out there to work with nowadays.

    • Climate, and climate models are not chaotic. Weather is. Your statement reduces to a claim that the average temperature in July in New York can be the same as the average temperature in January. Wanna bet? The rest is simple persiflage.

    • Tomas writes:
      3) Weather is chaotic, nobody disputes that.
      The “climate” is exactly the same system, obeying to the same laws and described by the same equations like weather.
      Same laws, equations that describe chaos?
      As for the statement that nobody disputes that weather is chaotic, try reading Marcel Leroux “Dynamic Analysis of Weather and Climate” Springer 2010 2ed.

      • Averages are much more constrained than single realizations which is why, although weather and climate obey the same physics, weather at a particular point and time is much less constrained than climate over the years and a larger area.

        To claim that climate is chaotic is to ignore the fact that it does not snow in Chicago in July. If you want a simple simile, climate is the distribution that weather is drawn from.

      • Tomas Milanovic

        This comment tells more about misconceptions than about the issue discussed.
        Chaos doesn’t mean that “anything goes”.
        It doesn’t mean that “chaos=randomness” either.
        It is not because these misconceptions are very common that they are less wrong.

        Averages are neither less nor more constrained in temporal chaos.
        It can be done as exercice to prove that if an orbit is chaotic, then its time averages over any arbitrary averaging period are chaotic too. (Hint: Consider that the chaotic orbit has at least 1 Lyapounov coefficient >0 and prove that the time average of this orbit has also at least 1 Lyapounov coefficient >0).

        Time averages of temporal chaotic orbits are chaotic, this is just a fact that doesn’t stand to discussion.

        Of course it would be ridiculous to say that it will snow in Chicago in many Julys just because the system is chaotic.
        If the temperature and precipitation variables in Chicago followed chaotic orbits, they would be constrained to an attractor.
        Everything outside the attractor are forbidden states that the system would never visit.
        Each state along the attractor would be visited with a certain frequency because chaos is not “anything goes”.
        So clearly the warm snowless states in July in Chicago would be visited very often.
        However the chaotic nature of the orbit would manifest itself by the fact that the July states would be different every year (the no intersection theorem) and that one could not exclude snow in July even if this state would be preferentially chosen by the system in December.

        The reality is however much more complex.
        The climate (or weather) in Chicago is not and cannot be temporal chaos. It is spatio-temporal chaos.
        Spatio-temporal chaos is to temporal chaos what QFT is to a free particle dynamics.
        To spare a lengthy development about what spatio-temporal chaos is and is not, here a very informative link :

        Be very sure that if the problem of spatio temporal chaos even in a very “simple” case like Navier Stokes could be trivially and deterministically solved by making averages, then T.Tao, who understands this question infinitely better than all of us, would have solved it long ago.
        And it is literally a million$ question.

      • Rabett: it does not snow in Chicago in July.
        Tomas: would be ridiculous to say that it will snow in Chicago in many Julys

        A poorly chosen example. 90% of the past 500,000 years it has snowed on Chicago year round. Perhaps not during the current interglacial, but this will change.

      • Tomas Milanovic

        What makes you think that I don’t know Leroux thesis?
        If I didn’t mention it, it is because I am not at all convinced.

      • Omniscience is just a past time I presume?…

      • Your comment on chaos was very interesting. Could you briefly expand on your take on Leroux work? thanks

  29. Judith,

    Again, I thank you for taking the time to lead this open discussion of uncertainties in the experimental evidence for CO2-induced global warming. That is something we can all agree on.

    The Climategate scandal and the subsequent “investigations” exposed a more basic problem that threatens the most basic goals of free societies.

    Former US President Dwight D. Eisenhower warned us about this threat in his farewell address to the nation on 17 January 1961:

    “Yet, in holding scientific research and discovery in respect, as we should, we must also be alert to the equal and opposite danger that public policy could itself become the captive of a scientific-technological elite.”

    “It is the task of statesmanship to mold, to balance, and to integrate these and other forces, new and old, within the principles of our democratic system – ever aiming toward the supreme goals of our free society.”

    It is the lockstep mentality of this “scientific-technological elite” – rather than the uncertainty monster – that seems to explain why high quality empirical evidence has been, and still continues to be, ignored of:

    a.) The Sun’s natural variability.
    b.) The Sun’s influence on Earth’s changing climate.
    c.) The Sun’s origin, composition, and source of energy.

    The UN’s IPCC, Al Gore, and the army of climatologists that shared a Nobel Prize will have a hard time convincing us that they are serious about understanding the causes of climate change if they continue to ignore Earth’s heat source – the Sun [].

    With kind regards,
    Oliver K. Manuel
    Former NASA Principal
    Investigator for Apollo

  30. Judith,
    I’m trying to reconcile your claim that “A sustained and systematic enquiry of how to understand and reason about uncertainty in climate science has not been undertaken by either climate researchers or philosophers.” and the very existence of the Royal Society workshop you mention. Tim Palmer and co wouldn’t have put together the workshop unless they were already engaged a sustained enquiry into uncertainty in climate science. Certainly you’ll have seen Julia Slingo’s talk at the RS workshop where she explained how techniques used in NWP have been adapted for dealing with uncertainty in climate simulations (and I’m sure she would have mentioned the UK Met Office’s long-standing research program into these uncertainties).
    (For those who missed it, she gave it again at the CCSM workshop this summer: )

    And I’m sure you’re aware that many of the other participants of the RS workshop have also been grappling with these issues for years, as they discussed in the the session at the last AGU meeting on model validation:

    Your suggestions that you’re treading new ground here are more than a little overblown.

    • Steve, the IPCC lists uncertainty locations, and then summarizes all this with statements based on expert judgment (e.g. very likely). The talk given by Julia Slingo scratches the surface of how we can characterize some elements of climate model uncertainty. There are much broader issues, including model structure uncertainty that gets short shrift.

  31. Judith,

    Very interesting. You may want to also research finance literature where there is a rich body of work on this point – or more specifically on risk versus uncertainty. Those of us who have worked in financial modeling have been facing the issue for years (and not always successfully!) with the first discussions in the literature dating from the 1920s. In colloquial terms, financial professionals define risk as something you can hedge / insure while uncertainty cannot be protected. Eli and Hunter in their comments unfortunately miss the distinction here.

    We saw with the Wall Street meltdown that some of the financial models built by first rate mathematical minds simply broke down (the bulge bracket banks and the quant hedge funds are teeming with high level PhDs and former university professors – google James Simon as an example). It is not difficult to see potential parallels with climate models. I remember speaking to the risk manager for a very large Wall Street bank responsible for synthetic CDOs and asking him what went wrong with his models. He replied that there were no intrinsic errors to the models – but that the correlations used proved, well, less than robust, in a stress scenario. We use volatility as a proxy for risk, but volatility is itself a stochastic process. As is the volatility of volatility and so on. This all well captured in Taleb’s best selling “The Black Swan” where he expounds on why the standard tool for financial risk management modeling (Value-at-Risk) is worthless. I don’t quite agree with him here – VaR is not perfect but it is still the best we have, though I think we all agree it does not account for uncertainty. Blind reliance on financial models without understanding uncertainty helped create a financial disaster. Your posting suggests that the IPCC may be similarly blinkered.

    • Gens, in the financial sector, it seems like uncertainty got replaced by Knightian risk, without thought to the broader issues of uncertainty and black swans. There is some analogy here with the IPCC, although the IPCC relies more on expert judgment whereas the financial sector relies on statistical analysis of the Knightian risk. The end result is the same tho, insufficient consideration of the broader uncertainties.

  32. I am sorry but as I read how the major players falsified, adjusted, withheld and lost data I see no uncertainty.

    • Yet notice how skeptics are called Merchants of Fear.

    • You are right, NC, there is no uncertainty of the plan to sell one particular point of view as “scientific” – case closed!

      That is exactly what former US President Eisenhower warned us about in his farewell address to the nation on 17 January 1961:

      “. . . , we must also be alert to the equal and opposite danger that public policy could itself become the captive of a scientific-technological elite.”

  33. Bart,
    There would rarely be a good reason for a need to bail out of a serviceable aircraft, with a crude parachute. On the other hand, accurate altimeters are often found to be very useful.

  34. Yeah, as Christopher Booker said:

    ““…Academics who dare to question the scientific establishment’s consensus on Darwinism or global warming increasingly find themselves ostracised and demonised.”

    Or ridiculed.

    As Monbiot wrote:
    “To dismiss an entire canon of science on the basis of either no evidence or evidence that has already been debunked is to evince an astonishing level of self-belief.”

    And then complaining about being demonized and left out of the debate and such. Playing the underdog; the media will love you for it.

  35. Well, we always have to keep in mind Barry Marshall, Nobel Laureate in Medicine, who bucked the entire medical establishment which held which held for decades that ulcers were caused by stress, spicy foods, and too much acid (instead it is caused by bacteria.)

    • And he did so by doing research, publishing in journals and demonstrating the evidence.

      Though it should be noted that not all ulcers are caused by H. pylori. There are still a portion of ulcers that are of unknown cause.

  36. And some physicists are still fighting quantum mechanics to this day.

  37. Judith

    I read AR4 cover to cover and I found it hard to see how the level of confidence set out in the summary for policy makers reflected the contents of the report. I was particularly concerned by the level of scientific understanding (LOSU) detailed in table 2.11 of chapter 2 in relation to forcing agents. As a non-scientist, it is my understanding that forcing agents are the things that drive the climate (and presumably also the weather). You touch on this very point in one of your postings above, I think.

    There has also been much discussion in these comments about climate models.

    Is it not the case that climate models would of necessity need to reproduce the effects of ‘forcing agents’? If, as appears to be the case, the LOSU is low or very low on most of the forcings, how much confidence can we have in the outputs of climate models (quite apart from the points made by other commenters about the inherent problems of modelling a chaotic system).

    The other thing that causes me concern is that the output of climate models is very often presented to the public in the MSM as if it was scientific evidence.

    It’s things like this that lead to a sceptical view of the world – and climate science in particular.

    Surely we can do better?

    Kind regards


    • gary, your analysis is dead on. in the chapters there is some good analysis, however one chapter doesn’t know what the other is doing (inconsistencies) and the summary for policy makers, esp the confidence levels, seems inconsistent with much of the analysis in the chapters

      • Thanks, Gary and Judith, for excellent comments!

        Internal inconsistencies, hidden beneath slick PR (public relations), is another hallmark of bureaucratic-led science.

        That is why 99% of the public were once convinced that:

        a.) CO2 caused global warming, and
        b.) Neutrino oscillations solved the solar neutrino puzzle.

        If climatology and particle physics continue to receive public support after the recent PR fiascos, it will be because a few brave souls – like Judith at Georgia Tech and Janet Conrad at MIT – had the courage to slip truthful statements into the slick brochures designed to deceive the public.

        E.g., this statement of actual fact on page 15 of a beautiful brochure designed to convey the opposite message:

        “But do neutrinos oscillate? No experiment yet has proved that they do—at least not to the satisfaction of the very demanding scientific establishment.”

        Judith Curry and Janet Conrad would perhaps lose their research funds if they bluntly stated that “CO2 did not cause global warming” and “Neutrino oscillations did not solve the solar neutrino puzzle”, but most bureaucrats are unwilling or unable to read the fine print.

        Again, Judith, thanks for having the courage and the integrity to openly discuss the experimental evidence.

        With kind regards,
        Oliver K. Manuel
        Former NASA Principal
        Investigator for Apollo

  38. > the entire medical establishment which held which held for decades that
    > ulcers were caused by stress, spicy foods, and too much acid

    “Perhaps the greatest scam ever perpetrated by the pharmaceutical industry ….”

    Not the “entire medical establishment” — that confuses medicine with marketing. Where there’s money at stake, science gets pressured.

    • And climate science has stumbled into the mother lode of grant money.

      • The mother lode of grant money was discovered:

        First by a.) Nuclear Scientists;
        Then by b.) Space Scientists; and
        Recently by c.) Climate Scientists.

        The adjective changed: The flawed nature of “science” directed by Washington DC bureaucrats remained the same.

        That is why the:

        a.) Largest source of nuclear energy is still ignored;
        b.) Obsolete model of a H-filled Sun is still taught; and
        c.) Dogma of CO2-induced global warming is preached.

        Thanks to the continuing stench of the Climategate scandal, the public is becoming aware of the “scientific-technological elite” that Eisenhower warned us about on 17 January 1961:

        I expect that science will ultimately benefit as five decades of constipated science is now flushed from the system.

        With kind regards,
        Oliver the Optimist

  39. In fact, that’s worth a quote:

    “… the U.S. National Institutes of Health (NIH) in 1994 issued an advisory to all American physicians to discontinue the prescription of cimetidine, omeprazole, and ranitidine for their ulcer patients. Instead, the physicians were advised to treat them with the appropriate regimen of antibiotics, antibacterials, and bismuth compounds to eradicate the H. pylori and permanently cure the ulcers(2).

    It would appear that both the medical community and the pharmaceutical industry largely ignored the advice by the NIH to the painful and costly detriment of their patients.

    The pharmaceutical industry reacted vigorously to the threat posed by the NIH advice to their $8 billion a year antacid market by making a swift deal with the U.S. Federal Drug Administration (FDA) to allow their antiulcer drugs to be sold without a prescription. With this master stroke the drug companies freed medical doctors from the ethical burden of prescribing ineffective, costly drugs while at the same time maintaining or perhaps even expanding their market ….”

    Follow the money, follow the market — with suspicion.

    • Hank, is this very different from the climate experience, where huge energy comapnies are funding climate research and supporting cap-n-trade schemes?

      • > huge energy comapnies are funding climate research and supporting cap-n-trade schemes

        This message was posted from a parallel universe.

      • Dave,
        You are entitled to your opinion, but not your own facts.
        BP is a large supporter of cap-n-trade.
        Cap-n-trade was invented and promoted by Enron.
        Shell and Exxon Mobil fund huge dollars into AGW research and alternative energy research, as well as to NGO enviro groups.
        It is not parallel, but perhaps you are less familiar with the one we all live in than you think.

      • Hunter

        > You are entitled to your opinion, but not your own facts.

        Am I allowed to savour the irony of following that up with:

        > Cap-n-trade was invented and promoted by Enron.


        You might want to double check which source you just gullibly repeated that from, because it’s blatantly untrue. Just look into the history of the clean-air act in the US.

    • Dave H,
      Who is in the parallel universe?
      and more on Shell’s former chairman, Ronald Oxburgh:,_Baron_Oxburgh
      And BP directly funds Greenpeace:
      And for a nice summary of big oil supporting enviro groups and specifically funding AGW research:

      Could it be that AGW believers do not know what they do not know?

      • Hunter,

        It is obvious that energy companies spend money on greenwashing, and on investigation of alternatives. They are not stupid, and can see the inevitable collapse of their current business in a few decades. This is orders of magnitude less than what they spend on thinktanks and lobbying to preserve the status quo as long as possible.

        I took issue with you drawing a direct comparison between pharmaceutical companies engaging in well-financed political gamesmanship to ensure that profitable drugs could still be sold against medical advice, and energy companies funding for renewables. This is a bizarre comparison that makes no sense, when the obvious comparison is one of energy companies preserving their existing profits and manufacturing controversy to discredit scientific advice that harms their business. You chose to overlook the obvious comparison in favour of a tenuous (nay, nonsensical) one that suited your own personal bias.

        It is *that* that I was drawing attention to.

      • Dave H,
        Greenwashing must be expensive, since energy companies are spending vastly more money on the green side than the skeptic.
        The implication to me is that the skeptical side must be really strong to take so little support and use it to result in what we see today.
        It is odd that you are sticking to your story that energy companies are funding skeptics more than greens. I suggest you re-read the links I provided. The fact is that energy companies are funding green causes directly and alternative energy directly orders of magnitude more than AGW skeptical causes. The best I have seen supporters of the AGW narrative come up with to support the idea of a vast big energy conspiracy supports my point: the money spent on skeptical causes is trivial, even using your definition of support.
        I am not sure where the pharma comparison came from, but this is a better one: is big pharma spending money to fund holistic or nutrient based therapies and to support promoters of non-pharma therapy alternatives?
        As to the clean air act, we are not talking about the clean air act. We are talking about using the idea that CO2 is causing a ‘global climate disruption’ to impose dramatic changes on our energy economy and everything depending on energy.
        You do keep savouring irony in my statements but apparently feel your beliefs are so strong they need no proof.
        Which comes back to the idea that Dr. Curry inspired, the monster promoter.
        The monster promoter depends on the credulity of those who listen to his or her promotion, and needs the audience to listen to their narrative exclusively.
        Here is an accounting of Enron’s role in developing the idea of cap and trade:
        Perhaps it would be more credible if you addressed the links- heck even read them- rather than dismiss them.

      • The amounts given by energy and other companies to alarmist and skeptics are by-theby – mere drops in the ocean compared to what the the state spends, virtually all of which goes on “consensus” alarmism – a cause in which it has a huge vested interest.

  40. One reaction to the unknown is phobia. When people fear something and they can’t see it (e.g., pesticides) they may demand that the level of that thing in their environment/food be reduced to zero, which of course is impossible. Similarly, a level of damage from climate change that might be a minor hassle (e.g., a few more droughts/tornados) because of its uncertainty may lead to demands that all coal plants be shut down, out of proportion to the risk but due to the psychic response to uncertainty.
    The unknown-unknown comes into play when we identify a process (human effect on climate) by subtracting the knowns from the historical record. But if there are unknowns then this is fallacious.
    Depictions of uncertainty such as the faint gray color of error bars can be designed to give the appearance of greater certainty than truly exists.
    Finally, uncertainty can directly affect an estimated quantity. If paleo data has dating error (and most of it does) then combining multiple proxies to create a reconstruction leads to a damping out of any peaks and valleys (a smearing of the signal)–Loehle, C. 2005. Estimating Climatic Timeseries from Multi-Site Data Afflicted with Dating Error. Mathematical Geology 37:127-140

  41. “A third type is the merchant of doubt, who distorts and magnifies uncertainties as an excuse for inaction for financial or ideological reasons.”
    The opposite is also true, where the uncertainty is used to claim that something truly horrible “might” happen, and thus we must take urgent and drastic action now, just in case.

    • Steve Fitzpatrick

      Hi Craig,

      I completely agree.

      Some time ago you published an analysis of ARGO heat content, 0-700 meters, confirming several other published reports of reasonably flat OHC since 2003. A recent article (I think a French group) suggests that the story told by ARGO since 2003 from the surface down to 2,000 meters is completely different, and their analysis appears to confirm very large increases in OHC from 2003 to 2009.

      Have you considered extending you analysis of ARGO to current data and down to 2,000 meters?

      • I am waiting for the update, coming soon. I have questions about the French group’s work, but not time to look into it right now.

    • ‘monster promoter’

  42. Does the Irreducibility of Uncertainty, provide evidence of the certainty of Irreducibility a climate paradox ?

    The paradox of the climate problem is the more research , or additional parameters etc that have been included into GCM experiments the greater the level of uncertainty and error in these experiments. This was succinctly defined in Ghil et al 2008.

    …The difficulty in narrowing the range of estimates for either equilibrium sensitivity of climate or for end-of-the-century temperatures is clearly connected to the complexity of the climate system, the multiplicity and nonlinearity of the processes and feedbacks it contains, and the obstacles to a faithful representation of these processes and feedbacks in GCMs. The practice of the science and engineering of GCMs over several decades has amply demonstrated that any addition or change in the model’s parametrizations” i.e., of the representation of subgrid-scale processes in terms of the model’s explicit, large-scale variables may result in noticeable changes in the model solutions’ behavior.

    • Steve Fitzpatrick

      “The paradox of the climate problem is the more research , or additional parameters etc that have been included into GCM experiments the greater the level of uncertainty and error in these experiments.”

      If true, then the only economically rational approach is to stop this work completely, or at a minimum reduce it’s complexity what maximizes useful information. I would not hold my breath….

    • There are no such thing as “GCM experiments”. Perhaps there are GCM runs or GCM scenarios, but there are no experiments.

      Or if you wish to characterize each scenario run through the GCMs as an “experiment”, then please don’t try to assert that the GCM results are anything more than fiction which doesn’t relate to anything we experience living here on the earth.

      • How many climate scientists are computer scientists or physicists, that understand the impossiblity of the long term predictions of what the ‘runs’ produce… or understand the impossibility of long term predictions using a computer model..

        Computer models do have their uses in climate science, think hurricanes, cloud formations, but the whole concept of GCM’s is to be frank laughable. Events/hypothesis in isolation, ocean temp cycles, cyclones, even moddelling longer regional cycles, may give useful answers (but the real world will not necessarily follow) but putting them all together. No

        It is the nature of the uncertainties involved, we are still at the assumptions stages, which make this doubly impossible…

        Even if, all factors were known, no unknowns, to a fraction of a tenth of 1% error, thes long term prediactions/simulatiuons are impossible to call experiments and a collosal waste of energy and time.. Getting bigger and faster (of course more expensive) computing power will not resolve this issue…

        In Working group 1, this is half acknowledged, but more time money and computers seem to be their solution…

      • There are no such thing as “GCM experiments”. Perhaps there are GCM runs or GCM scenarios, but there are no experiments.

        Or if you wish to characterize each scenario run through the GCMs as an “experiment”, then please don’t try to assert that the GCM results are anything more than fiction which doesn’t relate to anything we experience living here on the earth.

        This is well described in the literature eg Vladimir Arnold problems, and Yasha Sinai Russian problems,and were robustly debated at the first Andropov conference on open problems in nonlinear dynamics .

        Gallavotti summarizes well.

        The first disturbing and striking feature of 3DNS is that only a non constructive existence proof of a solution is available for general and smooth data and forces. This amounts to saying that unless one invokes statements about something being “physically obvious “ the numerical experiences concerning turbulence phenomena are in fact experiments on the program generating them, rather than on the idealized NS fluid.

        Models are by definition fictions,eg Arnold

        The ugly building, built by undereducated mathematicians who were exhausted by their inferiority complex and who were unable to make themselves familiar with physics, reminds one of the rigorous axiomatic theory of odd numbers. Obviously, it is possible to create such a theory and make pupils admire the perfection and internal consistency of the resulting structure (in which, for example, the sum of an odd number of terms and the product of any number of factors are defined). From this sectarian point of view, even numbers could either be declared a heresy or, with passage of time, be introduced into the theory supplemented with a few “ideal” objects (in order to comply with the needs of physics and the real world)…….

        ….At this point a special technique has been developed in mathematics. This technique, when applied to the real world, is sometimes useful, but can sometimes also lead to self-deception. This technique is called modelling. When constructing a model, the following idealisation is made: certain facts which are only known with a certain degree of probability or with a certain degree of accuracy, are considered to be “absolutely” correct and are accepted as “axioms”. The sense of this “absoluteness” lies precisely in the fact that we allow ourselves to use these “facts” according to the rules of formal logic, in the process declaring as “theorems” all that we can derive from them.

        It is obvious that in any real-life activity it is impossible to wholly rely on such deductions. The reason is at least that the parameters of the studied phenomena are never known absolutely exactly and a small change in parameters (for example, the initial conditions of a process) can totally change the result. Say, for this reason a reliable long-term weather forecast is impossible and will remain impossible, no matter how much we develop computers and devices which record initial conditions.

        In exactly the same way a small change in axioms (of which we cannot be completely sure) is capable, generally speaking, of leading to completely different conclusions than those that are obtained from theorems which have been deduced from the accepted axioms. The longer and fancier is the chain of deductions (“proofs”), the less reliable is the final result.

      • Tomas Milanovic

        Fully agree with everything in the post above.

  43. A little off topic but thought it was appropriate:

    In all science, error precedes the truth, and it is better it should go first than last.
    Hugh Walpole

    • And another, often thought -incorrectly- to be limited to the medical profession, “First, Do No Harm!”

  44. Dr. Loehle’s paleo data temperature reconstruction and Dr. Korte’s (from GFZ, Potsdam). paleomagnetic reconstruction produced (surprisingly) similar results.
    Is the degree of uncertainty within both reconstructions greatly reduced, or is there an ‘unknown unknown’ at play?
    Scientists are far too often afraid to apply the old adage ‘speculate to accumulate’, to promote the expansion of knowledge.

  45. As someone who works fairly regularly with reservoir models and reservoir simulations, which are simpler and which we have been doing for longer than climate models and which we understand the lower number of input variables better, I always like the somewhat crude saying we apply to the modelers belief in the models output. Models are like masturbation, the more you do it, the more you come to believe it represents reality, especially when you are first beginning.

  46. Tomas Milanovic

    This is off topic but it is something that could be important for larger discussions.
    As there is a possibility to have trees within trees by replying to replies, it becomes nearly impossible to follow the whole discussion when it goes beyond some threshold.
    The purely linear format has a better time arrow.
    But if trees within trees are possible then there should be a way to see where the last contributions are.
    Of course unless there is something that I have missed.

    • Tomas, I am trying to figure out better ways to do the comments, i hope to combine (linear) comment numbering with trees. I have someone working on redoing the website using sandbox, but that is taking time. in the mean time trying to implement some plug ins or something. I share your frustration

  47. I second Tomas’ motion about trees within trees, if there a feasible way to track the latest entries.

    • If not, please stick with the Trees. It’s not only Eco-Friendly but so much easier to seperate the wheat from the chaf.

  48. To track all posts – be them new posts or replies, – could simply a counter be used that counts the comments coming in, and that counter number be placed in front of the date stamp for each comment. Then the ‘threads’ or ‘trees’ of replies can remain (which I also like), but an instant reference to the ‘order’ of comments is right there also.

    I would think it should be fairly easy thing to do . . .

  49. Steve Fitzpatrick

    Hi Judith,

    The indenting of replies causes loss of continuity and confusion. Something much simpler (just enumerated comments for example) would be much better.


  50. “A third type is the merchant of doubt, who distorts and magnifies uncertainties as an excuse for inaction for financial or ideological reasons”

    Merchants of doubt come in two flavors. The one you’ve ommitted is the one who exploits uncertainty (abuses the precautionary principle) to advocate for action for financial or ideological reasons. This type is exceedingly common – funny they were overlooked.

    • That is what I have been trying people to see for sometime. Only you have said it better.
      I have tried naming them as ‘AGW promoters’ and now, thanks to Dr. Curry, I think the label ‘monster promoter’ may be worth a look.
      The classic musical, ‘The Music Man’ takes an entertaining look at this human foible.

  51. fizzy water solution

    Reply to Tomas at September 24, 2010 at 4:42: I will try to be a little more helpful.

    Reading the obvious papers of interest from the Google search result would reveal that, contrary to your assertion, the issues you refer to have been discussed extensively. One would also learn that forced systems, like climate, can have predictable components even if their unforced counterparts are chaotic. (Also, aspects of chaotic systems may be predictable for long periods of time, despite their formally chaotic nature.) So your assertions regarding the unpredictability of climate, based on the chaotic nature of weather, are not generally valid.

    For instance, if it was known that the solar constant was increasing at a rate of 30% per 100 years, would you say that the future average temperature of the Earth could not be quantitatively evaluated, because climate is chaotic?

    Perhaps you were restricting your comments to decadal variability. In that case I would agree with much (but not all) of what you said.

  52. Dr Curry
    Do you believe the strong ‘anthropic’ principle applies to the climate system?

    As in, “all climatic event possibilities can only act to enhance the flourishing of life-forms’?

  53. Shub, I am unfamiliar with the strong anthropic principle, but as stated I don’t believe it (mostly because I am unfamiliar with it and the arguments on both sides). Ask the dinosaurs, check out this cartoon .

  54. Thanks Judith for addressing this critically important and poorly understood issue in climate science and politics. You note:

    A scenario is a plausible but unverifiable description of how the system and/or its driving forces may develop in the future. 

    If it is “unverifiable”, is a “scenario” even within the realm of “science”?

    science: A branch of study in which facts are observed and classified, and, usually, quantitative laws are formulated and verified; involves the application of mathematical reasoning and data analysis to natural phenomena.

    McGraw Hill Dictionary of Scientific and Technical Terms 6th ed.

    Has Climate Science considered applying “scientific forecasting principles?
    See J. Scott Armstrong on Standards and Practices for Forecasting

    One hundred and thirty-nine principles are used to summarize knowledge about forecasting. They cover formulating a problem, obtaining information about it, selecting and applying methods, evaluating methods, and using forecasts.. . .

    Green and Armstrong applied these principles to: Global Warming: Forecasts by Scientists versus Scientific Forecasts Energy & Environment, Vol 18, No. 7+8, 2007. They found:

    “The forecasting procedures that were described violated 72 principles” (out of 89 of 140 for which they found information). “Are these forecasts a good basis for developing public policy? Our answer is “no”. . . .The forecasts in the Report were not the outcome of scientific procedures. In effect, they were the opinions of scientists transformed by mathematics and obscured by complex writing. . . .We have been unable to identify any scientific forecasts of global warming.”

    Could Climate Science learn and apply these scientific forecasting principles? cf
    Principles of Forecasting: A Handbook for Researchers and Practitioners J. Scott Armstrong.

    You note:

    Statistical uncertainty can be described adequately in statistical terms. An example of statistical uncertainty is measurement uncertainty, which can be due to sampling error or inaccuracy or imprecision in measurements.

    Most of the discussion here on “uncertainty” appears to address “Type B” uncertainty without ever mentioning these formal uncertainty classifications and quantifications. Could Climate Science bring itself to incorporate and apply the formal scientific definitions and methodology of “Uncertainty”?

    See NIST on Uncertainty of Measurement Results

    Type A evaluation
    method of evaluation of uncertainty by the statistical analysis of series of observations,
    Type B evaluation
    method of evaluation of uncertainty by means other than the statistical analysis of series of observations.
    Representation of uncertainty components
    Standard Uncertainty
    Each component of uncertainty, however evaluated, is represented by an estimated standard deviation, termed standard uncertainty with suggested symbol ui, and equal to the positive square root of the estimated variance
    Standard uncertainty: Type A
    An uncertainty component obtained by a Type A evaluation is represented by a statistically estimated standard deviation si, equal to the positive square root of the statistically estimated variance si2, and the associated number of degrees of freedom vi. For such a component the standard uncertainty is ui = si.
    Standard uncertainty: Type B
    In a similar manner, an uncertainty component obtained by a Type B evaluation is represented by a quantity uj , which may be considered an approximation to the corresponding standard deviation; it is equal to the positive square root of uj2, which may be considered an approximation to the corresponding variance and which is obtained from an assumed probability distribution based on all the available information. Since the quantity uj2 is treated like a variance and uj like a standard deviation, for such a component the standard uncertainty is simply uj.

    (See the original for the proper math notation with subscripts )
    See NIST’s Uncertainty Bibliography, especially Guidelines for Evaluating and Expressing the Uncertainty of NIST Measurement Results, B.N. Taylor and C.E. Kuyatt, NIST TN 1297

    Thanks for the link to David Spiegelhalter:

    We need clear language to honestly communicate deeper uncertainties with due humility and without fear

    . I particularly enjoyed the quote:

    “All models are wrong, and some are useful.”

    George Box

  55. *****
    fizzy water solution says:
    September 24, 2010 at 9:26 pm

    For instance, if it was known that the solar constant was increasing at a rate of 30% per 100 years, would you say that the future average temperature of the Earth could not be quantitatively evaluated, because climate is chaotic?
    I’m not so sure the answer is in the bag. Would the extra heat cause more clouds causing the albedo to go up? This is why clouds are so important and their effect must be elucidated.

    • fizzy water solution

      Jim – Yup, there would no doubt be feedbacks. But one would expect that a strong response to the forcing would be identifiable despite the coexistence of a chaotic component; that was the point of my comment.

      • This is an interesting question. A biological population equation used by Robert May did not exhibit chaos until the growth rate exceeded a certain magnitude. I wonder if more energy into the climate system would increase the “magnitude” of chaos?

        For the diagram showing May’s equation results as a function of growth rate:

  56. I would like to hear comments on this survey from climate scientist who believe global warming will be between 2 – 6 C/century and will lead to disaster.

    “Only 12% agree or strongly agree that data availability for climate change analysis is adequate. More than 21% disagree or strongly disagree.

    Only 25% agree or strongly agree that “Data collection efforts are currently adequate,” while 16% disagree or strongly disagree.

    Perhaps most importantly, only 17.75% agree or strongly agree with the statement, “The state of theoretical understanding of climate change phenomena is adequate.” And equal percentage disagreed or strongly disagreed.

    Only 22% think atmospheric models deal with hydrodynamics in a manner that is adequate or very adequate. Thirty percent (30%) feel that way about atmospheric models’ treatment of radiation, and only 9% feel that atmospheric models are adequate in their treatment of water vapor–and not one respondent felt that they were ‘very adequate.’

    And only 1% felt that atmospheric models dealt well with clouds, while 46% felt they were inadequate or very inadequate. Only 2% felt the models dealt adequately with precipitation, and 3.5% felt that way about modeled treatment of atmospheric convection.

    For ocean models, the lack of consensus continued. Only 20% felt ocean models dealt well with hydrodynamics, 11% felt that way about modeled treatment of heat transport in the ocean, 6.5% felt that way about oceanic convection, and only 12% felt that there exists an adequate ability to couple atmospheric and ocean models.

    Only 7% agree or strongly agree that “The current state of scientific knowledge is developed well enough to allow for a reasonable assessment of the effects of turbulence,” and only 26% felt that way about surface albedo. Only 8% felt that way about land surface processes, and only 11% about sea ice.

    And another shocker–only 32% agreed or strongly agreed that the current state of scientific knowledge is developed well enough to allow for a reasonable assessment of the effects of greenhouse gases emitted from anthropogenic sources.”

    • After reading the results of the survey, I have some comments. (I’m not a climate scientist.)
      1. A fairly high degree of confidence is placed in the temperature element while there is less confidence in some elements that affect temperature, especially water vapor, precipitation and clouds.
      2. In spite of an admission that there is a 4, generally 5s and 6s.

      These observations seem to point to an inconsistency between the science and the conclusions. The science seems less certain than the conclusions and that seems backwards to me.

      • “2. In spite of an admission that there is a 4, generally 5s and 6s. ” is supposed to be:
        2. In spite of the relatively low certainty of the science, the confidence in the conclusions tends to be greater than 4, generally 5s and 6s

      • At first glance I would agree that this seems backwards, having more confidence in the conclusions than in the data and science. However when I stopped to think about it for a minute from my own experience, I have to say I would come to the same viewpoint. I know the limitations and inaccuracies of the data I work with and I know the state of knowledge of the science, so I have to accept their limitations. But I use that science and data to come to conclusions and make predictions that I am fairly confident of. I’m making a conclusion of the most likely correct scenario from a known set of limitations. Of course almost as soon as I reach that conclusion and then test the prediction I am often proven wrong, which means I go back, re-interpret, re-evaluate and come to a new, hopefully better conclusion and also advance my understanding of the science. This is how science works, constantly testing and re-evaluating, a progression to build on.

      • I wonder if Jim Hansen was as confident as you when he made the prediction 20 years ago that parts of New York would be under water and some species would be lost by now. I think he has harmed climate science as well as science in general.

  57. Judith – I looked through quite a few comments, and did not see this mentioned, so I had to add it to the discussion. Your inclusion of Rumsfeld’s discussion of known and unknown information was certainly appropriate – it’s the first thing I though of as I began to read your post. I’m no fan of Rumsfeld, but I find it unfortunate that you describe his discussion as ‘infamous.’ The word means ‘deserving infamy.’ Don’t you mean famous? When he was criticized at the time, I thought the people who did so embarrassed themselves. His statement was both clear and perfectly reasonable. The criticism – and ridicule – were disingenuous at best.

  58. I don’t know when we will cool significantly due to the normal glaciations cycle, but cooling certainly is a bigger problem than the putative warming. This is an interesting twist from the last Bilderberg meeting. They speak of global cooling as a problem. If this isn’t a typo and we see climate scientist get off the global warming bandwagon and on to a global cooling one then climate scientists in general will have zero credibility.

    “Which is what makes one particular item on the group’s discussion agenda so tremendously significant. See if you can spot the one I mean:

    The 58th Bilderberg Meeting will be held in Sitges, Spain 3 – 6 June 2010. The Conference will deal mainly with Financial Reform, Security, Cyber Technology, Energy, Pakistan, Afghanistan, World Food Problem, Global Cooling, Social Networking, Medical Science, EU-US relations.

    Yep, that’s right. Global Cooling.

    Which means one of two things.

    Either it was a printing error.

    Or the global elite is perfectly well aware that global cooling represents a far more serious and imminent threat to the world than global warming, but is so far unwilling to admit it except behind closed doors.

    Let me explain briefly why this is a bombshell waiting to explode.”

  59. There is no “uncertainty monster” in science; there is just uncertainty. Using the typical requirements for putting conclusions in the abstract of a scientific journal, uncertainty in science typically persists until statistical analysis of experiments yields a p value of <0.05.

    Uncertainty only becomes a "monster" when activists want to used science to advocate for legislation. Terms like "more likely than not", "likely", and any "level of scientific understanding" below "high" are indeed monsters. It is a joke to refer to a scientific consensus about areas where the level of scientific understanding is not high. Does anyone really believe in the idea of a SCIENTIFIC consensus when the science isn't reasonably well known? Why are the IPCC's SPMs are littered with these terms?

    When it comes to public policy about AGW, the REAL uncertainty monster is guessing what issues will be important to society 50 and 100 years from now.
    To paraphrase Arthur Eddington, not only will the future be stranger than we imagine, it will be stranger than we can imagine.

  60. I see a far greater trade by “merchants of doom” than by “merchants of doubt”. For every merchant of doubt (I’m not sure if I could even identify one good candidate) — i.e., people whose primary income is based upon sowing seeds of doubt — you can produce I’m sure I can produce 10 individuals who live high on the hog doing little but promoting doom and gloom climate scenarios based on the increasingly feeble AGW hypothesis. It’s a many-billion dollar industry. Perhaps hundreds of billions. Perhaps even higher, if you take into account tax and trade schemes designed to do nothing more than regulate “carbon”. I’m just thinking of the government largess that’s been pumped into the “climate change” community through governments participating in the UN, through handsome research grants like the one recently announced at the university where I teach, which proudly asserts that this will make the school a world leader in Arctic Ice research. Many new staff, new infrastructure (buildings) on campus, new equipment, new remote facilities, etc.

    The individual being brought in to the research chair thus created gave an interview in the local press, and was asked to sketch out what they hoped to accomplish. The very FIRST thing he said was that they were going to mount a campaign to counter skeptic and “denier” claims. Message received, loud and clear: a giant, richly endowed propaganda engine. There’s nothing like this in the whole skeptic landscape. Likely the total national funding for individual scientists whose work runs contrary to “the AGW narrative” would fit in the back pocket of the budget for this new center, one of dozens like it springing up around the world. Please don’t lecture about “merchants of doubt” — I’m far more concerned about the other kind.

  61. Alexander Harvey

    I find that the (un)known (un)known characterisation to be very wooly and opaque. The terms known and unknown (whether adjective or noun) are too broad and can be lazy stand-ins for better terms particularly (perhaps universally) when in repetition (adjective followed by noun). Much seems to depend on how one defines “known” both as adjective and noun and how we use “to know” as a verb.

    What do I know I don’t know? What can I describe or characterise that I don’t know?

    I might say that I know that I don’t know French, but I would be a liar, I do know French, it is the language spoken in France. My knowledge of French is limited, very limited. I can say that I know my knowledge of French is very limited but that is not the same as not knowing French or worse maintaining that I know I don’t know French. I simply cannot not characterise or describe things of which I have no knowledge.

    You write: “The unknown unknown corresponds to ignorance, …” I am not sure how class of things we are ignorant of, differs from the class of things that are we don’t know (the unknowns) except in one important case.

    What things do I not know that I do not know, beyond what I don’t know?

    The set of things that, I both know and don’t know I don’t know, is (or contains) the set of things that I know that are false.

    To me, it would be better to characterise your unknown unknowns as just the things I have no knowledge of (unknowns) , plus things that I erroneously know to be true.

    That is part of my point; the set of known knowns (knowns) includes the most dangerous parts of the set of unknown unknowns: the things that we know that ain’t so (Rogers/Twain/Traditional).

    So my point is that I can only understand your phrase when ignorance is suplemented with fallacies (there is a difference), so I must subdivide the knowns (or known knowns as they are here known to be known) into those that are true and those that are erroneous. This is of some practical imporance for should I wish to know more, I would not neglect to examine what is known for error.

    I do hope that this is not perceived as pure pedantry, it is not intended that way.


  62. Alexander Harvey

    You wrote: “The climate system is stochastically uncertain because of its chaotic nature, i.e. small differences in the initial conditions of a global climate model can yield very different results.”

    That statement is open to challenge on many fronts. Perhaps few would be bothered with it in the way that I am. Perhaps you could clarify.

    Are you saying the climate system is stochastic (non-deterministic) because it is chaotic (deterministic)?

    Are you saying the climate system is stochastic (non-deterministic) because climate models are chaotic (deterministic)?

    Are you saying that macroscopic climate metrics are stochastic (not-determined by past values) due to the amplification of an inherent indeterminacy by the chaotic nature (deterministic) of the climate system?

    Are you saying that macroscopic climate metrics are stochastic due to the underlying chaotic nature (deterministic) of the climate system?


    • More coming soon on the complexity of the climate system and challenges to climate models

      • Alexander Harvey

        To curryja:


        what bothers me in those dark moments is that the metrics are chaotic as opposed to stochastic. To me there is a big difference between uncertainty due to randomness and due to lack of knowledge when it comes to statistical inference. If the metrics can be modelled as the result of signal plus noise applied to a process (model/filter) then we can use statistical inference from stochastic models, if they are deterministic but unpredictable due to lack of complete knowledge then statistical inference based on stochastic models is highly questionable. The question “How often could this have happened by chance (confidence intervals etc.)?” in a chaotic (deterministic) system is (besides being meaningless) quite a different question to when considering stochastic systems. It becomes a question about the initial conditions (and externals).

        In chaos theory, as taught way back when, uncertainty was due to lack of knowledge and none of it due to randomness so classical statistical methods were not applicable.

        Obviously I do not know if someone has proved that the physics of the climate system gives rise to metrics that are essentially indistinquishable from stochastic behaviour, if not this is a painful and debilitating known unknown.


      • Alex, we seem to be in a situation with climate models characterized by painful known unknowns. I am working on a post about this, and it is quite a struggle.

      • More to Alexander than Judith. If you conceive of the weather as a trajectory through a parameter space it is true that trajectories arbitrarily close to each other at some point can diverge, on the other hand there are regions of the space where the trajectories cannot go, and others which are frequently visited. The description of the rules governing recurrence time is climate. This is a somewhat different take on the old saw that weather is an initial value problem and climate a boundary value one.

      • Alexander Harvey


        No salient disagreement bar a little wording, we all have differences in emphasis.


        “… on the other hand there are regions of the space where the trajectories cannot go, and others which are frequently visited.”

        Becomes my:

        … on the other hand there are regions of the space where the trajectories are “sparse or extremely sparse unless due to external perturbation” , and others “in which trajectories are dense”.

        With regard to boundary conditions I personally dislike the distinction as I find it unhelpful. The constriants are the same for both climate and weather. In both cases one must consider the tendency for the system to be in states that are trajectory rich, the problem being that it is not obvious if ones choice of initial state represents a dense or a sparse region of the system space.

        In climate modelling I beleive that they spin up for a few decades to remove the effects of inital value perturbation. Whether that is an indication of the time constants embodied in the acutal climate is moot.

        If the system is chaotic but reasonably well damped then some of the long term wiggles may reflect the underlying trajectories, alternatively they may just be stochastic artefacts. In the later case we can apply classical statistics and argue about significance; if they are representative of underlying trajectories then we must be cautious about how we apply statistical techniques.

        My main point was that we make use of the assumed stochastic nature of climate metrics frequently in statistical analysis without (as far as I am aware) any proof of the validity of such an approach and with an awareness that it may be lacking in rigour.

        It is never clear how metrics can exhibit stochastic behaviour when the underlying system is deterministic albeit chaotic but it seems to be the case just as times arrow seems never to waver.


      • It would be interesting to hear your WAG on the recurrence time of the Earth’s climate system.

  63. Judith,
    this is a great post. I think the phrase “(t)here can also be epistemic uncertainty about how a physical, chemical or biological process works is extremely important. Scientists suffer from “completeness fallacy” meaning they need to carry through work to some meaningful level of completion (for reasons such as need to mark progress, need to publish, need for peer recognition, need to meet deadlines for IPCC assessments, etc.). A model which lack sufficient treatment of all suspected independent variables – and their interaction is – under- ie., mis-specified. Result is that the model will tend to over-emphasize or under-emphasize attribution for the variables that are fully treated. I have long suspected this to be the case in climate modeling for global warming. There are many forcings and some are known to be underrepresented in the modeling such as aerosols / clouds and black soot.

  64. There’s uncertainty everywhere !

    Taking ‘stumper’ to roughly mean, We don’t know.

  65. I actually think so too:P I have been looking around the internet for some time this week, and its kinda hard to find anything interesting to read on blogs=) Maybe its because there are too many of those around =) But this website actually keeps catching my attention:P Great stories, and kawai design ^__^. Ill be sure to give it more visits from now on =P

  66. Hi! I found your blog on AOL.It’s really well written and it helped me a lot.

    Continue the good work!

  67. Andrew Ansari

    Firstly may heartfelt best wishes and congratulations to Judith Curry.
    Great Blog! I am not a scientist, however a practising medical specialist.
    All through my medical training “science ” led us to believe that estrogen was beneficial to postmenopausal women. The dogma was so loud that anyone suggesting otherwise would be labelled a crazy heretic. The science was overwhelming. A single study by the WHI unexpectedly found that the opposite was true and the there were greater risks than benefits. No such study can be conducted in climate science. So we are left with uncertainty.
    My gut instinct is that herd mentality is driving climate scientists.
    Results supporting the theory are published , results negating the theory are not making it into publication.
    I was interested to see the vitriolic attacks on the Lindzen and Choi paper.
    In the presence of uncertainty I would suggest we pick the low hanging fruit and reduce emissions where the cost is minimal.
    I suspect there will be no warming in the next ten years and less than one degree in the next hundred years.
    Global warming will turn out to be a scientific fad , some scientists will make a great career out of it , and be lauded in the mean time , but in a hundred years this will be long forgotten .
    When i started medical career duodenal ulcers were a stress related illness.
    Now we know H pylori is the culprit.
    In hundred years it will be accepted that climate change is natural and a boring subject once again.
    Sadly scientists are no different the rest of the human species and behave badly and are prone to corrupt the peer review process.
    Climategate was a sad and stunning example.
    I will be insulating my loft, buying low energy lightbulbs and turning off the lights. However I wont be buying a toyota Prius or abandoning air travel.

    Hopefully those in power won’t tax me too heavily in the mean time.
    Keep on debating!

  68. Denis Diderot ( ) :
    “Ce qui caractérise le philosophe et le distingue du vulgaire, c’est qu’il n’admet rien sans preuve, qu’il n’acquiesce point à des notions trompeuses et qu ‘il pose exactement les limites du certain, du probable et du douteux.”

  69. Appreciating the time and effort you put into your blog and comprehensive info you provide! I’ll bookmark your blog. Thumbs up!

  70. Hi there. I found your site via Yahoo and your post looks very interesting for me.

  71. i like it..

  72. I want to voice my passion for your kindness giving support to those people who require help on that subject matter. Your very own commitment to passing the solution throughout had been amazingly valuable and have consistently enabled girls just like me to arrive at their ambitions. Your important guide indicates a lot to me and somewhat more to my fellow workers. With thanks; from all of us.

  73. Fantastic mate!
    Fantastic article ! You havemade some very astute statements and I appreciate the the effort you have put into your writing. Its clear that you know what you are writing about. I am excited to read more of your sites content.

  74. Whenever, I felt uncertainty, I use different strategies to minimize the uncertainty.