A commenter mentioned ‘groupthink’ in a recent post, and that reminded me of a piece that Judith Curry put on her excellent website a fortnight ago. The title there was ‘Mutually Assured Delusion’, an allusion to the ‘Mutually Assured Destruction’ notion of the hotter parts of the Cold War. Those who think together, you might say, can be assured that they will be equally wrong. Dr Curry thinks that too many of the orthodox suffer from ‘groupthink’, and that it is bad for them as scientists. It is hard to disagree.

A bit of intellectual history is in order.The term is a borrowing from George Orwell (‘doublethink’ in 1984), via William Whyte in the 1950s, who actually coined the word ‘groupthink’, to Irving Janis in 1972, who wrote a book about it, to a Frenchman, Roland Benabou, who won a PhD at MIT and now teaches at Princeton; he has written a recent paper on it (‘Groupthink: Collective Delusions in Organizations and Markets’), and that paper is where ‘MAD’ comes from.

What characterises groupthink? Janis offers eight symptoms:

* the illusion of invulnerability

*  collective rationalization

*  a belief in inherent morality

stereotyped views of out-groups

* direct pressure on dissenters


* the illusion of unanimity

self-appointed mind guards

Benabou thinks that it is in any organisation’s interest to make sure that groupthink is easily distinguished from group morale:  organizations and societies find it desirable to set up ex-ante commitment mechanisms protecting and encouraging dissent (constitutional guarantees of free speech, whistle-blower protections, devil’s advocates, etc.), even when ex-post everyone would unanimously want to ignore or ‘kill’ the messengers of bad news.

Dr Curry comments that ‘in the case of AGW, we have a scientific debate/disagreement about a highly uncertain and complex system.  Acknowledging the complexity and uncertainty is key to generating a willingness to listen to different ‘prophecies’ of what the future might hold’. She has no doubt been saying things like that longer than I have, but my abundant posts on AGW carry the same message: why do we only hear about the evil consequences of warming, and never of the benefits, which have been many already? And why do we not hear about the many different climate prospects for the future?

Benabou is really talking about how businesses operate in market settings, but the following little statement is easily applied to global warming and ‘climate change’: Groupthink is thus most important for closed, cohesive groups whose members perceive that they largely share a common fate and have few exit options. Amen. He says that a major ‘source of group error is social pressure to conform, and that can best be defeated if dissent is encouraged.

Benabou lists seven ‘patterns of denial’, which I found somewhat delicious, since for him the deniers are those enclosed in groupthink. Some of them are so familiar in the global warming domain, like ‘preposterous probabilities’, and ‘new paradigms’ (this time is different, we are smarter and have better tools. Every case also displays the typical pattern of hubris, based on claims of superior talent or human capital).

Here are some more: ‘wishful beliefs’, information avoidance’, and the ‘normalisation of deviance’ (How do organizations react when what was not supposed to happen does, with increasing frequency and severity?). Two more: ‘reversing the burden of proof’ and ‘malleable memories: forgetting the lessons of history’.

I recognise that we are all likely to prefer information that accords with our own view, but I was trained in university  to examine evidence critically, and the instruction went deep. What is more, while it is easy to picture only two sides in the debate, the orthodox and the dissidents, and I use these terms myself, the dissidents are hard to group into a single team. They come from all points of the compass, and often disagree with one another, as you can see on this website, let alone on Judith Curry’s.

In contrast — at least, as it seems to me — the orthodox have a church, a holy book (the current Assessment Report of the IPCC), and elaborate defence mechanisms to defend the writ against attack. All that makes them powerful and so far successful, but it does also make them vulnerable to new evidence. In some fields you can divert new evidence with subsidiary hypotheses, but in ‘climate change’ it is Nature that provides the new evidence, and it does so every day.

The failure of the models to track what actually happened, and the prolonged pause in warming, can be diverted for a while. But they cannot finally be ignored.

Join the discussion 7 Comments

  • Mike O'Ceirin says:

    I think groupthink is the most important and the most damaging
    characteristic of humanity. When I say group I mean all groups, Greens,
    Communists, Catholics, Left, Right, Christians, Moslems, Democrats, Skeptics,
    Republicans, ABC, Pastafarians and so on. All need a common belief system which
    cannot be questioned if the group is to prosper. Those who lose faith and
    question it are cast out. This common faith produces strength and community but
    can be very damaging to free thought and those outside the group. Skeptics if
    they truly are the doubting Thomas’s who question all and apply scientific
    method make poor members of groups. The Skeptics Society of Canberra is very much a group driven by belief which is only skeptical about a narrow spectrum of
    matters. Groups can also be blind to the dangers of their belief system.

    • Don Aitkin says:

      I agree. One of my sons has reminded me that I am not sceptical about everything — life is too short. It’s probably true that we all ‘take for granted’ a great deal about the world and humanity without ever examining it closely, just as we assume that those who agree with us on X are likely to agree with us on Y as well, and we are surprised when it turns out not to be so.

  • David says:

    Yes prediction is important. But it is not the only aim of a statistical model and in my view you place too much emphasis on the capacity of climate model to “predict”. There are two reasons why a scientist constructs a statistical model.

    (a) Is to predict

    (b) Is to explain a causal relationship

    The two goals are related but distinct. Prediction is measured by the R-squared. Explanation will be measured by the p-value around and individual coefficient. A model with a low R-squared, can still useful, if the p-value around a coefficient of interest is statistically significant.

    For example, the statistical models for comprehensive car insurance have an R-squared of “only” 10% to 15%, which nominally is quite low. However the insurance firm will not care if they can not accurately predict the probability that an individual driver will have a traffic accident. What they do care about is the ability of their statistical model to quantify relationships between data they collect such as age and probability of an accident, so as to set premiums which generate a profit.

    Similarly, climate models are not only interested in predicting actual the temperature is 2060, they will also want to say something about the affect that CO2 has on temperature. The utility of a climate model is not solely determined by its R-squared.

    • Don Aitkin says:

      I agree, to a point. And the blame for excessive interest in prediction should probably be levelled at those outside the modelling community who used the models for their own ends.

      • AlainCo says:

        prediction is today so desired by society that people pretend to predict details when that seldom can prove reality.

        It is wishfull thinking that push so much support.

        one effect well identified is the Streetlight effect.

        people use what they is available, and pretend the rest does not exist.

        You see that ins finance risk management, in climate science… very often the key parameter in models are just “estimated by the nose” (sorry french idiom), and there is room for wishfull thinking and recursive reasoning.

        let us assume I am right, then this parameter is such, and this is not impossible. so my model says I am right. victory!

        typical climatology.

        ther is a variant of that reasoning about scientific denial, that nearly get written and sanctifies by mainstream:

        it is impossible to work because physics have no explanation (forget to say that nothing forbid it except lack of imagination).

        -any positive results is thus an experimental artifact
        -if the artifact is impossible, then it is a fraud.

        -thus if a scientist find a positive result, he is incompetent
        -if he cannot incompetent because of his past credential, he is deluded and frauding (or paid by oil companies).
        -if an industrial is supporting them, he is manipulated.
        -if he cannot be manipulated, he is a scam artist.
        -if he cannot be a scam artist… he does not exist, does not say so.

        it is the official vision supported by top science, wikipedia, sciam, and high impact journals.

        -if a paper is positive you have to reject it during the peer review
        -if a peer-review is positive (? how)
        -then you have to dump the article and then dump any article before peer-review

        it is currently implemented by high impact journal

        not a joke.

        recursive reasoning is the basic of many consensus, and many models.

        the key is the asymmetry of checking critics. no checking is require because “extraordinary claims requires extraordinary evidences” (nothing more stupid. any claim requires good evidences, no less, no more, even if consensual).

        One of the tools to implement the bias, is the null hypothesis.

        in fact my method is dual analysis. assuming various hypothesis are true, and checking if they match the facts.

        When I ear a critic, I rebuild the history according to the assumptions, and see if all the facts match…

        and I compare with various alternate hypothesis.

        I notice that if you assume conspiracy (like that all participant in a domain are deluded or corrupted) then you can protect your brain from dissenters ideas…

        You may be right, but you cannot be sure, because both beliefs are not less credible, assuming they are right.

        Often to get out of that twin vision, you have to use psychology (but many bad psychology is spread in the media) to rule out some conspiracy style.

        black cabinet conspiracy are not common (all is simply public,clear & ignored like what Snowden and Mannings revealed), but group delusion and elite screening is very common in history.

        “Only puny secrets need keeping. The biggest secrets are kept by public incredulity.” (Marshall McLuhan)

  • AlainCo says:

    One point that make me react is that scientist are taught to be critic.

    In fact it is not the solution, but often the problem with groupthink.
    If you see the real way that critical sense is used by scientist, it is to bash the claims that they don’t like, to throw away innovators, and to critics others critics on ones work…
    Others profession are no better, but they don’t claim to be rationally critic… they just have practical beliefs, and try to avoid errors… result is not far.

    Every day I see people who are skeptical of what they dislike like true-believers believe in what they love.

    My experience, using models likes benabou, is to think that our rationality, critical sense, beliefs, and delusions, are sadly guided
    – by our incentives
    – by our group dependencies
    – by our history, and the history of our groups

    I have been luck despite my culture (very mainstream) to be allowed, by mistake to be aware of some inconvenient reality… just because I was not reading the news when finishing my MsC internship.. 3 years later, I looked on Internet, not the web, and found data that I could interpret without any beliefs. and 3 years later, all was clear, but not consensual. because the scientific consensus get frozen in groupthink, unable to adapt to new data… It is still frozen for academic scientist, but dissolving for industrialists…

    You talk of climate, but at least in climate data are ambiguous.
    I talk you of something sound as hard science, with signals above 50sigma, with measurement using 100 years old science, and yet recent instruments…
    and today scientists doubt on, instruments used by engineers and physicist across the planet, just not to admit the simple flat reality….
    That they screwed up, that that talked too fast, though too short, forget basic evidences, insulted too fast, ruined lives and careers, defrauded, behaved like mafia boss or mafia gunmen….
    Now, as Benabou explain, it is too late. they invested too much in their mafia gang, and they cannot go out, else in a wood box nailed by their brothers of error. and there are street-kids of science that follow their beliefs because the seems to be powerful and protected.

    really climate is not the worst scientific delusion in the scientific domain.

    and by the way it is an uninteresting problem, since it is solved without any cost.

    • Don Aitkin says:


      Most interesting comment, thank you. It reminds me of the difficulty that geologists and other scientists had in coming to terms with Wegener’s theory of plate tectonics, and how he and his idea were laughed at for thirty years and more.

Leave a Reply