Less Wrong is a community blog devoted to refining the art of human rationality. Please visit our About page for more information.

Evaporative Cooling of Group Beliefs

41 Post author: Eliezer_Yudkowsky 07 December 2007 11:08PM

Followup toUncritical Supercriticality

Early studiers of cults were surprised to discover than when cults receive a major shock—a prophecy fails to come true, a moral flaw of the founder is revealed—they often come back stronger than before, with increased belief and fanaticism.  The Jehovah's Witnesses placed Armageddon in 1975, based on Biblical calculations; 1975 has come and passed.  The Unarian cult, still going strong today, survived the nonappearance of an intergalactic spacefleet on September 27, 1975.  (The Wikipedia article on Unarianism mentions a failed prophecy in 2001, but makes no mention of the earlier failure in 1975, interestingly enough.)

Why would a group belief become stronger after encountering crushing counterevidence?

The conventional interpretation of this phenomenon is based on cognitive dissonance.  When people have taken "irrevocable" actions in the service of a belief—given away all their property in anticipation of the saucers landing—they cannot possibly admit they were mistaken. The challenge to their belief presents an immense cognitive dissonance; they must find reinforcing thoughts to counter the shock, and so become more fanatical.  In this interpretation, the increased group fanaticism is the result of increased individual fanaticism.

I was looking at a Java applet which demonstrates the use of evaporative cooling to form a Bose-Einstein condensate, when it occurred to me that another force entirely might operate to increase fanaticism.  Evaporative cooling sets up a potential energy barrier around a collection of hot atoms.  Thermal energy is essentially statistical in nature—not all atoms are moving at the exact same speed.  The kinetic energy of any given atom varies as the atoms collide with each other.  If you set up a potential energy barrier that's just a little higher than the average thermal energy, the workings of chance will give an occasional atom a kinetic energy high enough to escape the trap.  When an unusually fast atom escapes, it takes with an unusually large amount of kinetic energy, and the average energy decreases.  The group becomes substantially cooler than the potential energy barrier around it.  Playing with the Java applet may make this clearer.

In Festinger's classic "When Prophecy Fails", one of the cult members walked out the door immediately after the flying saucer failed to land.  Who gets fed up and leaves first?  An average cult member?  Or a relatively more skeptical member, who previously might have been acting as a voice of moderation, a brake on the more fanatic members?

After the members with the highest kinetic energy escape, the remaining discussions will be between the extreme fanatics on one end and the slightly less extreme fanatics on the other end, with the group consensus somewhere in the "middle".

And what would be the analogy to collapsing to form a Bose-Einstein condensate?  Well, there's no real need to stretch the analogy that far.  But you may recall that I used a fission chain reaction analogy for the affective death spiral; when a group ejects all its voices of moderation, then all the people encouraging each other, and suppressing dissents, may internally increase in average fanaticism.  (No thermodynamic analogy here, unless someone develops a nuclear weapon that explodes when it gets cold.)

When Ayn Rand's long-running affair with Nathaniel Branden was revealed to the Objectivist membership, a substantial fraction of the Objectivist membership broke off and followed Branden into espousing an "open system" of Objectivism not bound so tightly to Ayn Rand.  Who stayed with Ayn Rand even after the scandal broke?  The ones who really, really believed in her—and perhaps some of the undecideds, who, after the voices of moderation left, heard arguments from only one side.  This may account for how the Ayn Rand Institute is (reportedly) more fanatic after the breakup, than the original core group of Objectivists under Branden and Rand.

A few years back, I was on a transhumanist mailing list where a small group espousing "social democratic transhumanism" vitriolically insulted every libertarian on the list.  Most libertarians left the mailing list, most of the others gave up on posting.  As a result, the remaining group shifted substantially to the left.  Was this deliberate?  Probably not, because I don't think the perpetrators knew that much psychology.  (For that matter, I can't recall seeing the evaporative cooling analogy elsewhere, though that doesn't mean it hasn't been noted before.)  At most, they might have thought to make themselves "bigger fish in a smaller pond".

This is one reason why it's important to be prejudiced in favor of tolerating dissent.  Wait until substantially after it seems to you justified in ejecting a member from the group, before actually ejecting.  If you get rid of the old outliers, the group position will shift, and someone else will become the oddball.  If you eject them too, you're well on the way to becoming a Bose-Einstein condensate and, er, exploding.

The flip side:  Thomas Kuhn believed that a science has to become a "paradigm", with a shared technical language that excludes outsiders, before it can get any real work done. In the formative stages of a science, according to Kuhn, the adherents go to great pains to make their work comprehensible to outside academics.  But (according to Kuhn) a science can only make real progress as a technical discipline once it abandons the requirement of outside accessibility, and scientists working in the paradigm assume familiarity with large cores of technical material in their communications.  This sounds cynical, relative to what is usually said about public understanding of science, but I can definitely see a core of truth here.

My own theory of Internet moderation is that you have to be willing to exclude trolls and spam to get a conversation going.  You must even be willing to exclude kindly but technically uninformed folks from technical mailing lists if you want to get any work done.  A genuinely open conversation on the Internet degenerates fast.  It's the articulate trolls that you should be wary of ejecting, on this theory—they serve the hidden function of legitimizing less extreme disagreements.  But you should not have so many articulate trolls that they begin arguing with each other, or begin to dominate conversations.  If you have one person around who is the famous Guy Who Disagrees With Everything, anyone with a more reasonable, more moderate disagreement won't look like the sole nail sticking out.  This theory of Internet moderation may not have served me too well in practice, so take it with a grain of salt.


Part of the Death Spirals and the Cult Attractor subsequence of How To Actually Change Your Mind

Next post: "When None Dare Urge Restraint"

Previous post: "Uncritical Supercriticality"

Comments (41)

Sort By: Old
Comment author: Doug_S. 08 December 2007 12:22:15AM 9 points [-]

If you eject them too, you're well on the way to becoming a Bose-Einstein condensate and, er, exploding.

Hmmm... to continue the metaphor more directly, their brains get frozen and can no longer update their beliefs?

Comment author: Carl_Shulman 08 December 2007 12:24:37AM 3 points [-]

Apparently debiasing techniques and encouragement of dissent played a role in the American re-evaluation of the Iranian nuclear program:

http://www.nytimes.com/2007/12/05/washington/05intel.html?ref=middleeast "Over the past year, officials have put into place rigorous new procedures for analyzing conclusions about difficult intelligence targets like Iran, North Korea, global terrorism and China.

Analysts from disparate spy agencies are no longer pushed to achieve unanimity in their conclusions, a process criticized in the past for leading to “groupthink.” Alternate judgments are now encouraged.

In the case of the 2007 Iran report, “red teams” were established to test and find weaknesses in the report’s conclusions. Counterintelligence officials at the C.I.A. also did an extensive analysis to determine whether the new information might have been planted by Tehran to throw the United States off the trail of Iran’s nuclear program.

One result was an intelligence report that some of the intelligence community’s consistent critics have embraced."

Comment author: Eliezer_Yudkowsky 08 December 2007 12:25:14AM 7 points [-]

There's no particular reason to want to continue that metaphor more directly. Only mathematically precise metaphors have that kind of power.

Comment author: Nominull2 08 December 2007 12:27:49AM 6 points [-]

You don't believe in the artistic value of a beautifully extended metaphor?

Comment author: denis_bider 08 December 2007 12:49:00AM 3 points [-]

Unquestionably, things get done a lot more by groups of people who are very much alike. Differences in opinions only tend to brake things.

The question is not whether you need people who are different in order to brake the group. The question is whether you're in the right group to begin with. As per Kuhn, things will get done faster and better if members of the group share a lot of commonalities.

If you're in the right group, excluding dissenters will allow you to progress faster. But if you're in the wrong group, then you're going to be making progress towards the wrong things.

Comment author: manuelg 08 December 2007 12:51:11AM 1 point [-]

Have you read Bion's "Experiences in Groups"? He was an English Freudian, so he was extremely passive while observing group behavior, which is fine, because he was also careful to record what was happening.

I am less satisfied with his analysis, because, as a typical Freudian, he always has ad-hoc reasons why any piece of evidence (or its exact opposite) perfectly confirms his theories. Absolutely impossible to falsify.

What I took from it was that, after you establish a concrete, positive goal for a group's interactions, for each and every sub-element of the goal, you can find some element of the human personal dynamic, or the human group dynamic, that will work against it.

It is a strong statement along the lines of Murphy's Law: "Whatever can go wrong, will go wrong". If it is a sub-element of a concrete, positive goal, there _will_ be, in opposition, some element of the human personal dynamic, or the human group dynamic. It is a strong statement you can use to predict failure modes of your particular work group.

*** If great progress is being made by the group, the individual identities of the members will lessen in importance. So people will assert their individual identities through disruption, gaining attention.

*** If progress of the group demands full attention on the final objective, the group will become paranoid and invent some internal or external bug-a-boo to focus on instead.

My personal take, informed by Buddhism, is that this is not necessarily a bad thing. There _should_ be some push-back on the goals of working groups. 99 out of 100 ideas turn out to be terrible ideas, and so it is a _good_ thing they die from the irrational failure modes of the human personal dynamic or the human group dynamic. If it is a good idea, then it is worth to take care from assaults from human irrationality.

And also, attempting to attack head-on any particular irrational failure mode will only make it stronger. For example, a troll will never have so many defenders as when the group leaders focus in to remove him. Better to use Jujutsu. (Trolls are best countered with neglect leading to boredom.)

Bion's "Experiences in Groups" will give a good sample of failure modes. You can attempt to skillfully steer around them, and keep the group working on the positive goal.

Yours is an interesting idea, keeping a "token" troll around. I would make a rule: any discipline against a troll will be matched by identical discipline against anyone who engages that troll, even in attack. "Feeding the troll" will have precisely the same sanctions as trolling itself.

Comment author: Adirian 08 December 2007 01:05:11AM 1 point [-]

I've believed this was so for a long time, after reading about the Klu Klux Klan. Which, it is weird to think, was once a (for the time) moderate and respected political institution. After the Grand Dragon got caught in a sex scandal, membership dwindled rapidly, leaving a core of fanatics - with results that I think we're all quite familiar with.

I think the schism in Objectivism ended differently - because I don't think on the one hand you have the fanatics, but rather simply two groups, each of which ended up defining Objectivism differently. One treats Objectivism as a complete value set, by which every aspect of their lives can be defined. I don't think they're fanatics, but, rather individuals who already lived in agreement with that value set. And on the other hand are those whose value system has a different set of values, who treat Objectivism as an approach to thinking rather than the definition of it. The latter definition is best expressed in Atlas Shrugged and We The Living; the former is best expressed in The Fountainhead and her philosophical essays. It's a distinction in definition. I am, incidentally, of the latter persuasion - I consider myself an Objectivist, but disagree with Ayn Rand on the matter of values. (I, for example, reverse her logic - life is important because it is necessary to reason, rather than the other way around.)

Comment author: Doug_S. 08 December 2007 02:07:03AM 1 point [-]

There's no particular reason to want to continue that metaphor more directly. Only mathematically precise metaphors have that kind of power.

Well, is it better to continue a vague analogy, or mix your metaphors and end up with something that reads oddly? ("Evaporative cooling" leading to an explosion? Isn't freezing the relevant phase transition here?)

I'm just nitpicking language here, so maybe I should stop hijacking the thread. ;)

Comment author: gwern 30 September 2009 05:00:36PM 1 point [-]

Well, is it better to continue a vague analogy, or mix your metaphors and end up with something that reads oddly? ("Evaporative cooling" leading to an explosion? Isn't freezing the relevant phase transition here?)

When I followed the link to the page, I watched the gif of a condensate cooling down; at a certain point, when enough of the 'red' (warmer) points had disappeared, the white column seemed to surge or 'explode' upwards.

Comment author: Strange7 17 February 2010 08:01:20PM 7 points [-]

Evaporative cooling could lead to a nuclear explosion. Imagine uraneous ore, dissolved in a fluid medium, surrounded by a potential barrier, and heated. Non-uranium particles escape first, followed by lighter isotopes. The end result would be a tight concentration of the heaviest available isotope: critical mass.

Comment author: Luke_A_Somers 18 October 2011 05:10:08PM 3 points [-]

As it approached critical, it would generate heat. There would be an equilibrium. In this case, the group itself becomes super stable - but its ideas become unhinged.

I really can't think of any comparable system, except condensation.

Comment author: JoshuaZ 18 October 2011 05:30:25PM 5 points [-]

An almost identical system actually happened in nature.

Comment author: Paul_Crowley2 08 December 2007 02:17:14AM 1 point [-]

So the point is that the idiots who are directly useless - make no useful contributions, have no ideas, spark nothing good - may be useful because they give shelter for others who want to raise controversial ideas?

I'd want to see a group not already mad that suffered for not having an idiot in their number before I believed it...

Comment author: Roland2 08 December 2007 02:20:41AM 3 points [-]

As a former member of a lot of different "christian"(the quotes are there for a reason) churches I can only say: you hit the nail on the head! Btw, I'm not a member of any church nowadays.

If you get rid of the old outliers, the group position will shift, and someone else will become the oddball. If you eject them too, you're well on the way to becoming a Bose-Einstein condensate and, er, exploding.

The real problem is that some group-leaders really want their groups to become BECs, with them at the center of control. So they will do all that is possible to eject dissenting voices. Alas, I speak from experience here.

Comment author: Ian_C. 08 December 2007 02:24:04AM 2 points [-]

Maybe those cultists were so brainwashed that they no longer judged their ideas by reality, but reality by their ideas? So in their eyes, strictly speaking, the non-arrival of the space fleet did not count as evidence either way. Scary thought, I know.

Comment author: TGGP4 08 December 2007 04:21:04AM 4 points [-]

Is suvine.com just being ironic after reading the stuff about trolling? If so, you really had me going there for a while.

Comment author: Caledonian2 08 December 2007 04:31:26AM 8 points [-]

I would make a rule: any discipline against a troll will be matched by identical discipline against anyone who engages that troll, even in attack.

In practice, such rules create echo chambers where any discussion that the moderators don't agree with is silenced. It's theoretically possible for a human being to be utterly impartial when deciding whether conversations are useful, but it requires near-superhuman patience and tolerance.

Quite a lot of the problem is that the category of 'troll' quickly expanded beyond its original meaning; in everyday Netspeak, it generally refers to "a person who persists in saying things I don't wish to tolerate".

Comment author: sboo 14 April 2014 06:58:23PM *  1 point [-]

well, it's tone too.

e.g. say sauvine.com had said: "this is why i think the scientists who believe in global warming have formed a BEC..."

i bet people would downvote, but i doubt they would label them as a troll.

Comment author: Stuart_Armstrong 08 December 2007 08:44:48AM 11 points [-]

If you have one person around who is the famous Guy Who Disagrees With Everything, anyone with a more reasonable, more moderate disagreement won't look like the sole nail sticking out.

Like the Devil's advocate in the catholic church? He seems to have done his role admirably; according to Wikipedia:

[The Devil's Advocate] was abolished by Pope John Paul II in 1983. This abolition streamlined the canonization process considerably, helping John Paul II to usher in an unprecedented number of elevations: nearly 500 individuals were canonized and over 1,300 were beatified during his tenure as Pope as compared to only 98 canonizations by all his 20th-century predecessors.

Comment author: Jerry_Jones 08 December 2007 02:20:19PM 18 points [-]

Actually, Jehovah's Witnesses and their predecessors have set one date after another since 1843, and are right now trying to figure out the next:


Comment author: Eliezer_Yudkowsky 28 November 2009 09:45:29PM 9 points [-]

Wow is that sad.

Comment author: Al_Fin 08 December 2007 02:57:59PM 1 point [-]

Most commenters refer to cultists as if they are not them. Wrong. Understanding how the brain works--how it clings to belief--does not place one outside his own brain.

It is our emotional core of thought that leads to irrational attachment to belief. If machine intelligence can precipitate around a different "thought model" it may escape most(?) of the irrationality of its forerunners.

Comment author: Nastunya 08 December 2007 08:07:46PM 1 point [-]

I think a better way to use Kuhn's idea to interpret the process and effects of the skeptics leaving is through an ontological reflection on the group's essence: We can say that the group or discipline really started existing in a relevant sense (living out its core beliefs, "getting any real work done") *only after* the original skeptics were gone. Perhaps the discipline's increasingly inhospitable environment to internal dissenters is a more relevant place to draw the line of exactly when that discipline "[abandoned] the requirement of outside accessibility" than the group's own idea of when it started existing. Put another way, the skeptics, while technically belonging to what I'm considering the not-yet-self-realized group *were* the outsiders whose access was, in a sense, curtailed by a later change in tone of discussion, etc


Comment author: Ray 08 December 2007 08:13:59PM 0 points [-]

There's no particular reason why you need to have a detailed knowledge of the psychological literature for fanatics to try to chase moderates out of a group. That particular piece of knowledge, that if you shout down and away anybody who disagrees with you then you are in control, seems built in to human understanding of politics.

Comment author: Patri_Friedman 08 December 2007 08:56:49PM 0 points [-]

Great post! I'd guess the effect is smaller than cognitive dissonance, but it definitely seems real. The tricky part, as you point out at the end, is how to apply it. But I don't think it's that hard to separate trolls from dissenters.

Comment author: Psy-Kosh 08 December 2007 11:41:29PM 6 points [-]

Hrm... I'm not sure having just "that one guy that's the Known Contrary Guy" would have the desired effect.

Maybe I'm completely wrong, but by personal expectation would be that having just one would instead be a convinient "bad example"... actually resulting in a reduction of dissent among the rest by having people partly wanting to, in a sense, avoid being like that guy. Having more than one may be better since then it wouldn't be "that one crazy guy that you don't want to be like" but "just some people that disagree."

Or am I completely and utterly wrong on the psychology of this?

Comment author: Nastunya 09 December 2007 04:05:36AM 3 points [-]

Psy-Kosh, having that some-people-that-disagree dynamic is clearly preferable to having that-one-crazy-guy-you-don't-want-to-be-like setup, but I'd expect that many such dynamics tend to start with having just one strong dissenter, don't you think?

Also, I can imagine another difficulty for that Known Contrary Guy: his input may unfortunately be less well received by others not just because of the clash of the views but also, and I fear even more so, because of ad-hominem-type dismissals by the rest of the group.

Comment author: TGGP4 09 December 2007 05:05:11AM 2 points [-]

Adirian: "...I consider myself an Objectivist, but disagree with Ayn Rand..."

Do you realize how ridiculous that sounds? Would that be more or less ridiculous that saying "I consider myself an Objectivist; I agree with Ayn Rand"?

Comment author: Ben_Jones 10 December 2007 12:31:07PM 2 points [-]

Let's make sure we're all discussing the same thing here: I'd say there's an appreciable difference between 'The Guy Who Disagrees With Everything' and your average troll. Note the word 'articulate' in the original post. A clued-up Devil's Advocate is an important way of keeping yourself honest. An annoying idiot (even just the one) is just a quick way of bringing a debate to its knees. Unfortunately, I fear there's no way to quickly discern which is which....

Comment author: ME3 10 December 2007 04:25:03PM 0 points [-]

The metaphor can be made mathematically precise if we first make the analogy between human decision-making and optimization methods like simulated annealing and genetic algorithms. These optimization methods look for a locally optimal solution, but add some sort of "noise" term to try to find a globally optimal solution. So if we suppose that someone who wants to stay in his own local minimum has a lower "noise" temperature than someone who is open-minded, then the metaphor starts to make sense on a much more profound level.

Comment author: Eliezer_Yudkowsky 10 December 2007 08:37:17PM 4 points [-]

ME, such an analogy is deeply inappropriate. Simulated annealing and genetic algorithms are rather nitwit, inefficient algorithms that AI programmers use because current programs can't abstract and compress and leap across the search space the way that humans do. Rationality is not about random noise and selection, or it would take humans millions of years to think anything significantly complex.

Comment author: ME3 10 December 2007 11:33:15PM 1 point [-]

Eliezer, you are right, what I really meant to say was, once a person finds a locally optimal solution using whatever algorithm, they then have a threshold for changing their mind, and it is that threshold that is similar to temperature.

Comment author: Ben_Jones 22 January 2008 09:22:52AM 4 points [-]

Just saw this.

Have a look at the first letter. Classic example of this post's subject. Newcastle had a good fanbase from all around England a few years ago, but "as soon as Keegan was gone [world not ending as predicted] so were the neutrals [skeptical followers] owing to the fact that they were never really accepted as 'real' Newcastle fans on account of they weren't Geordies. They don't have any 'outsider' fans to keep them calm and give them a sense of reality." I read that and had to stop myself saying "Evaporative Cooling of Group Beliefs!" out loud in the office....

I'd already thought of this myself before I saw the article. Not only is Evaporative Cooling in evidence at Newcastle United, they are caught in a Happy Death Spiral. For the philistines out there, Newcastle United Football Club recently reappointed Kevin Keegan as manager. Keegan was a star as a player for Newcastle thirty years ago, and had a spell as a manager about ten years back. They played brilliant attacking football under him, but Keegan's tactical naivete saw them throw away a commanding lead to lose the championship. He left, and it's been downhill ever since.

In the intervening years the team has struggled to show any real class, and rose-tinted hindsight means the fans still call Keegan 'The Messiah'. His recent second return has been greeted as if it were actually the second coming. He and the supporters have had a big old media love-in, reinforcing one another's sense of righteousness, and the club is now unnaturally confident about its prospects. When Keegan's faults are pointed out to fans, the stock reply is 'who cares, the Messiah's back, we're going to see some exciting football!'

Hope this makes sense in the states....

Comment author: srcs 10 August 2017 07:25:14AM 0 points [-]

You mean his tactical naivete caused them to throw away their lead. It didn't "see" anything.

Comment author: MichaelGR 11 October 2010 02:49:16AM *  0 points [-]

Typo/missing word alert:

*When an unusually fast atom escapes, it takes with an unusually large amount of kinetic energy, and the average energy decreases. *

Should be: "it takes with IT an unusually..."

Comment author: ciphergoth 20 May 2011 10:19:20PM 4 points [-]

Prophecy Fail: What happens to a doomsday cult when the world doesn't end? Vaughan Bell, Slate.com, May 20, 2011:

When Prophecy Fails has become a landmark in the history of psychology, but few realize that many other studies have looked at the same question: What happens to a small but dedicated group of people who wait in vain for the end of the world? Ironically, Festinger's own prediction—that a failed apocalypse leads to a redoubling of recruitment efforts—turned out to be false: Not one of these follow-ups found evidence to support his claim. The real story turns out to be far more complex.


In fact, so many studies have been conducted on unfulfilled prophecies from religions large and small that they were compiled into a fascinating book from 2000, Expecting Armageddon. None of the groups described reacted to the unexpected persistence of the world with a zealous drive for new members, and most made just minor adjustments to their beliefs.

Comment author: buybuydandavis 15 October 2011 12:14:12PM 8 points [-]

When the reasonable people start leaving, the center of mass of the group shifts to wack space.

As for the comments on moderation, the solutions described are overly constrained by the pitiful state of web forum software.

Back in the early 90s, I was on an email list with elaborate filtering mechanisms, including collaborative filtering. Usenet had the magic of killfile filtering. Today, we're lucky to get "thumbs up", "thumbs down", and "ignore". It's just pitiful. It's like grubbing around in the Dark Ages, after living in Imperial Rome.

Comment author: Odinn 06 March 2013 09:04:49PM 0 points [-]

This reminds me of a thought experiment where perfect averages skew towards one extreme when you eliminate one radical. It makes mathematical sense. Apparently, a village can come extremely close to guessing the weight of an ox by taking all of their guesses and averaging them, even if some indivuals are radically under or over. But change the scope and you may change the median's accuracy (or sanity, as the articles metaphor) Lock the village in a room with no clocks or windows and wait until 6 am, just before any hint of sunlight, then show them the sky and take their guesses. The radicals that guess 'midnight' won't change, but the ones who would have said 'noon' will, so your average would slide to ever more inaccurately early. Just a thought model though, I've never read this precise test being done.

Comment author: SeanMCoincon 31 July 2014 08:03:00PM 0 points [-]

"And what would be the analogy to collapsing to form a Bose-Einstein condensate?"

...All of them moving into the same compound and acquiring an arsenal seems about right, particularly when you consider the increased chance of violent explosion.