Hacker News new | past | comments | ask | show | jobs | submit login
Symptoms of Groupthink (2012) (washington.edu)
125 points by aleyan 3 days ago | hide | past | web | favorite | 56 comments

The problem with most of these kinds of analyses of groupthink, is that everything they say is a characteristic of groupthink, is also characteristic of achieving consensus of any sort. We don't call it self-censorship unless it turns out badly, because the absolute lack of self-censorship is cacophony where nothing is ever decided. "Belief in inherent morality of the group", is not what we call it when it turns out that, say, those in favor of abolitionism _were_ in the right. We only call it that when it turns out to be false.

I've seen cases where you actually lack all of the things listed here, and they are groups which are never able to agree on a course of action, or cooperate to achieve anything. You call it either "agreement" or "groupthink" depending on whether it turns out to be good or not.

Which means, that a list of "symptoms" like this will not help you to discover in advance whether the group in question has a problem, which is what is actually needed. It's something that can give you the comforting illusion of being able to detect groupthink ("it won't happen to me"), without in fact helping you actually notice when the group you're part of is heading off in the wrong direction.

> the absolute lack of self-censorship is cacophony where nothing is ever decided

I don't believe this, and I've engaged in processes where this was not the case (albeit on a relatively small scale).

Requirements for effective consensus without self-censorship are:

1. a strict, ordered and budgeted process for airing views. Your cacophony only occurs without such process (such cacophony also leads to a different form of censorship, simply by "survival of the loudest").

2. a deliberative decision making process: the best way to alienate (and encourage self-censorship in) someone is to assign finality to decisions they oppose. It's also the best way to delay decisions. Controversial decisions can be expedited if there isn't perceived to be a barrier to reconsidering them later.

Or to put it a different way: people who don't agree with a direction an org takes are more likely to remain committed to that org if their later "I told you so"s are taken on board.

This is an important distinction. Consensus does not mean agreement. Not everyone has to think the idea is the best or even good, they just need to all accept it as viable.

It seems that "group-think" is more about agreement than consensus. An insistence on agreement silences opposing views. A search for consensus entertains them all but recognizes the need for a group decision to be reached.

> An insistence on agreement silences opposing views. A search for consensus entertains them all but recognizes the need for a group decision to be reached.

Well put. I think a lot of the stigma around the word consensus comes from undertaking it without this understanding. A.K.A. design-by-committee, etc.

>s that everything they say is a characteristic of groupthink, is also characteristic of achieving consensus of any sort.

ALL of them? How about:

- Outgroup stereotypes? - Self-appointed mind guards to insulate leadership from challenging ideas? - Illusion of unanimity?

I don't see how any of those are necessary for consensus. Same for self-censorship really. One can voice disagreement with a course of action and then agree to go down it anyway to be a team player. People do this all the time.

"One can voice disagreement with a course of action and then agree to go down it anyway to be a team player. People do this all the time..."

Yes, and at the point you stop repeating your disagreement, you have begun to self-censor. People who say everything they think, are described as "having no filter". The difference between "self-censoring" and "having no filter", is whether you think it's a good idea or not. Groups where those opposed to an idea do not, at some point, stop saying so, do not cooperate well.

Which is not to say that it isn't often a bad thing that they stop saying so. However, knowing whether it counts as "self-censoring", requires knowledge of whether their criticisms are well-founded, rather than providing that knowledge.

>Yes, and at the point you stop repeating your disagreement, you have begun to self-censor.

Absolutely not and even making this comment seems indicative of really unhealthy communication dynamics in a team. Censorship is specifically suppression of statements due to concerns about their political consequences.

Simply not being a curmudgeon who keeps griping about decision long after it has been made. That's just called being a productive member of a team. The point of team communication is to accomplish a task up to standard, not to ejaculate personal commentary about whatever happens. The article is specifically talking about censorship of perspectives that have a material impact of outcomes during decision points.

> Yes, and at the point you stop repeating your disagreement, you have begun to self-censor.

At what point is choosing not to repeat yourself not self-censorship? If there is no such point, I'd say the term defined this way is meaningless.

"- Self-appointed mind guards to insulate leadership from challenging ideas?"

this one I can see in a consensus environment. At some point it's not useful when people keep questioning the decided upon course of action and it may be necessary to stop listening to them.

That's not really about griping after something has been decided. This was about obfuscating the actual risks or concerns.

Criticisms of groupthink and other anti-patterns are funny to me because they presume that if people only had access to these valuable insights they would change. Organizational anti-patterns are rarely broken, they work for someone, and someone's salary depends on sustaining that dissonance.

The work in large organizations is to achieve outcomes within the broken and inconsistent rules of their internal logic, so as to preserve the integrity of them, and not the outcomes themselves. The essence of being a geek, nerd, or technologist is to focus on the outcomes and problem solving to achieve them, whereas the essence of being typical is to be a working part of the system to sustain it, independent of it's outcomes.

Warren Buffett said, "I try to invest in businesses that are so wonderful that an idiot can run them. Because sooner or later, one will."

To me, the point of a startup is to defect from these dissonance machines and create something people actually want. Viewed this way, a startup is not a business. It's like an anti-business that gets absorbed by a business and releases stored potential revenue in the collision.

Groupthink might be more of a sign than a problem. It's possible that it is a sign of the growth stage of a business and whether it's ready to absorb something, or just be hollowed out in some long term private equity cost optimization hell.

Antipattern criticism seems more as a warning to others who aren't invested in failure as often the practitioners are too far gone in preserving their own egos over changing their course.

The "self-sustance" of the system is insanity akin to chopping down walls for firewood, never building new ones (while inefficient that would at least be a way of wear cycling) and then once they are left on a dirt lot bemoaning it was unforeseeable yet inevitable.

It's easy to be cynical but I believe some corporate cultures are far healthier than others. I believe lots of companies try to avoid the HiPPO effect (deferring to highest paid person's opinion) with varying success.

I worked for a few years at Shell though and now I'm wary of ever working for a large corporation again. The sweet spot I've found is mid-sized companies, where they're small enough where they can't just absorb the costs of useless software projects, but big enough where there isn't the mad scramble to ship. (I've never worked for a startup, though, so apologies if that's an unfair characterization.)

> I worked for a few years at Shell though and now I'm wary of ever working for a large corporation again

I didn't find them that bad tbh, I think that there's good parts and bad parts in all monsters like Shell - my take was that their size, heritage and criticality to the global economy has resulted in them developing certain bureaucratic traits akin to a governmental civil service, but even with that said they do some amazing stuff (which most civil services cannot get anywhere near) - and I worked with probably the smartest IT pros I've met in my career. They operate a massive global IT infrastructure at scale and they do it pretty well, with some aspects being indisputably world-leading.

Oh yeah, definitely. Shell is not a office but a zillion offices spread throughout the globe.

Probably my main problem was I worked for the data management department where the most experienced people were semi-technical and had geoscience backgrounds. Intelligent people, but internal software tools were still very new to them and many were still doing things they did in the 80s.

One of my favourite Shell memories is when Outlook malfunctioned and sent an email to thousands of people in error, many of whom then also replied saying 'me too' to an initial reply-all email that said 'I got this email in error'. It happened several times until the CIO was forced to send a company wide email saying 'Please don't reply all when you receive an email in error, it takes several days processing for Outlook to catch up'.

I know groupthink has a negative connotation and can be damaging but on the other hand it seems necessary to have some kind of groupthink in order to align an organization on a goal. Especially in areas where there is not a clear cut decision what’s best you have to form a group opinion that assumes certain things. For example in software dev at some point you have to decide on an architecture and live with it even if there are alternatives that are as good. People who constantly reopen that question are not really helpful at some point. The danger is probably for people to stick to that decision for too long even after weaknesses start showing up. But I think especially for a large organization it’s hard to find the point when it’s time to change course.

I guess all I want to say is that damaging groupthink is easy to spot in hindsight but that well functioning organizations also rely on groupthink. And that it’s quite hard to distinguish between positive and necessary groupthink and negative groupthink.

Consensus and common goals aren't groupthink.

Groupthink is specifically about when people stop thinking individually, and take the group's position to be correct-by-default. Consensus and common goals can be achieved without that assumption.

Groupthink, in software architecture, happens when we can no longer question the basis for a design or assumption. Perhaps the design made perfect sense once, but now we're trying to scale to 1000+ servers, not just 5-10. Communication overhead is killing us, groupthink promotes (for too long, ultimately to the detriment of the business and its customers) the idea that the architecture is beyond question. Consensus and common goals are different. If the architecture is limiting the business, then that can be an agreed upon idea, but not actionable yet. Maybe we realize that other things need to be addressed first (our build system, CI/CD pipelines, testing systems) in order to make architecture updates even feasible.

Groupthink would deny even addressing those concerns because they still won't accept that the architecture is limiting the business.

I believe that is cohesion which isn't inherently linked to group think but attempts to promote may lead to it.

To use a cliche combat context it is like loyalty effectively. There may be questions of whether to attack from the woods or an open field with tank support and working together is more important than one or another as even the situational "wrong" choice is better than a split effort.

However absolute blind loyalty ends only in needless tears from trying to shoot tanks with pistols, charging the no man's land into machine guns, massacre of villages, or military coups.

The dangerous aspect is the "delusion" in both.

Military is a bad example.

I can tell you from experience, it's not a place that runs on "groupthink", or "consensus" views. A decision is made by a decision maker, and that decision is implemented. In the military they even have the "mind guards" that "insulate leadership from criticism" that were mentioned in another comment. Only in the military, we call them "NCO's".

You can be diametrically opposed to a decision in the military if you like. Heck, you can think whatever you want in the military, because no one cares what you think.

Are you sure that’s groupthink that you’re describing?

Groupthink isn’t mere agreement, but specifically agreement contrary to the professional (not personal) options of informed experts.

“It’s not my first choice, but it will do” isn’t, I think, in that category.

I think often agreement gets relabeled as groupthink later.

I think it is more "obviously wrong in face of evidence and refuse to reconsider" than informed experts per se. Doctors being stuck up about handwashing because they heuristically interpretted "a gentleman's hands are alway clean" as a descriptive instead of a directive was certainly groupthink.

It is possible to be wrong and not be engaged in groupthink or right and lacking logical basis.

What about when all the informed experts are wrong?

Like consuming dietary cholesterol causing high blood cholesterol?

Or trans fats being healthier than saturated fats?

Or tariffs harming the US economy?

How can your group know something that informed experts don't? If you have special knowledge they don't, then they aren't informed. Your group might knowingly take a risk in guessing the experts are wrong, perhaps to test a novel idea, but that's not the same as refusing to recognize their opinion as valid, like in groupthink.

"Experts" are especially prone to groupthink, this has been shown numerous times.

People can be "informed" about something and still need to shut-up about it, when it's politically incorrect / represents a unpopular view, when it puts their income / reputation at risk, etc.

Experts have more to lose, so they are especially vulnerable to this sort of pressure.

Of course experts can be wrong, but you won't know that they wrong or how in any particular instance, so you won't be justified in agreeing on something contrary and silencing dissenters who bring up the expert opinion.

I'm just talking about obvious reasonable behavior here. For instance, experts probably agree that nuclear radiation causes cancer. It would not be sensible to believe that nuclear radiation doesn't cause cancer and then expose people to that radiation, believing it's harmless just because experts are at risk of being corrupted.

It might be interesting to list some contraindications as well. What aspects of group dynamics work against groupthink, or are signs of its absence?

I suspect all of them could be view as anti-patterns. For example:

- frequent dissent

- group frequently breaks ranks when new facts emerge

- lack of confidence

- lack of a clearly-defined enemy

- lack of consistent message/frequent flip-flopping

It seems likely that some aspects of groupthink are essential for any collective action. Without it, people can't or won't work together. With too much, the group makes mistakes no individual would have made.

Groupthink happens because humans are tribal by nature. Tribes provide protection as long as you comply with their principles. Doing something contrarian is questioning those principles which is a threat to the tribe. Groupthink keeps you from being ostracized from the group as long as you go along with it because it’s desirable to be accepted in that tribe in order to achieve your long-term plans.

I have a feeling this will be true for most, but I immediately thought of the increasingly common political bubbles we often exist in today, especially given facebook feed algorithms, etc.

Does anyone else go crazy in the presence of groupthink?

I've never been unpopular and I'm the kind of guy to make friends with individuals in various groups, but I've always been allergic to cliques.

Whenever I find myself part of an in-group that's rapidly converging on groupthink, I get deeply uncomfortable and don't last long there. When I'm on the periphery of a rapidly coalescing clique, I often find myself becoming the outspoken outsider.

I just can't stand people agreeing on everything -- I've got to play the contrarian. Universal agreement just doesn't seem safe or comfortable.

I feel I have a natural resistance to groupthink and it's a huge liability. Most people seem to love forming into tribes.

And frankly, how do you stop the ever-present march of groupthink? I guess key is creating a culture where respectful disagreement is encouraged and expected, where people can openly admit to being wrong without losing face.

I understand the feeling - I often feel like the world is insane and you are the crazy one for pointing out that something makes no sense and dismissed without even a half-assed hypothesis backing it. Finding out there is a good counterintuitive reason for apparent irrationality is a relief.

I suspect the easiest fix on a personal level is ironically self-selection but certain things are known to exacerbate groupthink like the sense of a threat - annoyingly completely separate from /actual/ threats.

What drives me over the edge though is that developers believe they are so imminently intelligent. Don't get me wrong, many of them are.

But sometimes they believe so strongly in their own biases and preconceptions that it is deeply frustrating.

For instance, there are some who believe in the gospel of whiteboard interviews and when you try to discuss things like individual variation and how the process could be mostly random, their minds are completely closed.

Honestly, I don't encounter these sorts of problems nearly as often with people who have had liberal arts educations... but then again, I don't debate those sorts of people online like I do with programmers. Maybe it's just people are more confident in arguing their opinions when they're anonymous.

To be frank I have encountered the problem /more/ with Liberal Arts Educations who gain a misplaced confidence im absolute nonsense and treat using numbers to decide anything with contempt as less human. I suspect that it may be more the culture to blame at that point per se.

There are two types of intelligence: fluid intelligence (specialized domains of knowledge) and crystallized intelligence (general knowledge). Those developers you encountered were probably very good at what they do (fluid intelligence), but crystallized intelligence is just another separate category that operates based on a different set of thinking patterns.

Stereotyping and the tendency of labelling are often the enemies to establishing a culture like the one you described. It's a product of our egos and cultural history unfortunately, and we are all susceptible to this flaw from time to time.

I can't tell if we are getting better or worse on that front. The attention economy and social media might have accidentally turned people into various giant bubbles, making it increasingly more difficult to find common patterns to agree on. Everyone all has different opinions on thousands of tiny different things. People in the past did not face this problem, the society of yesterday was a much more simple place with fewer variables, so groupthink was probably a lot more common.

Diversity is the obvious answer to counter groupthink, but managing diversity effectively is in itself another challenge. This is a modern problem of today and will become even bigger tomorrow.

Other than alignment, a safe environment for everyone to operate in as you have suggested is absolutely fundamental. Human will always have the tendency to attach labels to everything, it's the natural way for our brain to create patterns and make sense of the world around us. Without those patterns, we can't operate effectively either.

In that sense, the only way to counter that is to anonymize informational sources during internal meetings and data collection processes so people can't attach labels and link it to specific sources - people don't like to admit it but we all do this unconsciously. That way the data will be free from prejudices and biases since all sources are being given equal weight in consideration.

Organizations should run pilot programs to test this on their decision-making process. It should be easy to achieve with many tools available today.

Applying the "grey zone" mental model is also important here, as it constantly forces us to consider the pros and cons and weight everything in probability calculation - there is not a clearly right or wrong position in many cases.

Another method is by encouraging each group member to seek out two to three other directly or secondary related topics by themselves (so there is a higher chance for a more random selection which should yield better accuracy), and then asking each person to make a summary on the areas they had chosen, and to further explain their idea on why they have established those critical connections with the main topic. This way, it may foster the discovery of new patterns based on arbitrary links that no other individuals had ever thought of before.

Not to put too fine a point on it, but this probably makes you part of a group. There are probably people who will comment agreeing with you is what I'm trying to say.

We're all, kind of, members of the People's Front of Judea. (Except, of course, for those of us who are members of the Judean People's Front.)

I commented on this before. Basically, most of us are rebellious in almost exactly the same manner as every other "rebellious" person. Same taglines. Same clothes. Same beliefs. Same pass times. Etc etc etc. We've all become the brooding teenager, who believes him or herself to be unique and counter cultural. Yet somehow that teen still looks, sounds and dresses like every other unique and counter cultural teenager on the planet.

As humans, we haven't really come up with a way to escape groupthink. That fact has got me wondering if groupthink is simply a facet not so much of human behavioral psychology, but rather a facet almost of evolutionary behavior? Something we can't tear ourselves away from, because it's literally who we are? We basically only choose which group we are part of, whether or not we're aware of it.

Your comment reminded me of one my favorite posters:


That's a false dichotomy, though. You can be part of a group and not succumb to toxic groupthink.

In many ways, groupthink can be enormously useful. In a sense, universities, books, and other knowledge dissemenation mediums exist to solidify useful groupthink. Maybe one guy comes up with some small piece of math, another guy on the other side of the world comes up with another piece, and everyone makes their various contributions until someone comes along and rolls it together into theory that advances the world.

People who don't think like the group are reformed or weeded out so everyone understands how to "properly" study the world and you don't end up with a bunch of cranks coming up with intellectually lazy speculations.

So that's good groupthink (but people don't call it groupthink, because groupthink is generally considered to be the negative version of that). The problem is when groupthink becomes detached from reality.

But back to your original point: no, if I'm part of a larger group and I start to disagree with them because I think they are wrong, that does not make me part of a group. It's just me. I could organize a group of dissenters and form a critical mass to change the group, but that's not typically what I do.

Furthermore, the argument that "everyone is unique like you" to criticize counter-cultural movements is attacking a strawman. Maybe unintelligent teenagers think they're radically unique. But counter-cultural movements are very much groups. Humans work best in groups. We think best in groups. No one can accomplish very much alone.

All the thoughts I have, and all the thoughts you have, are plucked from the ether of the prevailing beliefs of our times. I can explain to you my personal philosophy, and you could sit down and trace my intellectual lineage if you really wanted to. You could observe how I picked the belief I'm arguing for after reading Carl Sagan at 16. He got it reading some other guy who got it from Schopenhauer or something. Maybe some of my Baptist upbringing is sprinkled in too, which comes from the Bible, in a line of intellectual descent that can be traced back to Zoroastrianism and beyond.

In the meantime, there are a thousand other youngish programmer dudes who read the same books that I did, and we talk to each other on the internet and reach group consensus. We bounce ideas off of each other and bring each other in line -- just as we're doing to each other at this very moment, on this very forum.

People who think they're radically unique are wrong. We're social animals. But that doesn't mean having the courage to dissent isn't important.

>> At the hearing, engineer Brian Russell noted that NASA managers had shifted the moral rules under which they operated.

I would like to juxtapose it with the current sway of moral relativism - It is okay for you but not okay for others. Where does one draw the line between an absolute moral code and an opinion.

> Self-Censorship

this one is interesting.

Human's are lazy by what once was necessity, and that is still deeply rooted in our behavior.

Most folks in tech can grok that Laziness can be a virtue.

However, modern society has perverted the image this trait reflects to appear somehow inferior to "hard work."

We live in a society where we are raised to "endure" adulthood. It definitely doesn't have to be like this now. Whether it did in order to achieve the tech at hand is a lovely debate to be had... if only more people could hear the opportunity is literally pounding at the door begging to come in.

Wishful thinking anyway...`

The economist Timur Kuran has an interesting theory on this topic called “Preference falsification”. Under this theory people lie about their true opinion due to social pressure of various kinds. The interesting part is that this phenomenon can produce dramatic and unexpected changes when a widespread preference is no longer falsified.

After reading the article, I realized that if a site like HN or Reddit wanted to mitigate groupthink, they would want to limit how much an item can be upvoted or downvoted (e.g. max 5 like Slashdot did).

Apparently people really like groupthink, it increases engagement.

In my opinion, groupthink on social media is to be understood as a Keynesian beauty contest. [0]

Suppose you‘d want to determine who among a group of entrants in a beauty is the most beautiful according to standards shared by the group. Such determinations are hard to make, as every judge would usually vote according to their private norms of beauty, not the groups. One way to overcome this, would be to give a price to the two judges who are closest to the final result. In this setup, you‘d be incentivized to vote according to the norms you believe to be the overall group norm, not your individual norm. The result would be „the groups belief about what the group sees as beautiful“.

Social media such reddit and Facebook incorporates this concepts on two levels:

- At the user level, users aiming to gather likes/karma/... are incentivized to submit content they believe the group finds most useful. Of course this is why we instate voting systems in the first place, so that great content (according to the communities standard for such) is submitted. On the other hand, this incentive scheme drives unpopular opinions to the margins. Hence, fringe content is more likely to be found on sites without a like system such as image boards. There, a different incentive exists: survival of your post directly depends on engagement, so it‘s more likely to stay afloat if it’s provocative.

- Then, at the platform level, sites faced with the decision of what to show their users, also face a beauty contest: they select what is most likely to be liked. The perverse result is an eternal perpetuation of groupthink, as contributors facing a „likability selection“ are unfree in what they submit. If it’s deemed not controversial enough/not likable enough, they will wither away. So they keep perpetuating the ever same content and jokes.

[0] https://en.wikipedia.org/wiki/Keynesian_beauty_contest

Sites select content that is most likely to generate engagement they can sell to advertisers. Content that is liked turns out to be a poor strategy for that. Content that generates violent reactions of all kinds is more effective. (See also tabloid newspapers.)

Karma is a very poor proxy for post or poster value because it's cumulative.

Example: Stackoverflow has no way to distinguish between a poor post and a good post that's too recent to have acquired many upvotes. Likewise karma provides no useful distinction between someone who has been posting for years and someone who has just joined. (Also applies to Reddit and HN.)

In fact markets are perfect examples of Keynesian groupthink. The goal of most investment decisions isn't to assess intrinsic value or potential, but to second guess how everyone else in the market will estimate their ability to profit from collective average estimates of value/potential.

This leads to investment distortions where investors would rather chase potential unicorns - because of the explosive IPO potential - than invest in unexciting but solid businesses with stable but undramatic growth prospects.

There's no reward for up-voting a post that goes on to receive a lot more up-votes, and there's no penalty for up-voting a post that goes on to receive a lot of down-votes. Therefore the judging mechanism itself (the votes) is not subject to the Keynesian beauty contest problem.

That‘s true, it‘s not present at the voting level. But I believe it’s present at the level of submitting and feed selection.

I don’t think you can get rid of groupthink while still letting everyone vote on everyone else’s opinion. People are going to inevitably use the up and downvotes as agree/disagree buttons and boom, you have crowd-enforced “correct” opinions. If your goal is to curate quality discussion, don’t crowdsource that curation. All voting systems I’ve seen confuse popularity with quality.

> If your goal is to curate quality discussion, don’t crowdsource that curation.

But what other resource than the crowd is readily available to curate?

Non curated public discussions are absolute garbage, so I'll take a flawed curation over none every time.

What about adding separate "I agree/disagree with this" buttons? With minimal or no influence on the position/color of comments. Would people be able to say "I disagree with this comment but it makes a good point"?

I liked good ol Slashdot’s system, where if you vote, you need to select a reason with a drop down menu. This tiny bit of extra friction might be helpful to discourage quick “drive-by” downvoting and brigading. Additionally, their meta-moderation system helped to make sure people were really voting in good faith. I think they were really ahead of their time in some respects.

Also, give people a vote budget so they think about whether they really need to vote. Maybe 5 votes per day. For whatever reason I can only write 5 posts per day here, so I try to use them wisely. Same should go for votes.

HN won't show a value below a -4 point score on a particular post regardless of how many times it's downvoted (though one could infer based on how low the contrast of the text is within that range).

In any case, the number of points for a particular post is only visible to the author of that post, so even if a post has a large number of upvotes, it won't really stand out other than being close to the top of the main thread (or subthread).

There's still the trouble of having many maximally upvoted comments all sharing similar sentiments

How do you avoid it ?

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact