What makes group decisions go wrong and right | Hacker News Reader

What makes group decisions go wrong and right

nautil.us

46 pointsdnetesn21 comments

> results—theoretical at this point, tested mathematically and with computer simulations

I was about to say "these results are extremely unlikely to be replicable" until I saw that phrase: there are no results. She had some models. This sort of thing is hard to devise an experiment for.

I was once in a very complicated experiment designed to "prove" that group decision making was better : your group was at a house in the woods and you were trapped in an Australian bushfire. You had to pick 10 items as being important to your survival, both on your own and collectively as part of a group. Then those results were compared to the "experts" ranking; the experts being Australian fire fighters with tons of experience in this.

  (Side note: I think of this every time there really IS an Australian bushfire.)

The results were pretty unconvincing; almost everyone, individually and collectively, got it wrong. Everyone thought "car keys" and "cell phone" were the most important. WRONG! Your car will get trapped in a traffic jam trying to get away and you'll burn, and the cell towers will be overloaded and your phone won't work. If you're curious, the most important was "long pants" and "long sleeved shirt" because exposed skin will burn most readily.

Afterwards, our boss said "I was really pleased that our group really came together and arrived at the results collaboratively."

I said "Yeah, but we all died!"

I find the "wisdom of crowds" vs. groupthink thing really interesting. Wisdom of crowds is definitely a thing but under fairly narrow conditions that I've never seen rigorously defined although I think people have some sense for it.

When it works it tends to work really well (e.g. http://bitmason.blogspot.com/2012/04/crowdsourcing-predictio...) but it's not very generalizable.

ADDED: To your example, I'm guessing there weren't any real experts in the group (or they were very outnumbered). So there was no "wisdom" of diverse experiences bringing something to the table.

> I'm guessing there weren't any real experts in the group

You're right. It was overwhelmingly SV types.

Have you read "The Wisdom of Crowds" by James Surowiecki? The book is focused on just the question of what conditions seem to be beneficial/detrimental to sourcing the "wisdom" from crowds.

I have, not sure about @ghaff but I suspect he has, too.

This "researcher" is attempting to go deeper and quantify it, maybe (I suspect) as a way of devaluing individual thinking.

I have although, as I recall, while the book touches on some of the conditions for the wisdom of crowds to work I didn't leave with a strong sense of a definitive list.

However, as noted upthread some of the things would seem to be independent solutions, an ability to meaningfully aggregate those solutions, and some level of expertise in the crowd (i.e. if everyone's just making completely random guesses it won't work).

Traffic jam? In the woods? Further: I could see how cell towers would be overloaded in the past, but these days everyone are internet connected 24/7, so there must be enough capacity for everyone.

There's a really simple test - try texting or calling somebody as clock strikes midnight on new year's.

Does cellular internet go down at midnight?

If we're talking about a really rural area (which we are, probably), then their cell capacity might be sized by "what if ALL our subscribers texted at once?"

So if we assume that's N, it still might not handle 10N.

Usually there are only one or two exit roads, so it's easy to get blocked (especially with emergency services coming in, falling trees, people leaving at the last minute because they ignored the repeated warnings etc.). If I recall correctly, there were quite a number of people who died in their vehicles in the https://en.wikipedia.org/wiki/Black_Saturday_bushfires (see also https://en.wikipedia.org/wiki/2019%E2%80%9320_Australian_bus... -> https://en.wikipedia.org/wiki/File:Bells_Line_of_Road_Bushfi... the road in that photo is one of two main roads through the mountains, when that's blocked, traffic all over the place is affected).

As for internet, assume it's going to go down quickly, if you had any initially (I'd be surprised if a battery powered radio tuned to the ABC wasn't on the list).

What you said.

It doesn't take much traffic to overload one 2-lane road.

I am pretty dubious about "GroupThink" as something distinct from "politics" and "social pressure".

The canonical example is Bay Of Pigs where apparently no-one in Kennedys cabinet thought "what if the invasion fails".

I bet loads of people thought it. In fact they all thought it. It's just that no one at the decision making table was secure enough to raise the problem.

(Well that is the definition of group think. it is also the definition of failed politics, or social / peer pressure and failure to create psychological safe spaces, to have adequate feedback mechanism, etc etc)

It's important - but having a one of the cabinet raise issues is not solving the whole political decision making structure.

Democracy is still needed

> It's just that no one at the decision making table was secure enough to raise the problem.

'The five dysfunctions of a team' is a good book about this topic. You need trust to have an efficient team. Collaboration does not work when people does not feel safe.

Probably didn't apply to Bay of Pigs but something that's somewhat distinct is a desire for consensus.

It's somewhat a cultural thing (where culture can mean country/company/team). But there can be situations where someone (and it may actually be multiple people) have misgivings about an apparent consensus but feel that unanimity is more important than having a knock-down argument over some minor point. This isn't even always the wrong call--again doesn't apply to Bay of Pigs.

Hypothesis to which I have given 30 seconds' thought: the problem with social media, is that the person who announces a provisional conclusion (like this one, I suppose) has now announced it to a large group, in a print medium that is hard to back away from later. It takes more mental effort to change your opinion on a widely disseminated, print statement than it does to just not repeat that verbal opinion you said to a couple people.

So, if the problem is that "if an individual learner never changes their mind based on new evidence, their chosen option could spread throughout the social network", then discussing in small groups in a format which leaves no record afterwards (at least until later stages of the decision) might help to avoid this problem. Maybe?

Santa fe university has some interesting papers on groupthink as a complexity problem, and their podcast 'complexity' deals with it in several episodes.

Mirta Galesic does research in social decisionmaking, and if you're interested i would recommend the Complexity podcast episode with her https://open.spotify.com/episode/7m3lEMqnkEDgZcKesiepcn?si=A...

Small correction in case others wanted to go down this rabbit hole (which I’d recommend!): It’s Santa Fe Institute, not a university

A nice book I liked is "The Wisdom of Crowds" by James Surowiecki.

From what I remember, the gist is you want people to be coming up with solutions to a problem independently, and then you want them to voice their proposals independently (so as not to be stymied by others' ideas). And then you want to aggregate these solutions/proposals. The beneficial dynamic breaks down if people around the table share their ideas in order one after another.

https://www.amazon.com/Wisdom-Crowds-James-Surowiecki/dp/038...

IMO, it's familiarity. As soon as the unknown is introduced group decision becomes more and more useless.

> social learners seem to be more interested in group harmony than in making an informed judgement.

> It is not the social learners who are the problem here: Individual learners reluctant to change their minds are.

Or maybe it is the social learners, who don't care about being informed before making a decision?

Later they just spin it around to say as much.

Wasn't there a story on HN today about how the false-dichotomy (binary choice) isn't constructive? Oh, they hit on that too. eyeroll Feels like a pointless exercise.

Reply on news.ycombinator.com