The Wisdom of Crowds

 Posted by (Visited 15550 times)  Reading
Nov 012005

I just finished The Wisdom of Crowds. Thanks to Jamie Fristrom for turning me on to it.

Those of us who have been watching groups of MMO players already knew, of course, that the aggregated intelligence of large groups can be awesome and frightening. Those who have watched the burgeoning genre of Alternate Reality Games know that the players in those games can solve deeply complex problems cooperatively.

But that’s not the sort of problem-solving that this book is discussing.

In those other cases, it’s the presence of a wide array of specialists that generally makes the difference. In Surowiecki’s world, it’s the presence of a wide array of people, period, that allows startlingly accurate assessments to come forth.

Ask one person to guess how many jelly beans in a jar, and they’ll prove fairly unskilled–even if they are a jelly bean expert (the most notable quality of the jelly bean expert will be that they will overestimate their accuracy).

Take a crowd of random people, ask them all to guess, then average the results, and somehow you’ll get an answer that is remarkably accurate. Like, within a couple percentage points.

In other words, SirBruce should quit trying to get developers to leak numbers to him; he’d probably get more accurate results by just asking everyone who comes to the site to provide an answer, then averaging the results.

The problems with this sort of approach, of course, are that people influence each other. When monolithic blocks appear within the group, you’ll start to get inaccuracies. When apparently authoritative sources of information start broadcasting their impressions of reality, it’ll distort the result. The results in markets are bubbles and crashes. The result, perhaps, in democracies, is ideological partisanship.

Based on my reading of the book, the sorts of problems that this collective wisdom solves are complex but limited. A huge amount of the examples in the book involve examples exactly like the jelly beans–quantitative assessment. The bulk of them can be divided into

  • coordination problems, which are essentially forms of self-organizing behavior, many of which seem to echo the power-law observations in books such as The Tipping Point or my personal favorite explanation of small worlds phenomena, Albert-Laszlo Barabasi’s Linked. (Got email from him today, I need to reply… hmm.< ---gratuitous name-drop.)
  • cooperation problems which largely echo discussions about tit-for-tat strategies, trust networks, etc.

One thing where I was left hanging was whether or not this wisdom of crowds phenomenon can assist with creative problems. After all, the repeated creative story is very different from what is described in this book; it usually involves intersecting disciplines or schools of thought, and then someone who essentially integrates things. Breakthrough products are often created by teams that are very much infected by groupthink–what in this book is decried is described as an effective way to create a skunkworks team in Peopleware.

All in all, this is a valuable read that may well illuminate many aspects of decisions you see taken around you, the nature of leadership, and so on, but I’d definitely want to put it in context with other worldviews before letting it take on too much significance in your daily life. These days are exciting, because it seems like every few months or years, we get another good, solid description of a chunk of the elephant; but as usual, the way the world works is proving to be quite like the pachyderm of the fable: hard to describe except in isolate pieces.

  9 Responses to “The Wisdom of Crowds”

  1. Blogroll Joel on Software Raph Koster Sunny Walker Thoughts for Now Sex, Lies and Advertising

  2. Plaguelands [HTML] [XML] [FULL] PlayNC [HTML] [XML] [FULL] PlayOn [HTML] [XML] [FULL] Psychochild’s Blog [HTML] [XML] [FULL] Random Battle [HTML] [XML] [FULL] Raph Koster [HTML][XML][FULL] Saavedra’s Blog [HTML] [XML] [FULL] Se7en Samurai [HTML] [XML] [FULL] Sifo’s Blog [HTML] [XML] [FULL] slashrandom [HTML] [XML] [FULL] [HTML] [XML] [FULL] Stargate Worlds News

  3. I haven’t read Peopleware, but I certainly understand where one would get the impression that group creative works flourish in an environment infected by groupthink. There are two reasons I see for this:

    1.) Groupthink enables a sort of shorthand between people that allows them to skip past explanations, preambles, and disclaimers that might otherwise slow down the free flow of ideas.

    2.) Creativity leaves people very exposed and vulnerable. People feel very close to their creativity. It can be very hurtful to have one’s creative ideas and works openly criticized. This is exactly why you’re not permitted to criticize ideas during brainstorming sessions. It causes people to shut down. The creative flow is smoothest when there is a strong bond of trust between all the parties involved. Improv acting troupes do trust-building exercises, for exactly this reason. If you have a good, trusting relationship with someone, you know that person isn’t going to stab you with brutal cynicism, right when you’re at your creative peak.

  4. I’d have to dig out Peopleware to verify my memory of it (which is quite old) but there was a fascinating section on skunkworks teams which basically described them as insular, cut off from the larger organization, passionately devoted to what they were making, often feeling like underdogs or unappreciated, highly intra-communicative, and so on.

    These are the exact characteristics that the book gives as poor circumstances for creating “wisdom of crowds,” where you want team diversity, a lot of independence in thought, differing agendas… you get the idea.

  5. Oh, yeah, I can see the underdog motif working very well. An underdog is going to be more inclined to take chances and shoot the moon than someone who feels comfortable and secure.

    I wonder if this means we should make our R&D departments feel neglected and unloved? šŸ˜‰

    I think, as you point out, only certain sorts of of problems that can be solved with “the wisdom of crowds” (just as only certain sorts of problems can be solved with neural networks or genetic algorithms). Crowds are, as an aggregate, entirely incapable of lateral thinking. In fact, I imagine that they will produce the exact opposite of lateral thinking, if there is such a thing. It is this very characteristic which makes it so very good at solving, for example, the jellybean problem.

    Even given the observations in Peopleware, however, we should be cautious before jumping to the conclusion that diversity is not beneficial for creativity. There are certain creative leaps that every monoculture may be inherently incapable of. I think it’s possible to take a group of people from diverse backgrounds, and turn them into the sort of teams that Demarco and Lister describe. It takes all of a single day to do the sort of team-building required to get an improv comedy performance out of a group of complete strangers. Shared challenges go a long way towards this. I’ve run high-intensity urban scavenger hunt races with complete strangers before, and people were literally inviting each other to their birthday parties, by the end of the day. I think humans WANT to trust each other. They want to be part of a team. They want that sort of bonding. In the right circumstances, it doesn’t matter what side of the tracks we came from, or even what country we came from.

  6. I actually think that diversity is essential to idea generation. But there’s a long road from idea generation to a shipping product, in any industry. And particularly in team-based endeavors, there does seem to be a certain level of passion and focus required, and a “pull together for the team” sort of vibe. So it’s not quite groupthink, but perhaps it’s selective groupthink.

    I’d certainly want to have that diverse team, but then I’d want them to bond into a single entity that has many shared assumptions.

    I agree that humans want to trust each other. Have you read much into stuff like iterative tit-for-tat and so on? I also highly recommend “Influence” by Robert Cialdini for great insight into the many ways in which humans trust each other.

  7. Have you read much into stuff like iterative tit-for-tat and so on?

    Yes — from a game theory perspective. Though, I’d be interested in learning more about them from an evolutionary psychology perspective. Hmm, that book sounds interesting. I’ll tuck it on my wishlist, so I can find it again later.

  8. […] Darniaq is referencing, in part, this earlier post of mine in which I discuss The Wisdom of Crowds by James Surowiecki. […]

  9. […] I’m discovering a great deal of effort by people to try and classify or qualify the wisdom of crowds in some way, explaining it using a particular single discipline. Complexity theory, collective intelligence (CI), game theory, network theory, organsational behaviourĀ (plus here)Ā and psychology are just a few of the ‘boxes’ people are attempting to put the concept in and close with packing tape. […]

  10. […] Running through the exchange are challenges about the supremacy of each idea. This isn’t surprising, and is echoed across the blogosphere with people pointing out the failings of the wisdom of crowds concept in various areas – can’t cope with lateral thinking, might have trouble with self-interest, to point to a few. […]

Sorry, the comment form is closed at this time.