James Surowiecki’s 2004 book The Wisdom of Crowds offers a somewhat contrarian view of what is typically seen as a widespread bias: the human tendency to follow the crowd and go along with what the majority says is true. Surowiecki argues that in many cases, this is actually a reasonable thing to do, as crowds and groups are often much smarter and more accurate than even their smartest members.
Well you won't fall behind if you follow the crowd. "Runaways" substitute short term reputability for longer term status, erroneously assuming the crowd will appreciate the value of insight.
The Wisdom of Crowds Paradox
Donn Parker is a security professional who wrote an article for ISSA Magazine a few months ago that asserted that risk management should be replaced by due diligence, compliance, and enablement (whatever that is). Of course, ignoring risk is simply one...
Bruce, I think the reason the "beans in a jar" example is supposed to work is that people do not have a systematic bias in any direction. I don't know if that requires an explanation, but it would seem to me that biases are what demand them. People would often be in error, but their errors are random and centered around the true value. Averaging everything together causes these random errors to cancel each other out.
I haven't read the book, so it's possible that you already knew that and were asking something different.
I found the Wisdom of Crowds a very stimulating book which raised a very important question: how does the WoC effect actually work, ie. what are the mechanisms? However the book made no attempt to address this obvious question, which I thought was strange.
On the other hand, I have since utterly failed come up with any plausible answer myself. Any ideas? How about the 'beans in a jar' example (assuming the evidence is correct): how is it that group average guesses are accurate?
Considering the opposite side of the issue, I highly recommend Cass Sunstein's "Why Societies Need Dissent". This book isn't nearly as entertaining of a read as Surowiecki's - it's pretty much just a series of lectures strung together as chapters. Throughout, Sunstein presents and dissects the pros and cons of many group behaviors. Filled with lots of examples, and plenty of research citations. Check it out.
Digg's Fatal Flaw - Its Users
Digg this, Digg that - crowdsourcing is the ultimate way to unearth what's newsworthy. Or at least it would be - if its user-base actually did some real 'digging' and checked the facts behind a story (or at least the...
Interesting. So a useful question might be "does the accuracy of a crowd's 'crowdsense' correlate with the way that its individuals perceive and analyse collective opinion?" Does Surowiecki discuss that? If you're trying to predict the accuracy with which a particular crowd will answer certain questions, and you want to find (and eventually tweak) the simplest possible parameters that will enhance that accuracy, then measuring things like the weight that individuals give to the collective opinion seems like a good bet.
An anecdotal look at the extreme cases would seem to support it... Accurate: people rave about the Deplhi Method, and that's an expert-oriented prediction market, structured in a way that very much suppresses runaway positive feedback. Inaccurate: bubble markets, where the only significant market force is the credibility of collective opinion, but the collective opinion still fails to predict when the market will collapse.
I'm glad Surowiecki wrote his book, in part because it called attention to these issues. My main complaint is that in many readers minds his book framed the choice of prediction markets as a choice of crowd wisdom over expert wisdom. In fact, prediction markets can and do work well in situations with lots of expert wisdom and little crowd wisdom. By choosing a prediction market you choose a mechanism which should do well regardless of who knows more; the people who know much should self-select to participate, and there is little harm (and often benefit) from allowing those who know little to participate.