The Wisdom of Crowds (Paperback)

Are crowds wise? Surowiecki fills his pages with unconvincing anecdotes. He has only a handful of real studies and he buries them randomly throughout the book. Worse, Surowiecki sometimes describes a study that would be easy to conduct, but instead of doing it he simply tells us what he expects the results would be. And despite the book's constant championing of dissent, Surowiecki offers no evidence that cuts against his argument. Instead, every failure of a crowd simply helps prove his thesis, since he claims it failed because it violated one of his vaugely-stated rules.

When are crowds wise? Surowiecki offers only untested speculation. He claims they need "diversity, independence, and a particular kind of decentralization" (oddly, by decentralization Surowiecki appears to mean aggregation). Surowiecki never defines any of these particularly clearly but instead gives lots of examples. This makes them useless as predictors of a crowd's intelligence which is probably why Surowiecki makes no attempt to test them.

Why are crowds wise? Surowiecki doesn't even bother to answer this one, even though it's the first half of the books subtitle. He considers the question briefly on page 10, only to spout some empty sayings (crowds are "information minus error") and wonder in amazement ("who knew ... we can collectively make so much sense") before finally concluding "You could say it's as if we've been programmed to be collectively smart."

Perhaps noticing these weaknesses, Surowiecki gets all this out of the way in the first 40% of the book. The remainder is dedicated to larger collections of anecdotes Surowiecki likens to case studies. But even they disappoint. While Surowiecki has lots of stories, few are particularly enlightening or even memorable. Surowiecki does little analysis of the stories and does not draw out larger lessons. He assumes he is right and only stops to look down upon those who disagree.

I'm especially disappointed since I expected the book to be good. I love Surowiecki's weekly column in the _New Yorker_ and I suspect he is right about a lot. But instead of making a convincing argument, Surowiecki just stirs together anecdotes from his columns. The result, not suprisingly, is an intellectual muddle.

One thing the book does teach (although not clearly) is the wisdom of _dissent_. You can ensure dissent by collecting a large group and keeping the members from talking to each other (since people are usually smart but afraid of going against the grain), by ensuring some members of the group vocally disagree (since they will force the others to better justify their positions), or by forcing them to try to justify all sides (since that will keep them from prejudging the question).

All of which makes it ironic that Surowiecki's book fails because of a lack of dissent. Nothing goes against the grain, he doesn't justify his positions, and he has clearly prejudged the question. It would seem he needs a crowd to make him wise.

One of the things about the book that hasn't been much remarked on is the light it sheds on the flaws in the way the U.S. intelligence community -- and, I would argue, the Bush administration -- approaches the problem of forecasting the future and making good decisions. The book's main subject is the wisdom of crowds, but Surowiecki spends a lot of time on how groups go wrong, and his discussion of how groups make bad decisions seems to me completely relevant to our current problems. When Surowiecki delves into groupthink, into the pressure that's exerted on lower-level employees to conform, and the perils of too little diversity of opinion, he's making a broader point about what good decisions require. But in the process, he clarified for me just why the current administration did such a bad job of figuring out whether Iraq had weapons of mass destruction and of planning for the postwar period. I was surprised, but it turns out this book has a lot to say about the state we're in right now.