“Wisdom of Crowds” by James Surowiecki


I am on page 5 and have seen nothing but anecdotes. Anecdotes have their place but I fear that Surowiecki thinks he is making his point with them. On page 5 he speaks of a group being 94% right. What the hell does that mean? When he notes that the average of a set of estimates is better than most of the estimates he is merely quoting a mathematical tautology, not an empirical observation. You can conclude that before you lay your hands on the numbers. Few of the anecdotes have distinguished between means and medians.

The Thiokol-Challenger anecdote is interesting and perhaps relevant but that is an example of the “crowd” being self selected ‘experts’ with a vested interest in the outcome. Just because you are not an insider in the sense of the SEC doesn’t mean that you don’t know understand the gross outlines of the failure trees. This is collective wisdom in Hanson’s original sense of where you loose money when you are wrong. Too much of that and you get out of the game and leave behind those with better sense. I buy that sort of collective wisdom.

On page 17 he finally gets to the Iowa Electronic Markets. I consider markets like this to be a democracy of money, not people. Money speaks here and people are in the background. That is why they work.

I have seen no inkling of members of the crowd with incentives to lie about what they think.

So far he leaves the process out of focus. It is like reading a biology book that fails to mention the difference between plants and animals.

To his credit he enumerates a good list of sources of counter anecdotes.

On page 20 he broaches the power of monetary incentives but notes that these incentives are only token. He fails to contrast these to markets in commodities futures where the incentives are more than token. These markets improve our economy. It is a pity that we have not found our way around the laws that limit the Iowa markets to market amounts.

On page 30 Surowiecki quotes Scott Page as recommending against uniformly smart people for that provides too little diversity. Without disagreeing I would say that perhaps the problem here is that ‘smartness filters’ are too narrow, not in the sense of letting too few thru, but in the sense of a single concept of smart. I like the emphases on diversity and polycentrism, especially many agents with their own money to invest. There is some talk of Darwinian evolution here, which I approve of. You get the best from diversity if under continuing play the losers remove themselves from the game. Surowiecki seems always to miss this point.