Some notes on information and policymaking.

I was going to post this on Facebook, but it got too long...

This post is to lay out some key points and ideas for my classmates, who apparently were disappointed by today's lecture.

First read Hayek's classic paper on tacit knowledge. This describes the essential informational problem in economics (and generally): information is dispersed and impossible to perfectly transfer.  That information needs to be aggregated into some form useful for decision-making.

This is an especially relevant problem for policy-makers, who (ideally) try to choose and design the best policy for the most people. Without good information, success in that isn't about skill but dumb luck.

The wisdom of crowds is the simplest method of aggregating information, relying on some statistical ideas (the most likely single value for a random variable is its mean, and that over- and under- estimates will tend to cancel each other out) and some cognitive psychology (lying is generally more cognitively burdensome than honesty, so people will tell the truth unless they have some other reason to lie).

The classic example is from a contest held in 1906 in rural England. Now apparently rural England is quite boring,  and there is a prize after all, so a crowd 800 people decided to try to guess the weight of a cow. Now along comes a statistician, Francis Galton, who decides that it would be fun to analyze the guesses of the crowd. He does this, and discovers that the mean of the crowds estimate (1197lb.) was almost exactly the measured weight of the cow(1198lb.) and even more interestingly, was closer than an individual person's guess.

The contest aggregated the diverse information of the crowd into information more accurate than what any single person possessed. This is one of bases of the argument that "diversity trumps ability" (there is some academic argument around that, but that it is not immediately obvious is itself somewhat startling).

The cow-contest also illustrates the problems of this method. Almost immediately people begin trying to cheat: sabotage, bribery of the owner of the cow, and all manner of rural shenanigans, since a person can better improve their odds by attacking the structure of the contest, instead of, say, practicing their cow-weight-guessing skills at home.

The cow-contest also illustrates the problem of groupthink very clearly. In the original contest, participants wrote down their guesses of the cow's weight and submitted them privately - each person using their private information. In other contests, where peoples guesses were not private, the wisdom of crowds failed. Why? People were communicating with each other, and while a lie is cognitively costly, social discord is even more costly - thus agreement is more important than honesty. I am sure we have all experienced the social pressures of the white lie.

I could explain this by references to human evolution (better to go along with the tribe in a bad decision than strike out on your own), to social dynamics (people like other people who agree with them, even about trivial things), and of course, to sabotage and intimidation (one way to rig the contest is to circulate false information, or to threaten to break the legs of everyone who doesn't guess 0).

Now, in general, markets also perform this function - especially when talk is not cheap, and agents are sophisticated or engaged in frequent and repeated transactions. This is the idea behind decision markets.

Informational problems are also fundamental to political systems. One way to think about democracy (and about the importance of secret ballots) is as a way of aggregating diverse tacit information.Even in authoritarian systems, like China, the Communist Party uses secret ballots and elections for Party and local elections.

I hope this has been fun, informative, or useful.