So you believe that experts in terrorism and hurricanes are motivated only by the drive for accuracy, and that the organizations for which they work reward accuracy over, say, adherence to a corrupt, morally indefensible political ideology or, I don't know, a complete disregard for the economically disadvataged?

Man, I wasn't surprised that Ezra was all down on the markets, but you, Tom? As a commenter on his post points out, the futures markets got every individual Senate race correct. One more time. Every single Senate race. Correct. But there's little to no information that might tell you something about ALL Senate races that you won't have built in to the individual races, leaving traders to multiply probabilities to determine an appropriate Dems Take Senate price.

A desire to make money will mean that people with more confidence in their opinions (say, people that are expert in a particular field or particularly well informed) will bet bigger and more often. If you have no clue what kind of hurricane season might be about to take place, you probably won't wager a great deal of money on the outcome, or if you do, you'll hedge. Not so for someone with a lot of meteorological training.

The other thing, though, is that markets are blind and anonymous. Experts have lots of reasons to believe lots of things and to speak about their opinions in varying ways, not all of which are conducive to determining what actually is most likely to happen. This was the idea behind the internal Pentagon terrorism market. The internal workings of such an organization certainly mean that lots of people are holding on to or dismissing information based on political or careerist calculations. An internal, anonymous predictive market would allow people and incentivize them to register their beliefs without any organizational repercussions. Result? Better information.

There's no need to run down markets. A lot of folks hate computers for a lot of reasons. They break and do things wrong and erase your work and allow people to send you obnoxious forwards. But computers are, essentially, the same thing as markets: they're super-efficient ways to process lots of information. So yeah, arguments might be more interesting and fun and romantic than futures markets, just like a Victorian office building filled with hundreds of guys with slide rules might be more romantic than a PC with Excel. But so what?

Also, you coming out next Thursday? I'll buy you a beer.

Felix Salmon, who's much smarter about the markets than I am, has a slightly different take.

And I think he's right - the Dems did get lucky. Montana, Missouri, and Virginia (sad to say) were all very close races, and could have gone either way - had just one of those gone 1% the other way, the Repubs would be running the Senate.

A market is not going to be an infallible predictor of the future. But, it will be the best aggregator of available data out there.

Oh c'mon, Ryan. You should totally expect this from me. Ezra knows way more about this stuff and is generally smarter than I am. You've been patient about educating me in the past, but by now you should realize that I have a not-so-proud track record of lacking the imagination to really believe in various economic concepts (I blame much of this on how econ 101 is taught, at least at UVA: the glossings-over and appeals to magic run throughout the course).

I hadn't considered the institutional advantages that markets might offer. As you and Emily point out, they may help to provide a forum where opinion can be related objectively, without fear of reprisal from within the institution. And as Mike points out, I can believe that they're very efficient aggregators of information.

Here's what I don't buy: that they can produce new knowledge. I can get behind them as a means of filtering opinions. But when it comes to, say, that hurricane futures market, I don't understand how the mechanism is supposed to work.

Is the market supposed to produce an answer that's better than the one produced by the best methodology employed within it? Or is it supposed to just filter between methodologies? If the latter, fine. If the former, I don't understand how the magic works. How can systems like this can produce original knowledge? Maybe this is the wrong way to look at it, but it's how it strikes me.

Anyway, your closing "so what?" is a fair response. But I do think there's something to be said for trying to focus on the model that's being used to make the prediction itself, rather than just the prediction. Behavior in markets is so complex that the results are context-free. Markets may be good for finding answers to specific questions, but they don't offer the explanatory power that a survey would, I don't think. That Felix Salmon link is interesting because it shows that a different type of polling *can* produce very good results. And by collecting data on the participants, you'd also be able to learn things about the problem at hand that a TS market couldn't tell you.

Markets can provide new information in the sense that they can tell the expert how much faith he ought to have in his opinions. They can't come up with a better hurricane model, but they can come up with better probabilities for different hurricane models, giving a better overall forecast.

So markets in no way replace the production of better knowledge or models or ideas, and to get the answer you want you have to pose the question properly and structure trading properly. As far as context goes, well, markets are best at providing certain kinds of information, and other things are better at providing other kinds. If market snapshots don't provide the explanation you seek, then I'd blame the journalists and not the markets.