[skates]

I'm with DS and pokervintage here, although I don't like removing tournaments from poker. The point I think that pokervintage was trying to bring up here is that poker is a game of incomplete information, where chess is a game of complete information. With this in mind, there are, hypothetically, equilibrium strategies for poker, and strategies that exploit those who are not following said strategies. MoP is a great book for understanding this idea, as it presents an actual *solution* (equilibrium) to high-blind heads up play with explanations on how to exploit a player who calls too tight or raises too loose.

The point is, we use Bayesian inference to take known information and extrapolate probable playing styles, ranges, etc. within a probabilistic distribution. It is that aspect that humans have not been able to solidify well into a program (though the UofA HU Limit bot did incorporate learning technology), and even if we could solidify it into a program, we would have major computabiliity problems just as chess does.

Greater point: Chess may not be solved yet, but from a philosophical stance, there is always always always a correct play. In poker, if we had exact information of everyone's ranges and playing styles, there would also be a correct play, but we are unable to know precisely what those ranges are at any given time for a variety of reasons. As such, we make our best guesses.

A president needs to be able to work with incomplete information and come up with political solutions based on perceived "ranges" and "styles of play" of those involved. Poker is a better test.

Sidenotes: I do not believe that the skillset used for poker is different from the skillset used for chess. I believe that both can be perfectly rationalized and treated mathematically. That said, it's a big jump for someone to think about a pro-gambler as a successful mathematician, so I won't try to make that argument. Also, for the sake of answering hitch1978's question about feasibility... 30k hand duplicate matches ftw, playing both sides by randomizing the order each time (go ahead, tell me that you'll know which A79r board you're looking at after 30k hands when you have the "unknown" cards). There are, of course, still problems with this, and you'd have to get some intense statisticians to come up with the actual sample size necessary to produce significant results.

For anyone in the tl;dr crowd:
Poker is a better test, much more relevant application of logic than chess.