« The Hidden Power of Boobs!
Netsketch Art Fight »

Skynet Flops the Nut Straight

Monday, July 14th, 2008 at 12:10 pm by Jamie

Johnny Chan vs Terminator

You’d almost expect Skynet to be a ninja master at chess. Given enough computing power, a computer can calculate every possible move on a chess board and extrapolate a proper counter against a human opponent. In essence, chess is an information based game. But what about a game like poker that’s based more on intuition and skill?

AZM regular Greg sent over an alarming article about Polaris, a computer program developed by Michael Bowling of the Department of Computing Science at the University of Alberta that beat Texas Hold-em pros. This was Polaris’s second time at the big table. The first time it went up against human opponents last year, Polaris lost twice, tied once, and won once. This time around Polaris lost once, tied once, and won twice. Bowling believes that given enough practice and computing power, Polaris could play a perfect game of poker every time.

You know, if Bowling is ever hurting for research funds, all he has to do is fire up Polaris and hit those online Hold-em sites. What better way to give Polaris some practice and raise funds to continue development and research?

[Via ars technica]

Share This

This entry was posted on Monday, July 14th, 2008 at 12:10 pm and is filed under The Future, Skynet. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.

5 Responses to “Skynet Flops the Nut Straight”

  1. Avatargreg
    Posted: Jul 14th, 2008 at 1:23 pm
    1

    As an interesting side-note regarding chess. A lot of the world grandmasters have memorized so many of the possible board formations that an alternative chess format has developed in which the back-row pieces are randomly distributed instead of the standard R-N-B-Q-K-B-N-R distribution, thus increasing the potential board layouts and openings exponentially. I wonder if Deep Blue ever played that way.

  2. Pingback Pingback:
    Posted: Jul 14th, 2008 at 1:32 pm
    2
    Skynet Flops the Nut Straight

    […] angryzenmaster.com wrote an interesting post today onHere’s a quick excerptSkynet Flops the Nut Straight Monday, July 14th, 2008 at 12:10 pm by Jami You’d almost expect Skynet to be a ninja master at chess. Given enough computing power, a computer can calculate every possible move on a chess board and extrapolate a proper counter against a human opponent. In essence, chess is an information based game. But what about a game like poker that’s based more on intuition and skill? AZM regular Greg sent over an alarming article about Polaris, a computer program develope […]

  3. AvatarKevin Bahrt
    Posted: Jul 14th, 2008 at 3:06 pm
    3

    It depends on how much of it is card counting I guess. The vast majority of poker is limited by the fact that there are only 13#’s by 4 suits by however many decks you add so unless the deck is reshuffled every time Skynet could always count cards. It could also be programed for pattern recognition and a camera to watch for facial tells. If it watches the pupils (I think they do something during a bluff that can’t be controlled) then it could be learning how to tell if someone is lying.

  4. Avatargreg
    Posted: Jul 14th, 2008 at 4:15 pm
    4

    Kevin - As far as I know, most online poker sites, and organizations like the WSOP use decks that are fully shuffled before each hand. Basically the program calculates probabilities based on what it knows it has, and what other players may have. It’s currently limited to only 2 other hands, whereas a full texas hold’em table can sit 8, maybe 10, players which alters the odds for having cards come into play on the table. That is really just a question of scale of programming though. What’s interesting is that the program ‘learns’ what its opponents do over the course of playing against them. I’m curious as to whether the program is coded to bluff and as to how hard it is willing to push when it has a weaker hand. That is something that doesn’t seem like it could be very flexible in a program, though I can’t say for certain without seeing it play.

  5. AvatarWayne
    Posted: Jul 15th, 2008 at 8:55 am
    5

    Greg, I think you’re talking about Bobby Fisher’s random chess, I probably have the name wrong. Pawns are set in their normal place. White sets a piece in the back rank, black mirrors it. I’m not sure if black then sets a piece, that would seem to be the fair way to do it. The only restriction is that you can’t have both bishops on the same color. Play then proceeds normally.

    The one problem that I could see with that is that you would need to use algebraic notation instead of standard notation for keeping your sheet. (B2-B4 vs KBP to KBP4).

    I’ve worked a lot of tournaments, got Garry Kasparov’s and Anatoly Karpov’s autographs, along with a lot of great American players. Never met Fisher, glad I didn’t.

SexyComments by BorkWeb
Close
E-mail It