Two representatives of Stoxpoker, the online training site, will battle against Polaris, a poker-playing series of computer programs developed by researchers at the University of Alberta. The match will take place July 4-6 during the Gaming Life Expo in the Rio Hotel and Casino in Las Vegas.
One of the two human competitors will be Nick "Stoxtrader" Grudzien, the successful high-stakes online player and co-founder of Stoxpoker. The other has yet to be named, though will also come from the group of Stoxpoker pros. The match represents a chance for Polaris — and its creators, the University of Alberta's Computer Poker Research Group (CPRG) — to avenge its narrow loss last July to Phil Laak and Ali Eslami.
The competition will follow a similar format to that used for last year's event which took place in Vancouver at the annual meeting of the Association for the Advancement of Artificial Intelligence. There will be a series of 500-hand "duplicate" matches of fixed-limit hold'em, meaning in each match the two human competitors will simultaneously play 500 hands of LHE against Polaris, with the same series of cards being dealt in both contests; only the hole cards will be reversed. The total number of chips won or lost by each team will then be added together to determine the winner of that match. Following the "duplicate" poker format lessens the luck factor, making the match a more accurate measure of the relative poker-playing skill of the humans and that of Polaris.
"My goal will be to stay aggressive and to avoid tendencies that can be exploited," says Grudzien. Such a strategy will no doubt be necessary. Prior to last summer's competition, Jonathan Schaeffer, chair of the Computing Science department at the University of Alberta and head of the CPRG, explained that Polaris is has in fact been designed in such a way that it "learns, adapts, and exploits the weaknesses of any opponent."
As Schaeffer told me last summer when I spoke to him prior to the Laak-Eslami match, "If we lose, we lick our wounds and go back to the drawing board to try to develop new and better technology." At the time he stated he fully believed that "one of these days — within 5 to 10 years — two-person, limit hold'em will be solved." It will be interesting to see in July how far the Alberta group's AI research has progressed.