Recently, the London Times reported on the increasing prevalence of “bots” in online poker.  These bots are, in essence, computer programs that some players run, which play for the player in a preprogrammed manner.  While against the rules of almost every poker site, the article claims that many players have managed to avoid detection of their bots, and are earning up to £50,000 per year using such programs.

Bot programs work by making use of game theory principles to make their decisions.  Unlike a game such as chess, poker is a game of incomplete information, namely the cards that an opponent holds and the cards to come on the flop, turn and river.  While the cards to come in the future can be broken down through basic probability, a poker player must use logic and deductive reasoning to try and determine an opponents holding.

Poker bots offer several advantages: they can use their near-unlimited memory to recall past hands an opponent has played, can play continuously without suffering fatigue, and allow the user to run them while doing something else at the same time.  However they are slow to adapt to changes in an opponent’s strategy, and can not account for external psychological factors such as tilt (the tendency for an opponent to play badly while upset).In 2007, a group of researchers from the University of Alberta pitted their poker bot, Polaris, against professional poker player Phil Laak, who edged out a victory.  In 2009, the revamped Polaris was able to beat both Laak and two other poker pros.  Despite their success at games such as Heads-Up Limit Hold’em (a game with fixed bet sizes played between only two players), “Modern programs are not strong at No Limit or multi-player games,” according to Dr. Darse Billings, one of the developers of Polaris.  This is due to the multiple additional variables that exist in such games (bet-sizing in No Limit and additional opponents in multi-player games).

Thus, while some online poker players fear that poker bots will make the games impossible to beat, Billings, claims that, “I started the research into poker AI in 1992, and we have had a very large team of excellent researchers working on the problem for many years. We have made a lot of progress over the past 16 years, but I can assure you, the sky is not falling.”

Sources:

http://www.timesonline.co.uk/tol/news/uk/article5822078.ece

http://www.pokernewsdaily.com/poker-bots-not-a-real-danger-to-online-poker-252/