The last few months I've been coming across a lot of interviews with Bryce Paradis. Bryce is mostly known for winning millions of dollars while playing poker on the internet. He is also an instructor at one of the internet's leading poker tutorial websites. More recently his name has been coming up because of the consulting that he's been doing with the Computer Poker Research Group at the University of Alberta.

The Group's goal was to create an artificial intelligence software program that was capable of beating the best poker players in the world. The latest version of this poker bot is known as Polaris. The group's AI work is pretty complicated, so I'm going to avoid talking about algorithms and things of that nature in this blog.  If you are interested in reading about the technical aspects of Polaris, you can visit the CPRG's website through the link below. There are also links there to the academic papers that have been published about the project. University of Alberta Computer Poker Research Group.

So why is Polaris becoming a hot topic all of the sudden? At a recent gaming expo there was a series of 6 separate 1 vs. 1 matches between Polaris and the world's best limit holdem players. Each of the matches lasted for 500 hands. During each game, the human player would effectively "switch seats" with the bot half-way through the match. They were then dealt the identical cards that the bot was dealt during the first portion of the game. The other relevant cards such as the community cards were also dealt identically for each hand as it was played the second time. This was done to try and eliminate any variance that may have resulted from a streak of bad cards.

Here's the results from the recent set of matches. Polaris-3  Humans-2  and 1 tie.

In order to qualify as a win, the player or the bot must finish the match with a minimum gain of at least 25 small bets. This was meant to prevent a small difference in chips from qualifying as a win or loss...mostly for the sake of the player's ego.

So what's the big deal about this? There is a lot of money being exchanged at the online poker tables. There has been some concern that Polaris might fall into the hands of someone who would want to use it to cheat. These concerns are not without warrant. The security teams at some of the online poker rooms have discovered bots in the past.
 
However, there are many reasons that it's unlikely that anyone will ever be using this software to beat the online poker games. The biggest reason is that Polaris is an extremely complex software program. It's taken a whole team of people several years to develop the software and learn how to apply it.

It's unlikely that there is anyone out there who would be interested in using Polaris to cheat, that also possesses the knowledge and resources they would need to implement the software.

If you would like to get an idea of how complex some of this stuff is, you can go to the CPRG's website and read Darse Billings' dissertation on Algorithms and Assessment in Computer Poker.

It's hard for me to give Polaris it's due diligence in this blog because of my lack of computer science knowledge. If you are interested in learning more about the Polaris poker bot or the recent series of heads up matches, you can hear a recent interview with Bryce Paradis through this link; http://pokercast.twoplustwo.com/pokercast.rss. Or visit the University of Alberta Computer Poker Research Group.


I'll leave this blog with a Mike Matusow quote; "As bad as people play nowadays, if you have to cheat to win at online poker you might as well just quit".