A few months back, scientists at the University of Alberta built a piece of software that could not lose at the game of checkers. Even if you played the game perfectly, at best you’d end up with a tie. Somehow I doubt that most online gamers are looking for games that they can’t beat, but it was still a pretty amazing technological feat to see accomplished.
After proving that checkers could be perfected, the team behind the software set their eyes on the high stakes world of Poker and in late July they entered their Polaris poker software into the Advancement of Artificial Intelligence Computer Poker Competition in Alberta, Canada.
The highlight of the event turned out to be a Polaris rematch against celebrity poker player Phil “the Unabomber” Laak. Laak previously had beaten the Polaris software, but not without quite a bit of difficulty.
For the rematch, the AAAI paired up Laak with fellow poker pro Ali Eslami and the two did battle against a number of different poker programs. The end result of the event proved that the pros can still beat the best software out there, although even the pros admit that the poker bots are getting better.
“And so it seemed a solid victory for team humanity. Was this more proof that the complexity of poker was still currently too much, even for a program that had been in the works for 16 years?
Not according to Eslami and Laak.
As the applause died down Eslami spoke to the crowd, “This was not a win for us. First of all there are a few things you need to know. One of the bots completely clobbered us. Another one had kind of a glitch in the second match that we won.”
Both players also agreed that they had played their absolute best poker and if there had been a time limit on the hands, they would not have been able to beat Polaris.”
As it turns out, the Polaris software wasn’t even the best bot to play in the tournaments. A piece of software called Bluffbot 2.0 couldn’t beat the pros, but was able to edge out the other robots in a tournament that pitted a number of different bots against each other.
The Bluffbot was built by
a couple of software developers Teppo Salonen in Claremont CA. The duo that created the software hasn’t released the Bluffbot 2.0 on their website, but they do have an earlier limit version, that is available for download. They also promise that they’ll have an online version of 2.0 up soon, so that internet surfers can test their own skills against the machine.
A lot of people find the idea of poker robots somewhat distasteful, but I’m fascinated by the technology. Trying to create the perfect chess or checker games is tough, but because there are only so many mathematical possibilities, it’s something that is at least possible.
When it comes to poker though, there are so many variables involved that I’m still not convinced it can be done. You can certainly analyze other players patterns for tells, but sometimes, it’s the little things that give away someone’s hand. I used to play online poker quite a bit, but after finding that I couldn’t win online, I stopped playing in the real money tournaments.
I’m not sure if people are gaming the system online or if I just can’t do well at poker in an online environment, but when I play live, it’s a very different game for me. Just being able to look for subtle tells like which card someone is eyeballing, can give you a real advantage in real life, but when you play online, there isn’t nearly as much info to go on.
Eventually, someone will come up with a piece of software that will be able to consistently beat real life players and when they do, it will make online gambling even less attractive. There have already been attempts to deploy this technology on real money sites, but there isn’t any good data on how effective these programs really are.
At the end of the day, I enjoy online games as much as the next person, but part of what makes internet gaming so appealing is that people actually make mistakes while playing it. It’s fun to be challenged, but it’s even more fun to win and when your opponent is a cold calculating machine, it takes random mistakes out of the equation.