Each summer, the computer-science researchers behind the world’s best poker-playing robots bring their creations together for a tournament. In recent years, three competitors have dominated: First is a group from the University of Alberta, which currently has about a dozen people working on poker programs. Next is a team from Carnegie Mellon University and their champion bot “Tartanian.” And finally, there is Eric Jackson, an independent researcher who has created a program named “Slumbot.”
The tournament consists of several different competitions, with teams tailoring their bots’ personalities to each one. Some competitions are knockouts, meaning two bots go head-to-head in each round, and the one with the smallest pile of chips at the end gets eliminated. To win these competitions, bots need strong survival instincts. They need to win only enough to get through to the next round—greed, as it were, is not good. In other matches, however, the winning bot is the one that gathers the most cash overall. In these competitions, bots need to squeeze as much as they can out of their opponents; to do so, they need to go on the offensive and find ways to take advantage of the others.
Most of the bots in the tournament have spent years in development, training over millions if not billions of games. Yet there are no big prize pots waiting for the winners. The creators might get bragging rights, but they won’t leave with Vegas-sized rewards. So what’s the lure of working on these programs?