Playing your cards right: poker comes out of the back room and into the computer science lab

Science News, July 18, 1998 by Ivars Peterson

After several rounds of successive improvements in probability calculations, betting and bluffing strategies, and opponent modeling, Loki can play reasonably well.

At the moment, "the limiting factor is Loki's unsophisticated betting strategy," Billings says. For example, when the program holds a strong hand, it always makes a bet instead of occasionally holding back. Good opponents can take advantage of such a pattern.

"It's absolutely amazing how strong players can pick up on any kind of pattern in your play and exploit it," Schaeffer says. "Whatever your decision-making process, you've got to make random choices sometimes."

"The next major thrust in the research is to have Loki weigh the relative merits of different betting plans and to randomize those actions in a manner that conceals information from the opposition," Billings remarks.

Loki has already faced a variety of opponents on the Internet, and it wins consistently. The problem is that these human players have just pride, rather than money, at stake.

"It's very difficult to evaluate the program," Schaeffer says. "In chess and checkers, you can go to a tournament, pay the entry fee, and play to see how good the program is. Poker is different because it's meaningless without money." If a computer plays for money, however, who's going to pay for the program's losses?

Further improvements may make it worthwhile for Schaeffer to bring Loki to Las Vegas. "I would love to say that a year from now, we'll be sitting down with some of the best players in the world, but that remains to be seen," Schaeffer says. Many technical problems have yet to be solved.

Moreover, the sorts of expenses associated with a trip to Las Vegas aren't normally covered by a research grant. "We may need to find someone to bankroll our venture," Schaeffer says.

Meanwhile, at the age of 36, Billings makes a fairly comfortable living on the Edmonton poker circuit--though he has to cope with large fluctuations in income from week to week.

"The modern game of poker is a game of small percentages," Billings says. "It's not like you're hauling off wheelbarrows of money all the time. It's an accumulation over time. You just earn more than you lose--maybe winning in seven sessions, breaking even in one, and losing in two for every 10 that you play."

The lifestyle associated with the game isn't particularly appealing to Billings. "I'm considering abandoning it," he says. He knows that if he were to play in Las Vegas casinos, for example, where the financial rewards can be much greater, he would have to improve his game considerably without a guarantee of an increase in income.

Billings is thinking about writing a book, particularly on how theory can fall short in practice. "I've always been very cognizant of the limitations of the theory and where the theory breaks down," he says. "That has come into much clearer light through the process of mastering the game in real life.

"The real essence of the game is knowing your opponent and adjusting accordingly," he adds.