The world of online poker
'I am going to raise this one, see if this guy gives me a bit of respect," mutters Victor Wong as he shoves a pile of virtual chips into the middle of the virtual poker table.
Instantly his opponent folds and Wong collects his winnings, allowing himself a brief smile before the next hand is dealt and fresh decisions must be made.
Wong is the face of 21st century poker. A software programmer with a passion for maths and statistics, he is about as far as you can get from the grim-faced, cigar-chomping cardsharp of popular imagination. For up to 10 hours a week he plays online the most popular variant - Texas hold 'em - using a disciplined, aggressive system he says earns him an annual profit of a few thousand dollars.
Many of the world's poker servers are in the Mohawk territory of Kahnawake, an Indian reservation near Montreal, and overseen by the Kahnawake Gaming Commission, which is exempt from North American gambling laws.
Richard O'Neill, the founder of the National Poker League, which organises poker games in pubs and clubs, claims a membership of more than 200,000. "Our research shows that 50 per cent of them also play poker online for cash," O'Neill says.
However, unlike pubs, clubs or casinos, you are never quite sure who you are playing against online. Indeed, there is a fair chance your opponent is a poker "bot" - an automated software program designed to fleece unwary players. So far the bots are relatively unsophisticated. "They have a very simple algorithm," says Wong, who suspects he has come up against them from time to time.
It turns out creating the perfect poker-playing robot is surprisingly complex. It has long fascinated artificial intelligence experts, who are making big progress. Scientists have already cracked backgammon and draughts, producing a machine that can play the perfect game - the best a human opponent can hope for is a draw - and even chess has largely been solved.
Poker presents special challenges because it is a game of "imperfect information".
"In chess everything is visible - one can see the board, there is no information hidden except for the opponent's mind," says Ann Nicholson of Monash University's IT faculty, who with her colleagues has done a lot of work on poker.
"But in poker, depending on the version, some of the cards are hidden. There is uncertainty not just in the randomness of the cards in the deck but also in what people have, what actions they are going to take and what strategies they are going to use."
This uncertainty makes card counting - calculating the odds of cards to come, based on cards already dealt - a much less useful strategy than in games such as blackjack.
A true poker-playing bot must learn from an opponent's playing style and allow for the opponent doing the same thing. "I've got to remember that you [the opponent] are … thinking what my actions tell you about what I might have and I'm thinking about what your actions are telling me about what you might have," says Malcolm Ryan, an artificial intelligence and games researcher at the University of NSW.
"There are these whole cycles of thinking about thinking about thinking."
Nevertheless, scientists are coming closer. The world leaders are University of Alberta scientists in Edmonton, Canada. Their bot Polaris last year beat the world's best players at a tournament in Las Vegas. Even though it involved a simplified form of Texas hold 'em, it was a big breakthrough.
The Canadian group is working on a Polaris that would be unbeatable at the unlimited betting version commonly played online. This would pose a big problem for promoters of the lucrative online poker businesses. If punters lose confidence in the capacity of promoters to detect and ban bots, they will go elsewhere.
"It's like an arms race," Ann Nicholson says. "People are trying to get their bots in there and not be detected, and people are trying to detect them. And people don't usually advertise how well they are doing if they are beating the systems because they don't want to get caught."
Wong is confident with his half-man/half-machine strategy. His program allows him to instantly assess the strength of his cards against those of many players he is likely to come up against.
This information allows him to target weaker players - "fish" in online poker parlance - while staying away from the profitable "sharks". Some programs even track individual fish, alerting stronger players when they come online.
"That's the beauty of playing online," Wong says. "You've got so many tables starting every minute you don't have to play with good people."
Understanding mathematics, he says, has changed the whole game. "In the long-term you can't beat the mathematical edge."
Sydney Morning Herald