The Raw Prawn

Online gambling, particularly poker, has been an very successful business, seeing tremendous growth:

Gambling over the internet has become big business in the United States over the past year despite the recession that caused an industry-wide advertising slump.

Internet gambling, once thought of in the same seedy vein as online pornography, had gained a level of legitimacy despite myriad controversies facing it, according to a BBC report.

It said the sector has become the fifth largest advertiser online, jumping to US$2.5 billion from 911 million advertisements in the last year.

But could the party soon be over?

MSNBC Online has an interesting article on the rise of "poker bots" -- computer programs that can play poker, or at least advise human players -- and what it means for the online poker industry. The worry is that these bots will become good enough to ruin the game, driving out human players who can't compete.

Why not? Poker can't be much harder than chess, right? Well, it's actually much more difficult to create a good rules-based poker strategy. The most important difference is that you don't have all of the information in poker. You know your cards, and the few cards that are face-up on the table, but that's it. The best you can do is calculate the probability that someone could have a combination of cards that would beat yours, and combine that with inferences about the strength of someone's hand based on their behavior.

Betting is also a problem. It's much easier to write a bot to play limit poker, because the only choices a player can make are check, fold, call, and raise $x (where there is only one choice for x). In the more popular variant of Texas Hold'em, no limit, you can raise any amount, until your chips run out. There are now many many more options for a player, making it much more difficult to calculate the optimal bet and anticipate the reactions of other players.

There's also the bluffing element. You'd want a bot to bluff, and to be able to tell when someone else is bluffing. There's no bluffing in chess... :)

The bottom line is that effective poker bots require very sophisticated artificial intelligence capabilities.

Despite all of these hurdles, there have been some successes:

The University of Alberta's Computer Poker Research Group has developed an artificially intelligent automaton known as Vex Bot", capable of playing poker at the master level, though as yet it can only apply its gambling genius to two-player games. Vex Bot has been used by researchers to test the frontiers of artificial intelligence -- and as the basis for a commercial poker tutorial program, Poki's Poker Academy -- but some fear it may become a blueprint for programmers with more sinister motives.

Darse Billings, lead designer of the Vex Bot said he believes the odds are better than 50-50 that other programmers have secretly unleashed bots on commercial poker sites, apart from the commercial bots. But he throws his chips in with the skeptics, saying it is unlikely they would be anywhere near as capable as the Vex Bot -- so named for its ability to frustrate human opponents -- which is the result of more than a decade's research by the University of Alberta team.

Here is a limited-use demo of Poki's Poker Academy. I tried it out, and learned a few things. There were a number of marginal hands that I would have kept, but Poki rightly advised to fold (which suggests that I'm probably a good person to play poker against :P). As far as I can tell it only allows you to play against the computer, and does not allow you to link up to any poker sites and have it play for you. BadgeHelp's Texas Hold 'Em Assistant, on the other hand, will connect to an online poker site and play for you. They also have an online demo, though the website is a bit dodgy... use the product at your own risk. I know I wouldn't be willing to give a random program the ability to throw away hundreds of my hard-earned dollars.

Of course, that's precisely why poker bots won't get a widespread following. You have to really trust the program to be willing to use it. But the danger isn't that everyone will use bots, just that enough people will use them.

Slashdot has a great discussion about the technical aspect of creating these bots, and just how feasible it is to create good ones. You'll have to sift through lots of chest-thumping by people who say they can create the perfect bot, and others who say they can detect any bot.

There is an interesting post at Freedom to Tinker about the different tactics online Backgammon players have been using to detect people who are using computer assistance. They boil down to seeing how closely someone's plays match what a given program would suggest, taking note if someone takes a long time before every play, and only playing no money/little money games.

What does this mean for the online casino business? The highly cynical have suggested that online casinos don't really mind bots. Why? The casinos make their money off of the "rake" -- a certain amount out of every pot that the casino keeps. All the casino cares about is how many people are playing, and how much they're betting. It doesn't matter the players are man or machine -- their money is still green. If you have a decent number of players who are very confident in their bots, and they're playing multiple tables at once all day long, the casinos can make a killing off of them.

The real danger is that human players will be driven away, of course. No one wants to play a game that's not fair. And it only takes one rotten apple to spoil the bunch. If a handful of sites get overrun by bots, it will likely tarnish the entire industry's reputation. Even those casinos that succeed in keeping bots out will have trouble attracting new players, because players will be paranoid, and prefer to just play poker with friends or in casinos, or play other forms of online gambling.

Given the fact that tightly-run casinos might not see much a benefit from keeping bots out, should they even try (aside from that pesky "ethics" thing)? Why not enjoy the bounty while you can? If casinos will be able to detect bots (which is fairly doubtful), I think that you may see 2 or 3 sites that get a reputation for being air-tight operations, which gamblers will gravitate towards. The others will milk the bots and players for all they can, raking in the cash before the party eventually stops.

I suspect that human players would only be driven away from the high stakes tables, however. People will be more willing to lose small amounts of money to machines. Bots would also have more trouble succeeding with those types of gamblers -- they won't act as "rationally", and even if the program can read them, the potential winnings are low. The high stakes tables will then be bots against other bots against highly skilled players who think they can out-wit the bots, and if they succeed they could make a killing.

The question is though... if people are scared off from online poker, how many will then turn to other forms of online gambling where you just play against the casino, like blackjack? That's the question that will make or break online casinos.

See Also
The Washington Monthly - The always excellent Justin Peters on why poker appeals to the internet generation
Vice Squad - for more coverage of vices and the public policy questions surrounding them