Slashdot Mirror
Artificial Intelligence in Poker
Markian Hlynka writes "The University of Alberta's research into Poker AI is featured in this New York Times article. There is also detailed discussion of the game of Poker, and the 'new breed' of players who have honed their abilities online. See the U of A's poker project for more information."
You can never learn the true art of poker unless you can read the other players tells, and unless the computer also throws in virtual signs I doubt cyber players could ever climb the ranks.
How do you bluff a computer?
Now, he is able to play every night between 10 p.m. and 3 a.m. while his daughter sleeps in the next room.
If this was UT or Quake, this entire article would be about how he was destroying his life, and getting ready to go on a rampage.
But instead, its just a game of cards, and he's gambling with his family's money, but thats OK.
If I have been able to see further than others, it is because I bought a pair of binoculars.
Poker is not a card game, it's a people game (aka don't play the cards, play the people). It's all about bluffing and reading other people's bluffs. I'm baffled that people even bother playing poker on the internet
Except that people play a certain way and develop easily (especially for an AI) recognizable patterns. Those patterns are just as recognizable, perhaps even more so online where the number of hands played per hour is so much greater.
Players can get broken loosely (particularly for Texas Holdem, but also for other games) into a small group of profiles and their play patterned according to that.
While an academic study may be new, commercial software to do this has been available for years. In particular Turbo Texas Holdem from Wilson software does an outstanding job of simulating different types of players and play conditions
and if you really believe that people skills and not card or math skills are all that you need, I'd invite you to come to Atlantic City and sit in any of the games I regularly attend. We'd love to have you.
If privacy had a tombstone it would read "We did it for your own good" . -- John Twelve Hawks
You're partly right. No-limit hold'em is about things like bluffing. The kinds of poker most people play in public cardrooms (mostly limit hold'em and stud) are more about the cards, although understanding your opponents is critical to doing well at either one. Either way, online poker has a lot to offer. It's still about understanding your opponents, you just don't get as much information by way of physical behavior. There are tons of bad players online -- players who make fundamental mistakes (i.e., don't play their cards well) and who don't adjust to their opponents well (both in failing to take proper advantage and in playing easily exploited strategies). It's true that the game isn't the same at all. But many of the differences are positive, and like live poker, it's still a game of skill (a fact that's well obscured by short-term variance).
(Blatant plug: I'm a little biased, the new edition of my book ("Serious Poker," an introduction to the serious game) has a chapter on online poker. But I do believe online poker has a lot to offer, and the sites do offer poker for play money as well.)
I didn't realize just how bad the problem was..
> But still, in lower-limit games, people are loose enough that bluffing doesn't really help
I recently heard a serious poker player on the radio explaining why it's worth bluffing sometimes.
If you don't bluff and lose sometimes, then when you _do_ have a good hand, you won't win much with it. You need your opponents to think "he could be just bluffing again, it's worth raising".
He was playing in high-stakes games though.
rant
I'm a big fan of fair use. I hate the DMCA. But behavior like this just makes me wonder. The free registration at the NYT is not that much of a pain. Sheesh. The newspaper world is being very cool, at least compared to the music and movie business. Let them make a few bucks on the ads so they can pay me.
You're just much more likely to need the best hand to win in low limit games. Bluffing may sometimes work (especially against better, or at least tighter, players) and is probably useful to at least occasionaly mix up your style.
Read Caro's Book of Tells for a good introduction to how to read people. You'll notice that some of the most powerfull tells give you information about how that player plays the game (tight, loose, passive, aggressive, etc.). A computer could get this information by keeping tabs on what its opponents do, and crunching the numbers. The only information it would lack is the tells people give off to inform you of what they are holding right now.
The edge you get from being able to read these "what do I hold right now" tells moderately well is a very slim one. It is an edge, and it will let you win money in games where you would otherwise lose. But knowing the odds, and reading what your opponents have based on the way they bet, and based on what hands you have seen them play in the past, are the fundamentals of the game. A good AI could master these fundamentals, and could probably challenge some of the better poker players in the world.
Note that I said the edge is slim if you can read these tells moderately well. Some of the people I see playing in the WSOP seem to be brilliant at it. I'm not qualified to talk about what kind of an edge it gives them, so I won't.
>he house, being a corporation, has an obligation to maximize its profits in any way possible
And when the last corporation uses the last gram of uranium to power the machine that sucks up the last drop of oil which they use to cut down the last tree on the planet to turn into paper money which they use to bribe the last honest politician, it will be a great comfort to us all that they are only doing it because capitalism obligates them to.
If you were blocking sigs, you wouldn't have to read this.
This reminds me of strategies for programming Ro-Sham-Bo (Rock, Scissors, Paper). The "safest" strategy would be to randomly choose rock, scissors, or paper every time. Your winning percentage would approach 50%, but so would your opponent's. Ah, but if you're competing against other pairs of players, and they're all following that strategy, then it's just dumb luck who will win. For there to be any point to the competition, you have to assume your opponent has some non-random strategy, such that you could beat it and get >50% wins if you could figure out what it was. Of course, your opponent is making the same assumptions about you. And so begins a world of strategies on how to make your opponent think you're being predictable, when you're really just fooling him into making a choice you can predict. Of course, if your opponent knows you're fooling him, he will then know you're intention and gain the advantage. And so on and so on (similarities to the Iocaine Powder sequence from the Princess Bride are more than coincidence).
I just wanted to point that out as a counter to the posts advocating a purely statistical approach in which the program folds anything not likely to win. In the optimal case (there is no house rake, no ante, and no bluffing) it is as interesting as flipping a coin to see who wins. And even a small amount of bluffing will cause it to lose.
Bluffing the attributes is part of the game. The true masters of poker can read a bluff. That's where the game becomes interesting.
Well, there are a few things to consider.
1) It's a violation of the rules to do this on any of the online poker sites. If you're busted, you'll lose ALL of the money you've paid in. (buy in, registration, etc.) They might even press charges.
2) Perfect poker strategy is only perfect in a vacuum. When playing against real players, you can make more money by playing imperfectly. In fact, if you always play perfectly, the game will change to conpensate for you, and leave you playing poorly.
"People who do stupid things with hazardous materials often die." -- Jim Davidson on alt.folklore.urban
It's about low stakes Texas Hold'em. As I'm preaching to people who don't follow links, I'll explain that in hold'em, your hand is drawn from the best five out of your two personal ("pocket") cards, plus five common ("board") cards that everyone can see and use. You can even just play the five common cards if they're better than that 8 in your pocket. You tend to get strong hands, but then again so does everyone else. Hold'em is generally played with big tables, so chances are that someone has a strong hand each round. You don't get extended rounds of raising, and there are no huge wins to be made. Coming out on top of a night of hold'em involves long term risk management, not a single guts-or-glory Hollywood dramatic climax.
As for bluffing, go ahead and try. There are only four rounds of betting on each hand. Experienced players will fold early, so you won't get much of their money anyway, and excitable noobs will tend to stick it out and call you out with their regular full houses and flushes, making it expensive for you to try to bluff. You'll quickly find yourself playing to your hand, not to the other players, and you won't (indeed, can't) get yourself into a steely eyes, car-keys-in-the-pot ego clash.
I wish, I wish, oh how I wish people wouldn't predicate their discussions based on what they've learned from Mel Gibson movies.
If you were blocking sigs, you wouldn't have to read this.
You're playing against other people. If you did play "perfectly" you would never lose at poker, but it's impossible to play perfectly, because perfectly means that you would have to know whether other players that are after you would bet/call/fold/raise (thus you would always know the exact pot odds), which means that already with this perfect play you must analyze the individual player. With this, actually, the computer can do fairly well if it can see a huge amount of the previous hands that a person has played and thus they will be able to know the person's general strategy. But, then again, part of the game is changing how you play. If the computer, or another person, has you pegged with a specific playing style, and either the evaluation is inaccurate for whatever reason (players do get incomplete information) or the player varies his playing strategy, which good players do. But what I mean by all this is that, in order to play "perfectly", the computer would have to have an algorithm that defines the perfect play at each point in the game. Statistically, there can exist a perfect way of playing particular hands so that you will win money in the long-run. However, you probably have to integrate a certain amount of randomness into your own play so that you're not predictable (because, otherwise you're only playing hands where you have a good hand already or a good draw, thus everyone knows what you have by what you bet and will fold when you have a hand, and bet when you do not bet, thus causing you to lose money or at least not make money) and human behavior analysis of other people so that you can look at a person's behavior and their betting, correlate it with the person's past behavior, and use that to help you to decide what to do. This is a big task, with a lot of human so-called "intuition" and mathematical analysis based on a person's past performance.
The other part of this is that even if you're playing in a statistically perfect manner and it's working (probably because you're against bad players, or you switch tables enough), it's not necessarily the best way to play against you're opponents. Because it's statistically perfect, you might be making money from your opponents. However, with another strategy, one that's tailored to your opponents rather than tailored to the statistics of the card, you could devise a strategy that can net you a lot more money than the statistically perfect strategy. And, of course, you're also not varying your strategy, so someone can pick up on it totally and play so as to not give you any money, or someone might pick up on little things that lower the amount of money you make off of hands.
You can't play "perfectly" because you're playing against people, and you're playing against them in a game with, literally, gazillions of combinations, both in card possibilities, in the betting of a particular hand, and in how your opponents act.
"The price of freedom is eternal vigilance." - Thomas Jefferson