Poker Bot piece

[UATrewqaz]

Most of the people here who would be good enough to write a profitable bot strategy are WAY more profitable simply playing themselves.

There are probably very few bots playing currently and most of them probably play a very tight, break-even'ish game.

[Pirc Defense]

[ QUOTE ]
There are probably very few bots playing currently and most of them probably play a very tight, break-even'ish game.

[/ QUOTE ]

But that's really the point, no? If "most of them...play a...break-even'ish game" then they're already better than the average poker player. That's no small accomplishment. And this "industry" is in its infancy. Wait until some realy money is thrown at the problem.

So crystal clear.

I think the are probably alot of bots out there playing mid-limit holdem. It really shouldn't be that hard to program one to play tight aggresive. Combined with alway knowing the correct action based on pot odds I would think it would be a profitable player. Yes if a good player knew he was playing a bot I think he could win but how would you know? Also if you look at the winholdem web site these bots can be programed to communicate with each other. If a person had 2 or worse yet 3 bots at the same table even a good player would be in serious trouble.

I think there is more to this than meets the eye and some creditable magazine should reseach this so the average player know what they are up against.

[ QUOTE ]
Most of the people here who would be good enough to write a profitable bot strategy are WAY more profitable simply playing themselves.

There are probably very few bots playing currently and most of them probably play a very tight, break-even'ish game.

[/ QUOTE ]

Next you'll be suggesting that someone needs to be able to play checkers well in order to write a winning checkers program - which is not true. It's handy to be able to test your work though.

The fact is that computers don't have the intuition or learning capability that humans have, but they also have discipline, patience, and undying staminal. The things that make being a professional gambler unattractive -really. The need to play a lot, the need to keep things in mind. Moreover, the learning ability of a human is often a liability since winning longshots effectively trains players to stay in too much.

With strong theory, and correct implementation, the computer can be superior to a human player.

[Pirc Defense]

[ QUOTE ]
With strong theory, and correct implementation, the computer can be superior to a human player.

[/ QUOTE ]

Of course it can, but don't tell Jimbo. He'll just ask you why you're seeing bots under the bed.

[eastbay]

For the record, Pirc Defense has the credible side of this argument and clearly has the knowledge to form a credible opinion. Jimbo does not, and frankly should just shut his trap, because he doesn't know what he's talking about.

eastbay

While I think that a good bot could do well in limit holdem, such programs will never be able to beat good/excellent players in no limit holdem or 7 card stud. Both games involve more than just simple logic and probability. The intuition factor is too big in these games for a computer program to excel above the good players. These games are not chess.

[Pirc Defense]

[ QUOTE ]
While I think that a good bot could do well in limit holdem, such programs will never be able to beat good/excellent players in no limit holdem or 7 card stud. Both games involve more than just simple logic and probability. The intuition factor is too big in these games for a computer program to excel above the good players. These games are not chess.

[/ QUOTE ]

Intuition, at least in this sense, is way overrated. What is intuition? Knowing without knowing why you know? Can it be implemented in a computer? If not, can a program emulate intuition enough so that it appears to have intuition? If so, then it's enough to be a successful poker bot.

I see very little inherent in an on-line limit game that can't be represented quite accurately by probablities and discrete rules. Who's to say that isn't how we're operating, anyway? What we call intuition can very well be just the end result of massive and fast, but still mundane computation, which I differentiate from cognition.

As an aside, if anyone is interested in the above paragraph, I wrote an article in the FEB 1993 edition of The World and I titled "The Artificiality of Artificial Intelligence," that covers the computation/cognition and syntax/semantic dichtomy, though it's not terribly technical as it was published in a general readership magazine.

As another poster kindly pointed out, I know a little of what I'm talking about, and yes, Jimbo is absolutely on the wrong side of this discussion.

[Matt Flynn]

The primary collusion would be seeing the other bot-users' hole cards in real time. You could then pull a squeeze play, but the real advantage (and the undetectable one if not ridiculously abused) is always folding the second-best hands within your group.

[Matt Flynn]

Interesting discussion.

I was surprised by the consensus that no limit holdem would be harder to defeat than limit. Limit is more complicated mathematically. Also, no limit heuristics and play variation would be easier to write. It would still be very time-consuming to do well. The lack of good heuristic reading skills would be compensated for by tiltlessness and identifying opponent game flaws. Most no limit players have major flaws identifiable over a few thousand hands or less. A simple example is the legions of players who never raise out of the blinds or reraise big preflop without AA or KK. There are many more.

Copies of a good bot would obviously end up on dozens of computers, accounting for thousands of hours of table play a week at little cost and with minimal concern for detection if one did not get too greedy. A sophisticated team could easily defeat methods like type-the-word, using humans to surveil dozens of computers. Bots exist now and in number. I do not think they are winning significantly at the 5-10 and up level - at least not shorthanded. There aren't a lot of players good enough to crush those games, and only a small subset of those could write mirroring rulesets.

Matt