Poker Bot piece

[Pirc Defense]

[ QUOTE ]
I'm not quite that old Pirc. I am old enough to know that no bot will be able to beat above average players either in a ring game or tournaments till I am long gone so unless I am reincarnated I feel no need to holler "The Sky is Falling". Why do you?

[/ QUOTE ]

No idea where you're getting this "Sky is Falling" crap. I merely said that it's a definite fact that at some point, and almost certainly before you're long gone, computers will be able to beat humans in poker, and nearly any other game, as well. You supplied the "Sky is Falling" bit all by yourself.

What is so mystical about ring games? It's math, man. Especially online. Although we'd like to think otherwise, all of our intuitions and heuristical heroics boil down to probabilities and math, things a computer does substantially better at that us. It's hole cards, the cards on the table, the action so far, any db of previous actions by the players involved, and the money in the pot...extremely straightforward calculations. What can a human do better? Have a "feeling" that the opponent(s) are bluffing? It's just a probablitity. Make a strange play to confuse the opponent? Computers can do it, too.

To me it seems clear as day that, sooner than later, bots will be beating humans in poker, even in the mystical ring game. If programmed correctly, they'll never make a -EV play, and I know no humans of which this is true. Do you?

[Jimbo]

[ QUOTE ]
To me it seems clear as day that, sooner than later, bots will be beating humans in poker,

[/ QUOTE ]

I am sure it seems that way to many sub par poker players. It is always easier to externalize your losses than to blame oneself.

I will make one stipulation to your premise. If a perfectly programmed poker bot had several hundred thousand hands played with the exact same oponnents sitting in the same position at the same time of day in the same financial position in life as well as in the game with the same amount of rest and the same amount of stress and the same amount of interest in the game the poker bot would win.

Please let me know when this spectacle occurs, I'd like to watch.

[Pirc Defense]

[ QUOTE ]
I am sure it seems that way to many sub par poker players. It is always easier to externalize your losses than to blame oneself.

[/ QUOTE ]

Not sure what you're implying here, but I'll put the issue to rest by saying that I'm pretty sure I've never played against a bot. If I have, I've not noticed it. Sorry to burst your bubble.

[ QUOTE ]
I will make one stipulation to your premise. If a perfectly programmed poker bot had several hundred thousand hands played with the exact same oponnents sitting in the same position at the same time of day in the same financial position in life as well as in the game with the same amount of rest and the same amount of stress and the same amount of interest in the game the poker bot would win.

[/ QUOTE ]

Strawman. These conditions simply are not necessary for a bot to win. Let's say that all of your conditions hold, but that the opponent is one seat away from where he was the other "several hundred thousand" hands. If a bot can't make that adjustment, then it's not a "perfectly programmed" bot. Besides, what kind of rocket science does it take to adjust play based on an opponent's position?

You also make reference to things like rest and stress and interest. I think these types of things are what the "bots will never beat a human in a ring game" proponents hang their hat on. What, only another human can make an adjustment for a stressed-out opponent? How could you tell online, anyway? All the computer would see is actions, which it would be turning into mathematical probablities, and it would make +EV plays based on math, not some "gut feeling" that so-and-so opponent just lost his girlfriend.

The more we talk, the more I'm convinced bots will beat us up even sooner.

Well, AFAIK the University of Alberta currently has the most sophisticated bot. It's good enough that good players have to adjust their game to beat it.

The way that poker bots currently work is probably still that they don't really adjust to the opponent, but rather that they play in a way that is extremely difficult to exploit, and that will take advantage of certain categories of player errors. There's no real incentive to profile players in the environment of, say, party poker, where you're not likely to play a whole lot of hands against the same person anyway.

The mathematics for no limit and pot limit play is feasible, consequently, there's no real reason to believe that effective no limit computer players should be particularly difficult to write.

P.S. For those who are so inclined the University of Alberta does have a low-limit bot that you can play against on line for play stakes.

[nmt09]

What I find very interesting about this type of article is how they never talk money. So this guy turns up at your room and plugs in a "poker bot" a few seconds later you're playing online and ...... the article finishes.

The missing paragraph would probably read:

"and after 13 hours of watching our little metallic friend play we had earned $5.60 - another great day at the tables!"

If these people have so much faith why is it always the small stakes tables, why do they have to sell them to make a living if the bot could clean up?????

Complete rubbish...

[On a side note don't forget that the best bots recently came together to fight it out and Phil Laak took on the winner and kicked its arse]

[Pirc Defense]

[ QUOTE ]
On a side note don't forget that the best bots recently came together to fight it out and Phil Laak took on the winner and kicked its arse

[/ QUOTE ]

Thanks for your reply.

I think though, that a lot of people are making an error by aruging that, since we haven't yet built a bot that can beat the best players, we never will. Given the history of computational development, this seems prima facie short-sighted. To my knowledge, the mighty University of Alberta is the major player in developing bots. Could be others. But what if IBM threw it's weight at the puzzle? The millions of dollars and all the brainpower spent on Deep Blue couldn't come up with a bot that'd beat ya'? Now, I have no idea why it hasn't yet been attempted, or whether it has.

Keep in mind that when chess programs were first being developed, programs played like programs; that is, you could exploit a chess program because, for instance, they tended to value pieces more than position and tactics more than strategy. They were exploitable. Deep Blue does not have these exploitable tendecies.

We're at that stage now with poker bots. They are exploitable, opponents of the theory of poker bots say. Probably. But will they stay that way for long? No way. We'll look back at this argument and marvel at how silly it was. Of course computers will be able to beat humans in poker. It's inevitable.

[subzero]

[ QUOTE ]
Keep in mind that when chess programs were first being developed, programs played like programs; that is, you could exploit a chess program because, for instance, they tended to value pieces more than position and tactics more than strategy. They were exploitable. Deep Blue does not have these exploitable tendecies.

[/ QUOTE ]
Chess is a game of perfect information. There is a finite set of moves that a chess player can make and a computer can calculate/anticipate/analyze this. Poker is a game of incomplete information (randomness). Nobody knows what cards are coming next. This makes it difficult to program (especially for NL holdem).

If someone can write a poker bot AI that can beat NL holdem every time, despite the unknowns, that technology should be used for more important applications (e.g. military operations).

[subzero]

I read this article. More reason to play NL over Limit. This is the part that was really disturbing to me:

[ QUOTE ]
"Flick on Team mode and you can collude with other humans running WinHoldEm at the table."


[/ QUOTE ]

Hopefully the poker sites can detect when two people/bots are colluding (e.g. two people/bots raising/re-raising with trash to force people out and steal pots).

[Pirc Defense]

[ QUOTE ]
If someone can write a poker bot AI that can beat NL holdem every time, despite the unknowns, that technology should be used for more important applications (e.g. military operations).

[/ QUOTE ]

Good point. Keep in mind, though, that the test to determine whether a bot is at a world-class level is not that it can beat NL holdem every time. Impossible, as we all know. The goal is a bot that can hold it's own against world-class players, one that can earn it's $x/hr, like any professional player hopes to. At that point I'd say you have a pretty successful bot.

[ QUOTE ]
Nobody knows what cards are coming next. This makes it difficult to program (especially for NL holdem).

[/ QUOTE ]

I believe this is another example of the argument that there is something magical and mystical about the human mind that, in this application, a bot could never overcome. Hogwash. If anything, the computer has the advantage with unknown variables. Example: you and the computer both have a draw, say, a gut shot. There's a lot of money in the pot. The next card to come is unknown for both you and the bot, but the bot is able to calculate more accurately it's chances of hitting it's draw, so it's going to make the right decision every time.

"Well," you might say, "a good human might raise and ruin the bots' odds to draw." The human programming the bot is too thick to not realize this and program accordingly, and maybe even have the bot do such a dastardly thing every now and then to ruin the humans' odds?

I definitely don't see the major leap it takes from playing against a single opponent to moving up to a ring game, and I'm starting to not see why no-limit is such a difficult hurdle, vis-a-vis limit.

Can someone explain to me why no-limit is so unsurmountable?

[CCovington]

I think if the bot was really capable of making a ton of money, there would be no way he would be offering it for sale to the public.
However, the more people that hear about this the less attractive online poker will become. I hope all the online poker rooms are taking major action to sniff out and get rid of the bots.