Bots - quote from a chessmaster to discuss

[SNOWBALL138]

[ QUOTE ]
Nice find. Good post


Quote:
--------------------------------------------------------------------------------

bots are more difficult in poker because there is a bigger randomness factor with what cards can fall and the actions to each situation. Chess bots have only to know the basic principles, opening strategies and the like. For instance, knowing that controlling the center of the board above all else is easy enough to program, but how do you program a computer to make the best possible move in poker knowing that any number of random cards can fall on 3 different streets. In poker, you would just have to program it to make the "mathematically correct" decisions solely based on odds.


--------------------------------------------------------------------------------



Solid point

[/ QUOTE ]

Bots can beat low lvl SNGs w/o too much trouble.

[MrBlueNose]

In before OrangeKing

[stigmata]

[ QUOTE ]


Bots can beat low lvl SNGs, low stakes NL & can play strong HU poker w/o too much trouble.

[/ QUOTE ]

It's a matter of when, not if....

[crazy canuck]

I would also like to point out one other difference between chess and poker. Poker is played for money. So the methodology to develop a succesful bot is way less likely to be published. Academic finance suffered and suffers from the same problem. The best models to forecast prices and to price derivatives are not published. This is why many academics believed in the efficient market theory for decades. Now there are very few empirical researchers belive
in the EMH.

Same goes for poker bots. So if you were an academic, would you rather get the recognition from a few other AI researchers or make millions. Or even better you can publish your marginal results, get the academic recognition and still make millions.

[Sniper]

Canuck,

Sorry to inform you, but chess is also played for money [img]/images/graemlins/wink.gif[/img]

[FlFishOn]

Who would be foolish enough to play internet chess for money? Blitz excepted maybe.

[citanul]

[ QUOTE ]
Who would be foolish enough to play internet chess for money? Blitz excepted maybe.

[/ QUOTE ]

hahahaha. cuz computers can't think fast.

I don't think he ment versus computers.

[Sniper]

[ QUOTE ]
Who would be foolish enough to play internet chess for money? Blitz excepted maybe.

[/ QUOTE ]

Online... There are fish everywhere [img]/images/graemlins/wink.gif[/img]

Offline... there are many chess tournaments with cash prizes, and you can generally find a cash game in the park or a local chess club, if thats your thing!

[UATrewqaz]

I have a degree in CS, work as a software developer, and took several AI courses during college... so I have SOME info on this subject but not a deep expertise by any means.

Nobody will ever build a program to play poker at the same level programs exist to play chess, because as everyone points out every single thread about bots, poker is a game of incomplete information.

Any game of complete information you can easily build a program to play perfectly. The only thing stopping a computer from playing perfect chess is the processing power/computation. Even the best computers in the world can't analyze every possible move, and then every possible countermove, and then ever countermove to those countermoves (you get the idea) say 200 hands deep. The state tree simply grows too large, too quickly (even for a computer).

Poker is a game of incomplete info, which makes it even harder... HOWEVER...

You could hypothetically build a program that plays exceptionally good poker. It would never make a single mathematically bad decision (in terms of pot odds, %).

The only thing a program cannot do as well as a human is hand reading, however if you gave the program the ability to analyze hand histories of opponents (that is it does not play each hand in a vacuum) then it would soon SURPASS any human's ability to hand read for the following reasons

1. It would spot EVERY pattern in your play, no matter how slight.

2. Even if you attempted to "vary" your play it would know how often you vary it and could incorporate that into its decision. (there is a x% chance he was playing this hand in this way, do I still have him beat/right odds, etc.).

There already exists much expertise in the field of opponent modeling AI.

Ultimately the bottom line is this...

The program may not be a "big" winner but it would still be a terrible terrible opponent to have in your game.

Poker profit is based 100% on other people's mistakes (long term) and this program is simply not going to make ANY large mistakes and will make only a few small ones here and there.

BOTS WILL EVENTUALLY BECOME A VERY BIG PROBLEM ONLINE IF SITES ARE NOT AGGRESSIVE ABOUT STOPPING THEM, assuming that processing power of home computers continues to grow and serious development of commercial poker playing software continues.

They will steam roll weak players and turn once fish filled tables into tables of 9 TAG's who never make any large mistakes and the rake will be the only person making money.

Any algorithm/thought process/decision making a human can make at a poker table can be modeled in software. Saying "poker is a game of incomplete information" doesn't make the program useless. You are playing with incomplete information as well.