|
|
#31
12-02-2005, 03:05 AM
|
|||
|
|||
Re: Bots - quote from a chessmaster to discuss
[ QUOTE ]
[ QUOTE ] After decades of development, chess bots reached parity with the best human players. If even 1% of funding devoted to chess expert systems went into poker expert system, it would be capable of beating world class poker players. [/ QUOTE ] People always behave as if chess and poker are the only games people want to beat with computers. Significant effort has gone into Go programs as well, with only marginal success. Martin [/ QUOTE ] The problem with Go is that there are just way more variantions plus the fact that stones can leave and then reappear on the board. Plus evaluation function is harder there, but computers ill get there, trust me. And there is no dispute that chess program research by far had way more attention through the years. Chess programs don't outplay humans in terms of strategy, pressure or understanding of the game. Their evaluation function and computing capabilities, coupled with openings book and table-bases, just present human player with a problems he is incapable to solve most of the times. He makes a subtle mistakes, computer capitalize on that. Computers don't know [censored] about chess and they don't need to. It's only a matter of time an effort, before it happens with such a simple game as poker. You don't need to know what human is thinking. You just create a model, sofisticated enough that human would be incapable of figuring out, therefore he would not be able to play optimally against it. Computers will always surpass humans in competitions within limited context. Only a matter of time and complexity of context. 
|
|
#35
12-02-2005, 02:24 PM
|
|||
|
|||
|
Re: Bots - quote from a chessmaster to discuss
I don't think he ment versus computers.
|
|
#36
12-02-2005, 02:27 PM
|
|||
|
|||
|
Re: Bots - quote from a chessmaster to discuss
There are some areas where bots will excel at. Preflop opening standards when folded to, for example. Or capping nut flush draws / straight flush draws in multiway pots. Playing sets.
In fact, I do not think it is incredibly difficult to write a bot to beat low stakes loose limit games.
|
|
#38
12-03-2005, 04:06 AM
|
|||
|
|||
|
Re: Bots - quote from a chessmaster to discuss
[ 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!
|
|
#39
12-03-2005, 05:47 AM
|
|||
|
|||
|
Re: Bots - quote from a chessmaster to discuss
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.
|
|
#40
12-03-2005, 05:52 AM
|
|||
|
|||
|
Re: Bots - quote from a chessmaster to discuss
[ QUOTE ]
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. [/ QUOTE ] Exactly!! I agree with your analysis.
|
 |
|
|
Linear Mode
