chess genius is sick at poker

honestly its surprising that we don't see more high level chess players in poker considering how successful starcraft and magic players have been.

I guess the psychological part is not so big in chess. For example as Patrocle explained well, there are IA programs now unbeatable but we are far to have those in poker.
And in starcraft, you have to adapt so much to the other, something you find also in poker.
so not so surprising imo.

I had a radioshow with him while we were in Berlin, the guy talks like several languages and despite not talking it for months - he did the entire show in Dutch and was easy to understand. He's a very nice guy, always friendly and pretty good at poker I think.

Two of the guys I know who play the highest stakes at Stars weren't at all impressed by him I must say which was bit of a downer after hearing such great stories from everyone. Gotta say, these guys aren't easy to impress but they said he had some real big leaks to solve before becoming really superb.

@kingpowa: That is not correct, there are no unbeatable chess computers. But still, there is a reason computers play chess and Scrabble much better than they do poker or bridge: computers do not have a "theory of mind", i.e. a concept of how human opponents think which lets one develop strategies that exploit those quirks. There is also the fact that card games involve incomplete information, which means it takes more computing power to make the calculations, although I think this is a lesser concern: computers can still do a math problem involving ranges (and there are computers which play short stack NL exceptionally well).

@NMcNasty: I think it's because the kind of analysis required for poker requires some personal skills, which chess doesn't really develop. Because the board and chess pieces are an abstraction, you can ignore your opponent, and probably should, in order to focus. So being socially clueless is of no disadvantage in a chess game. In poker, the typical Aspie type who has chess skills has to develop that theory of mind, as ignoring your opponent is generally disastrous.

On August 16 2010 15:11 kingpowa wrote: I guess the psychological part is not so big in chess. For example as Patrocle explained well, there are IA programs now unbeatable but we are far to have those in poker. And in starcraft, you have to adapt so much to the other, something you find also in poker. so not so surprising imo.

Psychological part is huge in chess and that is one of the biggest problems for a GM when facing a strong computer, or was when those matches made sense and were held.

Starcraft and poker are games of imperfect information, while chess is a game of perfect information. The reason why the programs are so successful @ chess is that there has been enormous amount of interest in chess programming, mainly because chess was always considered something like an intellectual stronghold for people, a bit mysterious with all that "intuition", "sense" and patterns and things that seems you cannot program. So they cracked it
Modern chess programs are really cool.

starcraft should be (much) easier in theory, but you cannot do much in practice because there is no usable API AFAIK, like you cannot take over micro. Imagine a program with arbitrarily high APM which never does any mechanics mistakes. You have very little space left against such a program, i.e. you have to obtain a huge edge by tricking it with your builds, because you are at a huge disadvantage at micro/macro. Obv. it is not possible to do this for a long time, so as soon as the developers detect how to do proper scouting/countering to tricky builds, it would be over. Note that the computer should not be as good as humans in the metagame, it just would need not to suck balls and the mechanics would decide.


Poker strategies change with the stack size. For certain (small) stack sizes Poker is solved. For deep-stack poker I guess writing a competent bot would be hard. Several opponents is harder than HU probably. I believe it is still easier than chess, but it took them 30 years to solve chess in a way that a good program on a laptop rapes a GM. With poker it should come faster because after all there is so much money involved.

sick

I played chess until I was 12 and I was also a "prodigy".

Now that I have learned how to play go I marvel at how much depth people see in chess. From the point of view of go, it's such a limited game.

The reason why the programs are so successful @ chess is that there has been enormous amount of interest in chess programming, mainly because chess was always considered something like an intellectual stronghold for people, a bit mysterious with all that "intuition", "sense" and patterns and things that seems you cannot program. So they cracked it

The same holds true for go, and although enormous progress has been made, the best computers still struggle against your average club player.

The real reason computers outrank humans in chess by so much is that superior tactical foresight is worth a lot in chess and there is not enough strategical depth for humans to be able to outweigh that advantage.

What's happening with Chess is that it's gradually losing its place as the par excellence of intellectual activity. Smart people in search of a challenging board game might try a game called Go.
~Hans Berliner, The New York Times, Feb 6, 2003

I agree what you said regarding go and tactics.

In go, pieces dont move, so it's much easier for humans. And then again, there interest in chess has been much bigger than in go, maybe because so much more people know and play chess somewhat regularly compared to go.

The tactics thing regarding chess came after the programs started to play so well. So basically it was not part of the game before the computers came in. If you are a player: all those super-ugly openings they play today with grabbing the pawn and holding it in a way which seems very anti-positional (all those Moscow gambits or whatever it is called in Slav Defense, for example); those kind of things became possible because of computers. Before that, humans did not know they could hold it, and they didn't play it. Chess was much more "positional" in 60s and 70s than now.

What I'm saying-the programmers did not KNOW tactics was so important. They found it out: it was an engineering achievement, one of the many they made in the field.

it's not that simple, chess has alot more limited moves whereas go has alot less constraints.

On November 29 2009 18:06 Patrocle wrote: Even an iPod with shredder on it -it is a famous open-source chess-IA designed by an IM - can defeat anybody.

Opensourcewhathaveyoubeensmokinglolwut?


and it isn't designed by an IM. seriously.



(also... do you mean AI? or is the joke on me for that one...)

go-moku also has less constraints than chess and more valid moves on average probably. Accidentally, the variant with LEAST amount of constraints has been solved. Like checkers, which also has less constraints than chess and has been solved too.

to make it clear: I'm not arguing chess is harder than go from programming point of view. As of now go seems to be much harder, but we know better when people actually spending as much effort on go as they did in chess. I suspect right now there are an order of magnitude more active chess programmers than go programmers.

On August 16 2010 18:31 lucifer wrote: Show nested quote + On November 29 2009 18:06 Patrocle wrote: Even an iPod with shredder on it -it is a famous open-source chess-IA designed by an IM - can defeat anybody. Opensourcewhathaveyoubeensmokinglolwut?

http://itunes.apple.com/us/app/glaurung-chess/id305558605?mt=8

They say it's free.

Not tried it myself but I know the PC version and it is one of the strongest engines and it's free.

Edit: this is not what he meant obv. but the PC version is open-source, and it is rated 2-nd in the computer chess rating lists after Rybka 4.

pluzich so? He said shredder. I'm like wtf?




He also said it was designed by an IM.

Michael Feygin
Leonid Milov
Thomas Henrichs
Jan Michael Sprenger

are the four IM's from germany. The designer/programmer/whatever of shredder is not among them.



edit: the Stockfish engine is opensource yes. took forever to find out wtf you meant tho. Might help if you provided the name. (Stockfish)

Jeff is a great guy, I've known him for a over a year now and he's definitely an intense guy with mega talent in poker/accomplishments so far

On August 16 2010 18:46 lucifer wrote: pluzich so? He said shredder. I'm like wtf? He also said it was designed by an IM. Michael Feygin Leonid Milov Thomas Henrichs Jan Michael Sprenger are the four IM's from germany. The designer/programmer/whatever of shredder is not among them. edit: the Stockfish engine is opensource yes. took forever to find out wtf you meant tho. Might help if you provided the name. (Stockfish)

He confused everything there. The famous engine by IM is Rybka. Version 2.3.2 is free, which is quite old. I think there is no iPhone app for that.

Shredder sucks imo.

P.S. When you asked for a link in your initial post I thought you just wanna free program for your iPhone

On August 16 2010 16:05 pluzich wrote: Show nested quote + On August 16 2010 15:11 kingpowa wrote: I guess the psychological part is not so big in chess. For example as Patrocle explained well, there are IA programs now unbeatable but we are far to have those in poker. And in starcraft, you have to adapt so much to the other, something you find also in poker. so not so surprising imo. Psychological part is huge in chess and that is one of the biggest problems for a GM when facing a strong computer, or was when those matches made sense and were held.

I didn't know that.
Could you please explain why how ?

If you are familiar with the programs used (shredder or Rybka), do you know if (and if yes, how) it takes into account the moves of the other player during the match or the previous one ? How it "adapts" to the opponent ?

I just knew their existence but not how they work. I thought that they would almost take independently every move and calculate the best one, processing a huge amount of scenarios.

On November 29 2009 10:26 nolan wrote: it's not like an MTT player actually plays significant samples to have an accurate picture of how someone plays anyways. it would take years of grinding MTT with someone to actually have a grip on their game.

+12

playing yes, talking no

On August 17 2010 06:03 kingpowa wrote: Show nested quote + On August 16 2010 16:05 pluzich wrote:   On August 16 2010 15:11 kingpowa wrote: I guess the psychological part is not so big in chess. For example as Patrocle explained well, there are IA programs now unbeatable but we are far to have those in poker. And in starcraft, you have to adapt so much to the other, something you find also in poker. so not so surprising imo. Psychological part is huge in chess and that is one of the biggest problems for a GM when facing a strong computer, or was when those matches made sense and were held. I didn't know that. Could you please explain why how ? If you are familiar with the programs used (shredder or Rybka), do you know if (and if yes, how) it takes into account the moves of the other player during the match or the previous one ? How it "adapts" to the opponent ? I just knew their existence but not how they work. I thought that they would almost take independently every move and calculate the best one, processing a huge amount of scenarios.

I think he means the psychology affects the human players. It doesn't affect the computers much. Modern chess engines uses an evaluation function based on material advantage and how "active" all pieces are to decide a score for a position. It then basically "bruteforces" a LOT of positions trying to see what can get it the best score in the long term.