Vote Chess: Game 2

[LaurV]

Talking about some program that anybody has: how do you rank yourselves against the default Win7's Chess Titans?

I know this program is not on the first 50 or so in the world, but it is something free, which anybody has (it comes with win7 default games, as solitaire and freecell).

I usually win (almost) all games as white, regardless of the level, and I stand a good chance as black at level 7-8. I usually play fast, taking about the same time to think as the machine takes. When I think longer, it does not stand a chance.

I may be deluded, as Brian said, as I never played against a "strong" engine.

I played Go, Gomoku, Renju, against very strong engines (like Meijin, or respective Fiver, etc), given me enough time, they don't stand a chance (especially Go) (I may be "out of fashion" here too, last time I played Go against a computer was like 4 or 5 years ago). The only board game where I don't stand a chance against computer is Scrabble. And I was top-class scrabble player 20 years ago (best performance: coming third in the national championship). But playing Scrabble against computer is difficult, due to the large vocabulary you must have, and due to the huge amount of combinations you have for each move. Even if you play "open" (you can see the opponent's tiles), and extract the letters in a predefined order (a string is established at the beginning of the game, if you deploy 2 letters, you take the next two from the string), still 99.9999% of the humans don't stand a chance against computer (this remark includes a lot of personal experience with that game, too!).

Anyhow, I still think we do have a chance, as a team, to beat the computer at chess. Any computer. Call me an optimist, but we played here two very good games. These are like 100 times better than the games I usually play (and still win) against Chess Titans.

[Batalov]

Originally Posted by LaurV

..I may be "out of fashion" here too, last time I played Go against a computer was like 4 or 5 years ago...

You are. A lot of things happened even over the two years, before two years ago.

Go is the only game interesting to play (except maybe also long gammon*). We should play the next team game -- the game of Go. Chess is boring, like grammar. Computers spell better than most humans these days.

__________
*which nobody knows, and that's ok

[gd_barnes]

Originally Posted by LaurV

That's the fun, to find the "right" set of rules which does not change the game, but screw up the AI. That is why there are 50, or more, and we "vote" for few of them. Taking the one rule with the blocked pawns, for example, this would not pose any problem for a human player, and additionally, how many times do you find blocked pawns in a real game? And if so, ho many times would be actually "better" to jump over it? This does not influence the game too much, but it is enough to have a single pawn blocked by other pawn, and all the "backtrack" of the engine is screwed up. You even don't need a "blocked pawn" on board, it can appear after few moves, the AI will discard all the branches which could derive from jumping over the pawn, and get lost doing "standard" chess analysis.

I like this idea a lot. It wouldn't hurt to vote on some possible variations even if people ultimately decide they don't want to do it.

Related to the second part of the post, playing against a known engine, I don't believe any actual chess program can beat us, if we are allowed an opening book (from the web), "days" to think for each move, and a board to analyze. No way a computer could beat me under these circumstances... But you never know, and it will still be a lot of fun to try.

I couldn't disagree more. Yes you guys would likely beat it if you were allowed days to think and it was only allowed 2-3 minutes. But you would likely be beaten if you had days and it had days. After the 1st game was finished here and I had done and posted my own analysis, I allowed an older version of Fritz think for just a few hours on several of the moves of the 1st game here to see if it could improve upon my analysis and/or the actual game move. It came up with clearly better moves in a majority of situations, especially for white. I think you will be surprised by the power of programs today.

LaurV, I'll make you a deal. Give me some time to go buy a modern chess program and you can play against it. I'll post the moves here. We can make the time controls whatever you want although I feel best is to allow for game-based time controls so that you and/or the program can move quickly in forced lines to accumulate additional time allowing you to think for quite a while in tough situations. Unless you are of grandmaster talent (maybe I'm underestimating you), I predict the program will beat you. Regardless, what is your over-the-board rating? Do you have a correspondence rating?

Reference what you said here and something that Brian said about another one of the players using a program for some of the opening moves: It is legal in all correspondence chess to use powerful opening books online as long as a program does not analyze the moves. Chessbase has a very powerful opening book that shows percentage of games won for each response for white and black up to more than 20 moves deep in some variations. It is why I kept recommending against playing the closed Sicilian for white. Black equalizes quickly in most variations, winning at least half and sometimes a majority of points in high level play. Perhaps he was referring to someone actually using AI to do early game moves shortly after the game was already out of most standard opening books. Regardless, did the person admit to what he did? Was he expelled from the team after it was found out?

[Brian-E]

Originally Posted by gd_barnes

Regardless, did the person admit to what he did? Was he expelled from the team after it was found out?

Yes, and yes but not in the immediate aftermath of this particular incident.

The relevant posting (read only if you are feeling patient!) is this one:
http://www.mersenneforum.org/showpost.php?p=321650

[Xyzzy]

The first chess computer we owned that could beat us, all of the time, was this one we bought sometime around late 1991:

http://www.spacious-mind.com/html/module_polgar.html

We had that exact model in that wood housing. It was beautiful, and very good at chess!

[Brian-E]

Originally Posted by Xyzzy

The first chess computer we owned that could beat us, all of the time, was this one we bought sometime around late 1991:

http://www.spacious-mind.com/html/module_polgar.html

We had that exact model in that wood housing. It was beautiful, and very good at chess!

That was a famous one. Never had it but I would have liked to.

My first was this crumby one, about 1980. It was pretty weak (though that suited me). It would only recognise pawn promotions to queen. But it had a few nice features too.
http://tluif.home.xs4all.nl/chescom/EngCc10.html

[gd_barnes]

Originally Posted by Brian-E

That was a famous one. Never had it but I would have liked to.

My first was this crumby one, about 1980. It was pretty weak (though that suited me). It would only recognise pawn promotions to queen. But it had a few nice features too.
http://tluif.home.xs4all.nl/chescom/EngCc10.html

Ah I remember that one well and you're right, it did have a few nice features for the time. I was (very) thrilled to get it as a Christmas present my senior year in high school in 1979. The main thing that I remember is that it had a horrible end game, even at the longer time controls. So all I had to do was avoid losing material and I could beat it in the end. I do recall that the version I had would allow pawn promotion to any piece. Perhaps I had a slightly later version than you.

[Brian-E]

Originally Posted by gd_barnes

Ah I remember that one well and you're right, it did have a few nice features for the time. I was (very) thrilled to get it as a Christmas present my senior year in high school in 1979. The main thing that I remember is that it had a horrible end game, even at the longer time controls. So all I had to do was avoid losing material and I could beat it in the end. I do recall that the version I had would allow pawn promotion to any piece. Perhaps I had a slightly later version than you.

Yes, it was all very thrilling at the time, wasn't it, when chess computers came on the market.
Maybe I'm mis-remembering about the pawn promotions, or maybe you did indeed have a better version. I agree about its atrocious endgame technique!

[paulunderwood]

I have one like this. It will move out a knight and you can chase it all over the board, thereby developing all of your pieces.

I play against Gnu Chess nowadays -- I have beaten it once.

[Xyzzy]

We are still hoping that someday we will find a chess engine that allows us to play with just the pawns.

http://www.mersenneforum.org/showthread.php?t=17230

[gd_barnes]

Originally Posted by Xyzzy

We are still hoping that someday we will find a chess engine that allows us to play with just the pawns.

http://www.mersenneforum.org/showthread.php?t=17230

You can set up positions in many chess programs. In my older version of Fritz, I could set this up. What it does allow is for you to clear the board and then set up whatever position you want as long as it is a legal position. For example, I don't think it will allow you to put pawns on the first rank or have 10 queens or to have black in check with white to move. But it would allow 9 queens since (in theory) that is possible. I haven't tried it with 10 or 11 rooks. 10 rooks would (in theory) be possible; 11 would not.