Man vs Machine

[Teargarden]

Carlsen vs Houdini 3!!!!
Lets have one final match between Man and Machine!!

[tano-urayoan]

[quote=Teagarden]Carlsen vs Houdini 3!!!!
Lets have one final match between Man and Machine!![/quote]

Grandmasters acknowledge that engines are too much powerful for humans to play against.

[mschribr]

I also want to see this match. If Carlsen loses badly then I would make 2 changes to make it easier for Carlsen. Houdini 3 will on run a cell phone and draws count as wins for Carlsen.

[Don]

I also want to see this match. If Carlsen loses badly then I would make 2 changes to make it easier for Carlsen. Houdini 3 will on run a cell phone and draws count as wins for Carlsen.

I like the draws count as wins.

I don't like to handicap the computer other than using a position handicap, otherwise you are not really playing a "modern" program in some sense but pretending you are 10 years old in hardware. So let the computer play full strength other than the initial handicap to properly see what it can overcome. So one possibility is to start from the opening position with the computers castling rights turned off. Another possibility is to let Carlsen have the first 2 moves and start from there. If that is not enough you could combine the 2 handicap ideas or come up with others. You could swap the rook and knight positions for the computer to place the knights immediately on the worse possible squares and of course that spoils castling right too.

How about the right of takeback? Grant Carlsen N takeback moves which he must exercise before playing his NEXT move.

Another possibility is to let Carlsen stop the game at any time and replay from ANY point he chooses. That could be allowed once or twice. Some of these ideas are not my own, but this one is - I have always thought that would be an interesting handicap.

[mwyoung]

I also want to see this match. If Carlsen loses badly then I would make 2 changes to make it easier for Carlsen. Houdini 3 will on run a cell phone and draws count as wins for Carlsen.

I like the draws count as wins.

I don't like to handicap the computer other than using a position handicap, otherwise you are not really playing a "modern" program in some sense but pretending you are 10 years old in hardware. So let the computer play full strength other than the initial handicap to properly see what it can overcome. So one possibility is to start from the opening position with the computers castling rights turned off. Another possibility is to let Carlsen have the first 2 moves and start from there. If that is not enough you could combine the 2 handicap ideas or come up with others. You could swap the rook and knight positions for the computer to place the knights immediately on the worse possible squares and of course that spoils castling right too.

How about the right of takeback? Grant Carlsen N takeback moves which he must exercise before playing his NEXT move.

Another possibility is to let Carlsen stop the game at any time and replay from ANY point he chooses. That could be allowed once or twice. Some of these ideas are not my own, but this one is - I have always thought that would be an interesting handicap.

I don't like the idea of destroying the chess game by any kind of position handicap, or extra moves. And I don't like the idea of draws counting as wins for Carlsen, since chess is a drawish game. This could lead to a very dull match.

I would just limit the maximum points the computer could score to only 1/2 point on a win or a draw. If you want to give the computer a handicap, but still have a true chess game and match. This way Carlsen could not draw his way to victory.

So if the game is a draw, the computer and human player each receives 1/2 point, If Carlsen wins a game he receives a full point and the computer 0 points. If the computer wins, Carlsen receives 0 points and the computer again only 1/2 point.

[mschribr]

Chess is a drawish game at much higher levels. We are far from those high levels. If the computer is 450 points higher, then Carlsen could not force a draw. If there is draw, Carlsen will be fighting hard to get it.
Take backs are ok. I do not think Carlsen will gain anything from take backs.

[Don]

I also want to see this match. If Carlsen loses badly then I would make 2 changes to make it easier for Carlsen. Houdini 3 will on run a cell phone and draws count as wins for Carlsen.

I like the draws count as wins.

I don't like to handicap the computer other than using a position handicap, otherwise you are not really playing a "modern" program in some sense but pretending you are 10 years old in hardware. So let the computer play full strength other than the initial handicap to properly see what it can overcome. So one possibility is to start from the opening position with the computers castling rights turned off. Another possibility is to let Carlsen have the first 2 moves and start from there. If that is not enough you could combine the 2 handicap ideas or come up with others. You could swap the rook and knight positions for the computer to place the knights immediately on the worse possible squares and of course that spoils castling right too.

How about the right of takeback? Grant Carlsen N takeback moves which he must exercise before playing his NEXT move.

Another possibility is to let Carlsen stop the game at any time and replay from ANY point he chooses. That could be allowed once or twice. Some of these ideas are not my own, but this one is - I have always thought that would be an interesting handicap.

I don't like the idea of destroying the chess game by any kind of position handicap, or extra moves.

You are not really playing a modern program if you use hardware handicaps, which is equivalent to a big time handicap. We might as well play against 10 year old programs on Pentiums.

I talked to Larry about this and he said the same thing, that basically you want to play a modern program at it's best or it ceases to become a real match. There is rich tradition of handicap matches to equalize players in chess. In Go it's even built in to the ranking system. So with some sort of handicap based on starting from a different positions (instead of the opening) both players play the game at their full strength - that is what people want to see. I don't care how strong or weak Houdini is set to 1 ply searches.

Komodo and Stockfish and probably other programs have the ability to cripple the play with a user setting. So why not just use one of those settings instead if your intent is to cripple up a program? The answer is pretty obvious, it wouldn't really be representative of what the program can do. The principle is to see what the program can do playing it's best.

You probably remember the handicap matches Rybka played a few years ago. They were all played with Rybka at full strength on modern hardware and even with multi-processor for this same reason, even though they were given serious handicaps based on the position the started from.

And I don't like the idea of draws counting as wins for Carlsen, since chess is a drawish game. This could lead to a very dull match.

I would just limit the maximum points the computer could score to only 1/2 point on a win or a draw. If you want to give the computer a handicap, but still have a true chess game and match. This way Carlsen could not draw his way to victory.

So if the game is a draw, the computer and human player each receives 1/2 point, If Carlsen wins a game he receives a full point and the computer 0 points. If the computer wins, Carlsen receives 0 points and the computer again only 1/2 point.

[mwyoung]

I also want to see this match. If Carlsen loses badly then I would make 2 changes to make it easier for Carlsen. Houdini 3 will on run a cell phone and draws count as wins for Carlsen.

I like the draws count as wins.

I don't like to handicap the computer other than using a position handicap, otherwise you are not really playing a "modern" program in some sense but pretending you are 10 years old in hardware. So let the computer play full strength other than the initial handicap to properly see what it can overcome. So one possibility is to start from the opening position with the computers castling rights turned off. Another possibility is to let Carlsen have the first 2 moves and start from there. If that is not enough you could combine the 2 handicap ideas or come up with others. You could swap the rook and knight positions for the computer to place the knights immediately on the worse possible squares and of course that spoils castling right too.

How about the right of takeback? Grant Carlsen N takeback moves which he must exercise before playing his NEXT move.

Another possibility is to let Carlsen stop the game at any time and replay from ANY point he chooses. That could be allowed once or twice. Some of these ideas are not my own, but this one is - I have always thought that would be an interesting handicap.

I don't like the idea of destroying the chess game by any kind of position handicap, or extra moves.

You are not really playing a modern program if you use hardware handicaps, which is equivalent to a big time handicap. We might as well play against 10 year old programs on Pentiums.

I talked to Larry about this and he said the same thing, that basically you want to play a modern program at it's best or it ceases to become a real match. There is rich tradition of handicap matches to equalize players in chess. In Go it's even built in to the ranking system. So with some sort of handicap based on starting from a different positions (instead of the opening) both players play the game at their full strength - that is what people want to see. I don't care how strong or weak Houdini is set to 1 ply searches.

Komodo and Stockfish and probably other programs have the ability to cripple the play with a user setting. So why not just use one of those settings instead if your intent is to cripple up a program? The answer is pretty obvious, it wouldn't really be representative of what the program can do. The principle is to see what the program can do playing it's best.

You probably remember the handicap matches Rybka played a few years ago. They were all played with Rybka at full strength on modern hardware and even with multi-processor for this same reason, even though they were given serious handicaps based on the position the started from.

And I don't like the idea of draws counting as wins for Carlsen, since chess is a drawish game. This could lead to a very dull match.

I would just limit the maximum points the computer could score to only 1/2 point on a win or a draw. If you want to give the computer a handicap, but still have a true chess game and match. This way Carlsen could not draw his way to victory.

So if the game is a draw, the computer and human player each receives 1/2 point, If Carlsen wins a game he receives a full point and the computer 0 points. If the computer wins, Carlsen receives 0 points and the computer again only 1/2 point.

I was not clear, I don't want a hardware handicap either, I could live with a scoring handicap, that does not allow the human player to win the match without ever beating the machine once. Under my system, Carlsen would have to win one game to win the match, instead of all draws to win the match. And if Carlsen did win a game, the computer would have to win 2 back from Carlsen to draw the match, and 3 wins to regain the lead.

That is a big handicap for the human, that the computer could overcome.

[biscuit1953]

I haven't been keeping up with chess that much but I do realize that computers have come to the point of dominating even the highest rated human players. Bobby Fischer was concerned that chess would be played out in the top levels of competition so he started promoting Fischer Random Chess in hopes of preventing that possibility. Out of curiosity, do computers dominate against Grandmasters in FRC to the same degree as classical chess? Since there is no opening database, do the computers still have such an overwhelming advantage? What about correspondence chess? I read one time that a correspondence GM was actually rated much higher than their over the board counterparts.

[Don]

I haven't been keeping up with chess that much but I do realize that computers have come to the point of dominating even the highest rated human players. Bobby Fischer was concerned that chess would be played out in the top levels of competition so he started promoting Fischer Random Chess in hopes of preventing that possibility. Out of curiosity, do computers dominate against Grandmasters in FRC to the same degree as classical chess? Since there is no opening database, do the computers still have such an overwhelming advantage? What about correspondence chess? I read one time that a correspondence GM was actually rated much higher than their over the board counterparts.

I don't have an answer, but my educated guess is that computer would dominate even more in FRC.

Don