Is Rybka really that Great ?

[Dr.Ex]

A Grandmaster should be able to win this positions against God, provided God doesn't cheat:


I don't agree that the situation is so easy as you describe.
A GM has enough chances to win these positions against a perfect player, but it is not easy to do it.
To do it Benjamin had to open the positions most times and opening the position against a computer is not the most desirable thing a 2576 GM would want.

That does not apply for overwhelming positions like these.
White's King is perfectly safe in all this positions. In the first two diagrams there is no risk necessary, white just has to trade the right pieces and it's over soon.
The last diagram requires also no risk . Really, blacks position is pathetic.

[Michael Sherwin]

Joel was awarded a win in one game, because of a programming bug in Rybk. So the match may have been 2.5 to 5.5. Concidering that Joel had an extra pawn in every game then this is a dismal score. If Joel would recieve the same pawn odds against Anand, Joel would win the match against Anand by about 6.0 to 2.0 and that is allowing room for some luck on Anand's part.

Not in such a miserable form.

If anyone by now does not understand just how far Rybka towers above even the strongest human player, then they never will.

OK, so I never will.
Joel had total winning positions several times in the match.
Rybka didn't even play very well. It could have played much better several times.

Was it Alekhine that indicated, that to beat him, a person will have to win in all phases of the game?

How much more true is this of Rybka?

How much effort did Joel expend to achieve these good positions and how much time did he leave himself on the clock to win them? How short of time was his opponent? Okay, with an extra pawn Joel was able to outplay Rybka in the openings, but, computers are not at their strongest in the opening with out a book. However, they do not tire as a result of expending energy in the openings like a human does. Then after Joel expends all that effort in the opening, Rybka which survived the opening then enters that part of the game that it excells at, requiring Joel to dig even deeper into his mental reserves just to stave of the monster. Out of a database of long time control strong human grandmaster games, how many mistake free games by the winner of a won game are there? Not many. Therefore it is the norm rather than the exception that the human will make mistakes, even if his position is better. To beat Rybka, Joel would have to beat Rybka in the opening and the middle game and the end game and with out making any mistakes!

[Dr.Ex]

Joel was awarded a win in one game, because of a programming bug in Rybk. So the match may have been 2.5 to 5.5. Concidering that Joel had an extra pawn in every game then this is a dismal score. If Joel would recieve the same pawn odds against Anand, Joel would win the match against Anand by about 6.0 to 2.0 and that is allowing room for some luck on Anand's part.

Not in such a miserable form.

If anyone by now does not understand just how far Rybka towers above even the strongest human player, then they never will.

OK, so I never will.
Joel had total winning positions several times in the match.
Rybka didn't even play very well. It could have played much better several times.

Was it Alekhine that indicated, that to beat him, a person will have to win in all phases of the game?

How much more true is this of Rybka?

A technical win is a technical win. Rybka is not a wizard. It doesn't even play a lost position particulary good, since it practically plays against itself in that phase of a game.

A chess engine is not programmed to seek practical chances against a human.

[/quote]
How much effort did Joel expend to achieve these good positions and how much time did he leave himself on the clock to win them? How short of time was his opponent? Okay, with an extra pawn Joel was able to outplay Rybka in the openings, but, computers are not at their strongest in the opening with out a book. However, they do not tire as a result of expending energy in the openings like a human does.
Then after Joel expends all that effort in the opening, Rybka which survived the opening then enters that part of the game that it excells at, requiring Joel to dig even deeper into his mental reserves just to stave of the monster.[/quote]

I've played several test games my own.
Rybka without b-,c- or f- pawn is rather a kitty cat than a monster.
I don't own a multiprocessor version though.

[/quote]
Out of a database of long time control strong human grandmaster games, how many mistake free games by the winner of a won game are there? Not many. Therefore it is the norm rather than the exception that the human will make mistakes, even if his position is better. To beat Rybka, Joel would have to beat Rybka in the opening and the middle game and the end game and with out making any mistakes![/quote]

No, the positions I posted didn't require anything special.
Black has no counterplay at all. Can't you see that?

[pichy]

If we take a look at the recent match Joel vs Rybka with a pawn Odd, and take a look at GM Benjamin rating which is no more than 2670 and wonder why he got 3.5 point out of 8 games 3.5 to 4.5, Now If instead of Rybka, Anand was playing versus Joel Benjamin with a Pawn Odd the result would have been less than 2.5 out of 8 games or 2.5 to 5.5.

First Joel Benjamin is rated 2576 FIDE ELO.

Second he got a score of 3.5 - 4.5, but this is not the real score that indicates the strength difference between him and Rybka in Pawn-odds of a to h Pawn, since he got a "free" win in first game. We have to ignore that single game, for comparing Rybka and Joel performance.


In the first game, Rybka had a bug, but vas and Joel decided to continue the game from move 11 to see how the game would ended, and in the exhibition game or what if then (withouth the bug) the game ended in a draw, therefore the match would had ended 3.0 to 5.0.


Third, the statement that Joel would get less or equal to 2.5 out of 8 games against Anand, is just an unfounded assumption. You just can't know about it.

This is NOT an assumption but any GM with a rating difference of 200 points higher should be able to handle a Pawn Odd better than any computer program, since when a program detect that it is losing it tries hard NOT to take any further chances, and a human would simply sacrifice another pawn to open the position up knowing that the skill and knowledge difference would compensate later on his favour.



So we can't really know about the validity of your following conclusion:

Now Why are computer less effective with pawn Odd playing versus human GM when the difference in rating is more than 200 rating points

[Dirt]

Jaan played much better than Joel, both more aggressive and with fewer mistakes. That shouldn't be much of a surprise considering their rating difference and Joel's recent lack of high level competition, but some thought Joel had a special understanding of computers that would make up for it. That did not appear to be the case.

If Jaan had had this handicap he might well have won.

[Uri Blass]

Joel was awarded a win in one game, because of a programming bug in Rybk. So the match may have been 2.5 to 5.5. Concidering that Joel had an extra pawn in every game then this is a dismal score. If Joel would recieve the same pawn odds against Anand, Joel would win the match against Anand by about 6.0 to 2.0 and that is allowing room for some luck on Anand's part.

Not in such a miserable form.

If anyone by now does not understand just how far Rybka towers above even the strongest human player, then they never will.

OK, so I never will.
Joel had total winning positions several times in the match.
Rybka didn't even play very well. It could have played much better several times.

Was it Alekhine that indicated, that to beat him, a person will have to win in all phases of the game?

How much more true is this of Rybka?

A technical win is a technical win. Rybka is not a wizard. It doesn't even play a lost position particulary good, since it practically plays against itself in that phase of a game.

A chess engine is not programmed to seek practical chances against a human.

I believe that
Rybka was clearly programmed to seek practical chances against humans
by some contempt factor that did not encourage her to trade pieces.

I also disagree that it is a technical win.
I agree that these are positions that the GM should try to win but it is not simple enough to call it technical win.

Uri

[pichy]

Joel was awarded a win in one game, because of a programming bug in Rybk. So the match may have been 2.5 to 5.5. Concidering that Joel had an extra pawn in every game then this is a dismal score. If Joel would recieve the same pawn odds against Anand, Joel would win the match against Anand by about 6.0 to 2.0 and that is allowing room for some luck on Anand's part.

Not in such a miserable form.

If anyone by now does not understand just how far Rybka towers above even the strongest human player, then they never will.

OK, so I never will.
Joel had total winning positions several times in the match.
Rybka didn't even play very well. It could have played much better several times.

Was it Alekhine that indicated, that to beat him, a person will have to win in all phases of the game?

How much more true is this of Rybka?

How much effort did Joel expend to achieve these good positions and how much time did he leave himself on the clock to win them? How short of time was his opponent? Okay, with an extra pawn Joel was able to outplay Rybka in the openings, but, computers are not at their strongest in the opening with out a book. However, they do not tire as a result of expending energy in the openings like a human does. Then after Joel expends all that effort in the opening, Rybka which survived the opening then enters that part of the game that it excells at, requiring Joel to dig even deeper into his mental reserves just to stave of the monster. Out of a database of long time control strong human grandmaster games, how many mistake free games by the winner of a won game are there? Not many. Therefore it is the norm rather than the exception that the human will make mistakes, even if his position is better. To beat Rybka, Joel would have to beat Rybka in the opening and the middle game and the end game and with out making any mistakes!

I totally agree with you, plus according to vasik even he admitted that Joel used various anti-computer tricks to improve his chances which he would NOT be able to use against Another Stronger GM.

http://rybkaforum.net/cgi-bin/rybkaforu ... l?tid=1917

"If we substitute the exhibition game for the unfortunate game 1, then we can observe that in six of the eight games, GM Benjamin was in total control from beginning to end. In a few of these games, he slightly increased his advantage (games 7 & 8). In other games, his advantage slighlty decreased (game 3). In none of these games was there ever a moment where Rybka had started to genuinely untangle herself and actively threatened to win - Rybka simply defended from beginning to end"

[Dr.Ex]

A chess engine is not programmed to seek practical chances against a human.
[/quote]

I believe that
Rybka was clearly programmed to seek practical chances against humans
by some contempt factor that did not encourage her to trade pieces.[/quote]

What I meant was not such a simplistic approach as high contempt factor
or not trading pieces. I thought about setting traps in a lost position.
Sometimes you should play for tricks and sometimes grim defence is your best chance, depending on the position and also on the opponent.
A conventional engine will never try any tricks, even the sophisticated ones that are hard to spot for humans.

[/quote]
I also disagree that it is a technical win.
I agree that these are positions that the GM should try to win but it is not simple enough to call it technical win.[/quote]

Depends on the player, I'm sure for Ivanchuk it is very simple to win this positions.