Do we assume that Rybka is running on the same computer as the one which won the Computer World Championships? (2 x Intel Xeon 5500 series, 8 cores total, 16 cores virtual)
Rybka gains roughly 50 elo strength per doubling of processor cores.
Anyway, here's my guesses, though I am probably way too mediocre a chess player to be judging the greats!
Fischer -- 10-6-9;
Kasparov -- 10-5-10;
Karpov -- 10-4-9;
Morphy -- 10-5-6;
Carlsen -- 10-5-12;
Anand -- 10-4-11;
Do we assume that Rybka is running on the same computer as the one which won the Computer World Championships? (2 x Intel Xeon 5500 series, 8 cores total, 16 cores virtual)
Rybka gains roughly 50 elo strength per doubling of processor cores.
Anyway, here's my guesses, though I am probably way too mediocre a chess player to be judging the greats!
Fischer -- 10-6-9;
Kasparov -- 10-5-10;
Karpov -- 10-4-9;
Morphy -- 10-5-6;
Carlsen -- 10-5-12;
Anand -- 10-4-11;
Yes, I mean Rybka on the system which won the Computer World Championships.
Those look like good guesses, but I'm not sure whether or not people like Anand could put up fights like that against Rybka (11 draws in 25 games against the top computer in the world?). I would guess there would be a couple less draws in general at the lower levels (Anand would probably be more like 10-4-8).
Do we assume that Rybka is running on the same computer as the one which won the Computer World Championships? (2 x Intel Xeon 5500 series, 8 cores total, 16 cores virtual)
Rybka gains roughly 50 elo strength per doubling of processor cores.
Anyway, here's my guesses, though I am probably way too mediocre a chess player to be judging the greats!
Fischer -- 10-6-9;
Kasparov -- 10-5-10;
Karpov -- 10-4-9;
Morphy -- 10-5-6;
Carlsen -- 10-5-12;
Anand -- 10-4-11;
Wow, those are way too optimistic. Rybka 3 hasn't lost any games to humans yet in non-blitz games, and probably has an elo around 3200. My guesses would be, in a 14 game match:
Fischer -- 10-0-4;
Kasparov -- 10-0-4;
Karpov -- 11-0-3;
Morphy -- 14-0-0;
Carlsen -- 9-0-5;
Anand -- 10-0-4;
Kramnik -- 9-0-5.
Do we assume that Rybka is running on the same computer as the one which won the Computer World Championships? (2 x Intel Xeon 5500 series, 8 cores total, 16 cores virtual)
Rybka gains roughly 50 elo strength per doubling of processor cores.
Anyway, here's my guesses, though I am probably way too mediocre a chess player to be judging the greats!
Fischer -- 10-6-9;
Kasparov -- 10-5-10;
Karpov -- 10-4-9;
Morphy -- 10-5-6;
Carlsen -- 10-5-12;
Anand -- 10-4-11
Wow, those are way too optimistic. Rybka 3 hasn't lost any games to humans yet in non-blitz games, and probably has an elo around 3200. My guesses would be, in a 14 game match:
Fischer -- 10-0-4;
Kasparov -- 10-0-4;
Karpov -- 11-0-3;
Morphy -- 14-0-0;
Carlsen -- 9-0-5;
Anand -- 10-0-4;
Kramnik -- 9-0-5.
There are two things which you might be misunderstanding:
Oh I have missed the part about the match format. But those numbers were still way too optimistic because they suggest Morphy, who is at best at the level of a modern IM, has won several games against Rybka. Obviously he has almost no chance of winning, and drawing one game with white would be a huge accomplishment for him. The likes of Fischer, Kramnik, Anand etc on the other hand may sometimes hold with white, but still, they would have a very hard time drawing with black.
This is actually not that difficult to calculate. We can go by elo, or we can go by respective performances of Kramnik and Kasparov against various Fritz versions, Deep Blue etc (The fritz & deep blue that beat these two giants have no chance whatsoever against Rybka today), or even better, the latest human matches played by Rybka where it was comfortable giving exchange odds to a 2700+ GM. They all suggest an elite player would now have difficulty drawing against it and extremely low chances of winning a single game, and Morphy would probably not draw even one.
I expect rybka to play the most popular move as long as possible, then play the best move, if the GMs can exploit a line where the most popular move loses then they would have an advantage.
Oh I have missed the part about the match format. But those numbers were still way too optimistic because they suggest Morphy, who is at best at the level of a modern IM, has won several games against Rybka. Obviously he has almost no chance of winning, and drawing one game with white would be a huge accomplishment for him. The likes of Fischer, Kramnik, Anand etc on the other hand may sometimes hold with white, but still, they would have a very hard time drawing with black.
This is actually not that difficult to calculate. We can go by elo, or we can go by respective performances of Kramnik and Kasparov against various Fritz versions, Deep Blue etc (The fritz & deep blue that beat these two giants have no chance whatsoever against Rybka today), or even better, the latest human matches played by Rybka where it was comfortable giving exchange odds to a 2700+ GM. They all suggest an elite player would now have difficulty drawing against it and extremely low chances of winning a single game, and Morphy would probably not draw even one.
Perhaps you missed the part in my opening post where I said that all of these matches were making allowances for the era in which the player played. Hence, the accurate way to judge Morphy would be by his domination over his contemporaries. I have no doubt that Morphy would have little chance against any GM at all, much less Rybka, if his ability was unadjusted for the general increase in Chess ability since the 1850s. As a Champion, Morphy was far more dominant than most, and, if his skill is adjusted for the current era, he would outleague the likes of Anand. Many consider him the greatest player of all time (though not as many as Kasparov, Fischer and Karpov).
While the recent matches of Rybka are dominant, they are against stock GMs, not "elite" GMs. The domination of the likes of Fischer and Karpov during their very best tournaments is something that the Chess world has not been able to imagine coming from any of the contemporary GMs. Winning 2 games (losing the match 10-2 with plenty of draws) would be reasonable for the greatest tournament performance of all time by a human, in my opinion. Fischer, for instance, won multiple matches against players of levels higher than Anand by scores of 6-0 in his top prime. If Fischer in this state were to lose 10-0-4, then Rybka would barely ever even draw against humans (and when I say barely, we're talking at least 100 wins per draw against modern elite level GMs). Currently, I'm afraid Rybka is not yet quite there.
How reliable Rybka's Elo rating is ? Seeing as, it was formed from computer tournaments, and can we trust companies that produce computer programs to stipulate their ratings accurately.
come on!
Fischer -- 10-5-0;
Kasparov -- 10-6-0;
Karpov -- 10-3-0;
Morphy -- 10-3-0;
Carlsen -- 10-4-0;
Anand -- 10-5-0;
All this talk about Morphy makes me want to enter Capablanca during his "invincible" days. For a while, he was arguably the most dominant champion of them all.
In addition to talk of pure skill, you also have to talk about the tenor of the games involved. Personally, I'd be interested in seeing how Capablanca's pure simplicity, coupled with deep and comprehensive tactical awareness, would do against Rybka's pure brute force calculating power (especially in the endgames).
In addition, I think that the earlier players such as Morphy, Steinitz, Lasker, Capablanca, etc. will actually have better performances, assuming that you tweak Rybka's programming and preparation to reflect the theory and values of the times they played. I'm not sure how realistic this is from a programming perspective, but if you set Rybka to accept the same basic risk/reward assumptions concerning material sacrifices and the initiative that many of the Romantics possessed (complete with hasty and immediate sacrifices), and you might well see a win for the human player, just based on the positional breakthroughs and deeper understanding that those champions possessed. Afterall, what seperated Morphy and Steinitz from their peers was not their combinative prowess but rather their positional understanding, which was decades ahead of their peak performance period.
As I was stating, about Rybka vs. Morphy, this is actually an easier topic to discuss than, say, Morphy vs. Kasparov, which can only result in a crush. The reason is because you cannot seperate a Kasparov or an Anand from the chess time period from which they play. On the other hand, Rybka is, at its core, a computer program, and a calculating machine. Theoretically, and I admit that I don't know nearly enough about computer programing to support this assumption, but I'd think that that would mean one can realistically change Rybka's algorythms and programming concerning what constitutes good chess to the ideals of the Romantic school, which Morphy played against, with a prewritten emphasis on initiative over material, and a much more limited opening book centered upon wild, double edged openings covering only a handful of moves in depth.
In a sense, assuming my own preassumptions hold true, while it is problematic to seperate a Kasparov or an Anand from the time period that they play in, with a computer program that difficulty disappears, as a computer program, whether it be Fritz or Rybka or any other, is at its core, just a calculating machine.
How reliable Rybka's Elo rating is ? Seeing as, it was formed from computer tournaments, and can we trust companies that produce computer programs to stipulate their ratings accurately.
The ratings are given by independent testing sites such as CCRL or CEGT. They aren't "reliable" because the engines haven't played in human tournaments, but they are reliable I guess as a starting point for comparison, for instance 3200 for Rybka seems pretty reasonable to me.
I wonder how Petrosian would do.
Watch this start some kinda Rocky V sorta fight between kasparov and Rybka, that'd be awesome
pdela wrote:
come on!
Fischer -- 10-5-0;
Kasparov -- 10-6-0;
Karpov -- 10-3-0;
Morphy -- 10-3-0;
Carlsen -- 10-4-0;
Anand -- 10-5-0;
If the second number is the number of wins, then you clearly do not understand the normalities of top-level Chess. A draw against Rybka would be far more likely for most of those Grand-Masters to attain than a win. If the second number is the number of draws, that is more reasonable.
However, there is little chance that Kasparov will outperform Fischer. In his best days, Fischer was more dominant than any Chess player who ever lived (more so than Kasparov by a wide margin). It is like comparing Mike Tyson and Lennox Lewis. And on his best days, Karpov too was superior to Kasparov (just look at Linares 1994). These scores are not over large periods of time; they are with the Grand-Masters playing in their best ever form, down to the tournament or match. As for Morphy, he was far more dominant in his days than Anand could dream of being. We are also adjusting here for the era in which the GM played.
All this talk about Morphy makes me want to enter Capablanca during his "invincible" days. For a while, he was arguably the most dominant champion of them all.
In addition to talk of pure skill, you also have to talk about the tenor of the games involved. Personally, I'd be interested in seeing how Capablanca's pure simplicity, coupled with deep and comprehensive tactical awareness, would do against Rybka's pure brute force calculating power (especially in the endgames).
Go ahead. If you wish to do so and think he would do better than most of the others, go ahead and enter Capablanca into the tournaments. I would be inclined to think he would lose 10-3-6 or 10-3-7 (I wouldn't guess he would win any more than that, but that in itself would be a great achievement).
Gambitknight wrote:
In addition, I think that the earlier players such as Morphy, Steinitz, Lasker, Capablanca, etc. will actually have better performances, assuming that you tweak Rybka's programming and preparation to reflect the theory and values of the times they played. I'm not sure how realistic this is from a programming perspective, but if you set Rybka to accept the same basic risk/reward assumptions concerning material sacrifices and the initiative that many of the Romantics possessed (complete with hasty and immediate sacrifices), and you might well see a win for the human player, just based on the positional breakthroughs and deeper understanding that those champions possessed. Afterall, what seperated Morphy and Steinitz from their peers was not their combinative prowess but rather their positional understanding, which was decades ahead of their peak performance period.
No, the idea as regards adjusting for the era in which Morphy and Steinitz played does not mean to say that Rybka is simply a normal Chess player of those times with all the tendencies that normal players possessed, who simply had much greater calculating power. Actually, Rybka would be weakened in terms of calculating power to approximately the difference in standard between today's Top Grand-Masters and Rybka now. It would still not possess any of the tendencies that Chess players of those days had. And, of course, Rybka's opening book and endgame database would be shortened to those which were known in those days.
As I was stating, about Rybka vs. Morphy, this is actually an easier topic to discuss than, say, Morphy vs. Kasparov, which can only result in a crush. The reason is because you cannot seperate a Kasparov or an Anand from the chess time period from which they play. On the other hand, Rybka is, at its core, a computer program, and a calculating machine. Theoretically, and I admit that I don't know nearly enough about computer programing to support this assumption, but I'd think that that would mean one can realistically change Rybka's algorythms and programming concerning what constitutes good chess to the ideals of the Romantic school, which Morphy played against, with a prewritten emphasis on initiative over material, and a much more limited opening book centered upon wild, double edged openings covering only a handful of moves in depth.
In a sense, assuming my own preassumptions hold true, while it is problematic to seperate a Kasparov or an Anand from the time period that they play in, with a computer program that difficulty disappears, as a computer program, whether it be Fritz or Rybka or any other, is at its core, just a calculating machine.
There is no need to reduce Rybka to the inclinations of the Romantics. Simply delete the parts of its Opening and Endgame Databases which were not known in those days. Then, work out the difference in playing strength between Rybka and the Top Grand-Masters now (e.g. Kramnik, Anand, etc.), and then work out the strength of the Top Grand-Masters then, and add the difference. You will have a scaled down version of Rybka.
Bringing zombies back to life again...
Morphy will win his match, because we only see the flashy games, Fischer will probably win (both ajucated for expanded theory) I bet Kasparov will draw or win, but closely, Dissapointed no one considered Petrosian to join in because he would be a sure fire win. the rest will win some loose some
Here is the situation.
The latest and most powerful version of Rybka (which won the Computer World Championships) is to play a series of Grand-Masters from the past and present. The match format is as follows: both players in any given match have Classical time controls. The colors are switched every game regardless of the result, and there are no odds given of any kind. The winner is the first person to reach 10 wins, with draws not counting towards the match's conclusion, although they will count in the final scores. Scores are written as Number of Wins - Number of Losses - Number of Draws (e.g. 10-7-18).
Example: Rybka vs. Anand. If Rybka wins 10 games, loses 5 and draws 12, then the scores are written as 10-5-12.
Also, if the scores reach 9-9 (the number of draws makes no difference here), the winner is the first person to win 2 clear games ahead of the last person (e.g. 11-10 is not a final score, but 12-10 is, as is 15-13, etc.). If the scores reach 24-24, the person to win the next game is the winner (i.e. 25-24 is an acceptable final score).
----------------------------------------------------------------------------------------------
Rybka's first opponent is Bobby Fischer, in his top prime (the form he displayed at the tournament where he played his very best of all time - I'll leave this up to your judgement).
Its next opponent is Garry Kasparov, also in his top prime.
Then, Rybka can play Anatoly Karpov, Morphy (make allowances for the era in which Morphy played) and Magnus Carlsen (I would guess the Pearl Spring Tournament is his best performance, but it's up to you once again).
If you think any other players would do better against Rybka, feel free to post them and how you think the matches would turn out.