COMPUTER AT ODDS WITH BOOKED-UP FOE
By ANDY SOLTIS
July 22, 2007 — CHESS THE strange history of man-vs.-machine matches took a new turn earlier this month when Jaan Ehlvest tried for revenge against Rybka but received a strange type of odds.
Since Garry Kasparov was humiliated by Deep Blue 10 years ago, organizers of computer matches have tried to “level the playing field” by eliminating a machine’s allegedly unfair advantages, such as its opening “book” knowledge, which often runs more than 25 moves deep.
But the machines kept winning. So, in the first Ehlvest-vs.-Rybka match in March, the world champion gave odds – one pawn in each game. It won 51/2-21/2.
Fans of humanity found an excuse: Ehlvest, once the world’s fifth-highest rated human, is used to non-odds games, they said. He was the one at a handicap because he was forced into wholly unfamiliar positions against Rybka (!).
So in Round 2, Ehlvest got new advantages. He played White in every game and had twice as much time as the machine. Most important of all, Rybka’s opening “book” beyond the third move was wiped out.
In this week’s game, for example, the machine was on its own after 3 . . . Qd6. There are virtual beginners who have more opening knowledge than this.
It didn’t matter. Rybka simply outplayed Ehlvest again in middlegames and finished off 41/2-11/2.
Source: NY Post
Humans cannot defeat the best computer chess programs. Period. I think it’s rather silly to give human players odds, extra time, or taking away a lot of Rybka’s opening book just because it is better than the best human chess players.
Analogy: It’s the same as taking trying to fix a car to go no faster than 5 mph just to give a human runner a better chance of out performing the machine. It’s silly!
Just face the fact that computers are better and get on with life. It’s so funny to hear GM’s talk about how computers have “weaknesses” (like problems with long term planning), yet can never defeat them!
Why not just make it all even more silly and give the human 5 Queens and then just give Rybka the King and pawns. Hey…at least then somebody could say “I beat Rybka.”
These matches are just silly because the best GM’s (as was posted on the topic of checkers) are just too egotistical to face the fact that they cannot defeat a chess program.
Since Checkers has finally been solved and computers have been able to defeat their top GM’s for well over a decade, I followed the game for years and have never heard of a top checker GM wanting “revenge” against the computer and playing a match in which it was either an odds game or the computer had its opening book taken away.
They always played it with no assistance. In other words, they’ve accepted that computers are far superior and this begs the question:
Why can’t human Chess GM’s just admit that they cannot defeat and will soon never even draw against the best chess programs?
Ego. Just as the Checker Super-GM Don Lafferty once stated, “I never met a checkers player I didn’t like; they’re all even-tempered,” he says. “Chess players are egotistical. They think they’re intellectuals and that everyone else is beneath them. Checkers players aren’t like that…”
Since no human chess GM can defeat Rybka in a match on even terms, then why can’t they just admit it?
This, I think, is a most interesting psychologica question.
Does anybody knows how Rybka or any other software performs in the Fischer Random… Maybe its time to move forward and accept that Fischer vision of chess was espectacular… and play no more “classical” chess against computers
I have a feeling Rybka will beat anybody in Fischer Random Chess too! Rybka is just stronger than any human player.
We will soon know how Rybka performs in Fischer Random, or Chess960: Among the events organised by the Chess Tigers (Mainz), there will be a computer chess Chess960 tournament again:
http://chesstigers.de/ccm7/computer_wm.php?lang=1&kat=6&subkat=1
(August 14th-16th)
Spike and Shredder are the Chess960 computer world champions of 2005 and 2006.
A Chess960 capable Rybka has not been released yet. For the relative performance of other engines, see
http://computerchess.org.uk/ccrl/404FRC/
[1]
OLD QUESTION: Are humans stronger chess players than computers?
NEW QUESTION: Is a team of one human + one computer stronger than a computer alone?
One theory says “Yes”. The reasoning is that computers often win by avoiding errors. Yet it might be that the very best human moves are occasionally better than the move the computer would make. For example, the human might know the kind of endgame a variation might lead to, beyond the computer horizon.
[2]
(To anon 9:57)
Rybka slaughtered Ehlvest in traditional chess, even after Rybka’s opening book was almost entirely removed.
Ehlvest got to keep his memorized opening knowledge, which gave him an advantage.
Therefore I trust that Rybka would have an even bigger margin of victory against a human – if they were playing chess960 (FRC). There Ehlvest would lose his advantage of memorized opening knowledge.
GeneM
http://CastleLong.com/
To the 2nd anonymous:
I believe it’s a general consensus that (best) computer programs plays now better chess and I’m not sure if any GM is saying otherwise.
But I think it’s the PR of the event that requires to attempt to even the chances of the players. Why would people follow a match when a winner is obvious? Why would sponsor give any money?
the likes of kramnik and anand will beat rybka comfortably
elhvest is only an average player
There is absolutely no way that Kramnik or Anand (or even if they played as a team) could defeat Rybka in a match! As a team…they might, just might manage draws but I really doubt it.
However, neither will play Rybka for one reason: They wouldn’t stand a chance.
If Kramnik cannot defeat Fritz, which is an inferior program than Rybka, it is completely illogical to say he could defeat Rybka.
You can believe this if you wish. Some people believe the world is flat and that’s just fine. Yet, “belief” isn’t the same as “truth” or “facts.”
The truth and the fact is: Neither Kramnik nor Anand could ever defeat Rybka in a match…and I highly doubt even in a single game.
I’m glad a serious player is playing a machine without an opening book.
I’ve long held that the opening books aren’t that crucial to computer chess, but a lot of people don’t believe me.
I noted fairly weak computers will happily find 3… a6 in the Ruy Lopez at 15 seconds a move, which whilst it is a very logical move for black took humans a fair while to settle upon.
That said there are some “deep” opening traps, and I suspect that a GM might win the odd game by exploiting such if he can steer the game that way.
Ehlvest is not an “average player”, he is a Top-100 Grandmaster! I watched almost all games of that match live. I actually was impressed how strong he is. He just couldn’t keep that high level constantly, throughout a whole game. But he scored 3 draws (in 6 games). At least once, he achieved a winning position. The result alone doesn’t give a good impression of his play.
Giving the human twice as much time is stil ridiculous!
Give the computer only 15 minutes compared to the human a few hours and it might get even…
hehe, where are the times, when the comps were given the odds of using the most state of the art computing machines… now a regular off the shelf note-book will rip a GM apart if it is not dumbed down. …and even then it will..
I will give Ehlvest credit for one thing: He is the only top GM that will publicly play Rybka and, seemingly, not fret too much about the possibility of defeat. Kramnik cannot defeat even Fritz.
Why is it that the WCC and the #1 or #2 players in the world will not play the World Computer Champion? Everyone knows that these guys have Rybka at home and use it for study.
One reason is such a match, on even terms, would be a complete joke. No human player can defeat Rybka in a match…and probably not in a single game. Or, why not have a consultation game (or match) with one move per day, such as the “Kasparov vs the World” game? This way the best human players could consult each other.
However, this would require cooperation and considerting that Kramnik and Topalov completely hate each other…it wouldn’t happen.
Even if it did, Rybka would win unless some silly rule was in place to try and “even” the playing field.
Logically, considering that computer chess programs have been at the super GM level for over 5 years, it seems that the best chess players would show some improvement in their play due to the new theory that has come from rapid computer advances. I just don’t see it.
It’s complete folly and a patent absurdity to speak of a chess program’s “weaknesses” from those that can never defeat them. It’s akin to a club player describing a GM’s “weaknesses” even though he/she can never defeat the GM!
Since checkers has finally been solved, and considering the fast-paced development of computer technology, I certainly believe it is possible that chess, too, can be solved within, say, the next 100 years.
Today, we have absolutely no idea what breakthroughs in computer technology might occur in 75-100 years. I’m sure in the 1920, no one could have imagined the computers of today. So, it is quite feasible that chess will be solved within the next 100 years.
we have to accept the power of computers in chess. Making computers odds is not fine. Top grandmasters are being defeated by computers,from this we must understand that secret of chess is so deep,we have to improove in tactics.