Humans Are No Match For Computers In Chess World
By SHELBY LYMAN
September 9, 2008
For the public at large, the issue of man vs. machine was settled in 1997 when Deep Blue defeated Garry Kasparov 3 1/2 to 2 1/2 .
But top chess players were not convinced. They argued that Kasparov had inexplicably collapsed in the final game after missing earlier opportunities to pile up a lead in the match.
Although IBM‘s decision to dismantle Deep Blue a year later was a setback for the development of chess computers, they continued steadily to improve.
In 2006, chess computer supremacy was firmly established when the program Deep Fritz defeated the World Champion Vladimir Kramnik by a score of 4-2.
A further blow to the human chess ego occurred in July, when RYBKA 3 defeated the American grandmaster Roman Dzindzichashvili 2 1/2 – 1 1/2 .
Although the event was followed on the Internet, it was sited at the Potomac, Md., home of Larry Kaufman, one of its programmers. The venue attests to the fact that the contretemps of man vs. machine now offers the prospect of the same drama for the public as would a race between a man and a bicycle.
The special significance of the event lay in the fact that the human had the advantage of first move and an extra pawn in each game.
“Dzindzi,” who was ranked 10th in the world in 1980, makes a specialty out of beating chess programs. But he could not take advantage of the substantial handicap he enjoyed.
Here is the full article.
Kasparov is the best ever.
I loved all those human vs. machine matches, especially those involving Kasparov.
But the inevitable has happened and non-handicap matches between humans and machines have become obsolete, no human can realistically compete with Rybka or Fritz.
I believe 3200 ELO for the best machines compared to 2800 ELO for the best humans emphasize this thesis.
no human can realistically compete with Rybka or Fritz
Well, Rybka still can’t play a decent rook ending. And in closed positions it’s still rather dumb.
I think the biggest problem is the competition format. A programm will never commit any tactical mistakes and will be as wide awake after six hours than at the beginning — in stark contrast to the human player. So one would need longer time controls, breaks and perhaps even adjournment.
Also Super-GMs need extensive special preparation for playing against the machine. Only for a prize money in the millions do they do it. After all, they normally make money by beatings other humans, not computers. Rybka so far has only beaten second rate GMs which are much cheaper.
I believe 3200 ELO for the best machines compared to 2800 ELO for the best humans emphasize this thesis.
These numbers are not comparable, because almost all of the games computers play are against other computers.
I still remember the know alls at my old chess club insisting that computers would “never play to grandmaster standard”. At that time they were at about 2100 elo. Well short of course but NOT a million miles away.
@rgorn
of course computers still have their weaknesses, chess is more than lightyears away from being solved. but those weaknesses are highly compensated by their strengths.
and of course, 3200 vs. 2800 ELO cannot be directly compared, but it gives you a clear hint on what’s going on.
Correspondence matches between grandmasters & computers?
Right now, i believe that if Kramnik or Anand devoted the same efforts to play against a computer as to play the world championship; they would still win.
But this kind of matches are only for the show; and players go for the money.
of course computers still have their weaknesses, chess is more than lightyears away from being solved. but those weaknesses are highly compensated by their strengths.
Which in my opinion are tactical infallibility (well, almost), never getting tired, never loosing concentration, not getting nervous in time trouble, and beeing completely unimpressionable.
As humans are indeed no match here, you have to somehow compensate for this, especially by not using a competition format that was designed for games among humans and that gives all the advantage to the computer in a game human vs computer.
3200 vs. 2800 ELO cannot be directly compared, but it gives you a clear hint on what’s going on.
These different ELO points are collected and redistributed in different pools. Therfore they give no hint in which pool the better players are.
There is no way to compare the capabilities of computers and human players.
They do not play each other enough for any comparison based on their games to be meaningful.
Even when the computers force a human slip, or just benefit from one, people will say ‘we cannot use that evidence of superiority: the human made a mistake!’ … curiously, a bit like that GM who said Aronian ‘just made a mistake’.
The fact is that people who are unwilling to play will be accused of being afraid of losing.
I infer that the best GMs are not now willing to play computers because they are (rightly) afraid of losing. But ELO 3200 for the top computers is way over the top.
No Blow to our ego!
This idea has always seemed silly to me. When the bicycle was invented did were runners upset over the fact that human ingenuity ad once again found a way to enhance human capacity? No, he just kept on running races against other humans. Would you expect a sprinter to compete in the Daytona 500? Because humans have built machines that can crunch numbers better than humans alone, means… nothing. What’s amazing to me is that it hasn’t happened before now. What human skill keeps he best of us competitive against sophisticated software running in multi-processor machines.
here are thousands… ok so it’s hundreds of thousands, of players who can beat me at chess without difficulty, but that doesn’t stop me from playing and trying to improve. The strength of computers is only important in how they can help me, a lowly human, play better chess.