Man Vs Machine In Chess - Who Is Winning?

FFriedel writes "In a few weeks, the world's strongest player Garry Kasparov will take on X3D Fritz in a high-profile man-machine chess match. Who is the statistical favourite? Since computers have been steadily improving and are now holding their own against the very strongest human players, one would think it may be Fritz. Not necessarily, says statistician Jeff Sonas, who doesn't believe computers will inevitably surpass the top humans, and presents empirical evidence to support his claim as part of a series of articles for ChessBase."

Comments

Re:Infinite Chess

[jbellis]

read rec.games.chess.computer. the search space is WAY the hell too big.

Re:slashdot == sexist

[gnoblins]

Of the Top 50 female chess players, only six have reached the male grandmaster level, and only Judit Polgar has reached super GM level (>2700) which would be sufficient to challenge one of these machines.

Re:a contradiction?

[rakeswell]

No, it's not necessarily a contradiction. Simplest example: Players A and B have the same rating and win, lose and draw an equal number of games. Let's also say that both of their games (technique) is improving at an equal rate. Therefore, their ratings do not change, even though their play has improved. That's the tough thing about chess. You can play a very deep game and still lose. You don't get credit for having played well -- it all comes down to mate or a draw.

Re:Man v. machine? MACHINE.

[phr1]

Nah. Kasparov didn't get his ass kicked in '96, he WON that match, 3.5-2.5. He lost the '97 match by the same score. However, the '97 loss wasn't all that convincing. He certainly would have won a longer match, or a rematch against the same hardware and software. On the other hand, if a rematch had happened, it would not have been against an identical Deep Blue 2. The designers would have kept making improvements and speedups and gotten an even stronger machine, that might well have been convincingly stronger than Kasparov.

I think it will be several years before anyone builds anything as strong as Deep Blue II. At that point, the top grandmasters will have something to worry about. For now, the claims that PC programs like Fritz are as strong as DB2 was are mostly marketing hype.

For more info, see Deep Blue designer Feng-Hsiung Hsu's book "Building Deep Blue", about the work that went into the machine and how the Kasparov matches went.

Re:Sonas argument is silly

[rifftide]

When Deep Blue beat Kasparov, the computer had an unfair advantage (besides being a computer): it knew all about Kasparov's past games and tendencies, but not vice versa. Kasparov improved his performance in subsequent matches by insisting on having access to the computer in advance to play many test games. Under these more balanced rules the top human players are in rough parity with the computer even though the machines' raw performance has continued to increase according to Moore's law, or better.

AI researchers originally had high hopes of using chess as a practical test for machine intelligence. But AFAIK the current generation of chess playing machines rely mainly on brute force calculation, along with a substantial repertoire of "book" moves. Since the number of possible chess moves from a given position is subject to combinatorial explosion, improvement based on increasing dosages of brute force will eventually slow to a crawl. Chess masters aren't increasing their endowment of brain cells, but they can think intuitively and spatially and are steadily learning how to play more effectively against the machines. So I think the author's point is that grandmasters are likely to soon gain the upper hand, and will maintain this lead until programmers successfully implement more sophisticated approaches.

Re:Kasparov is a bad choice

[aurelian]

Chess is boxing with the mind. Also, if you make no mistakes, it's a draw. As if that ever happens...

Not necessarily; as far as I'm aware it's not yet known whether chess is a draw, a white win or a black win.

Re:Kasparov is a bad choice

[GlassHeart]

the game is initially equal.

No, somebody gets the first move. I don't believe it has been proven whether this is necessarily an advantage, disadvantage, or absolutely not a factor.

Yawn.

[TheLink]

Uh. Not worth a read, a waste of time. Summary: author says God cannot be omnipotent and infallible if humans have free will- coz if God knows what we will do, we don't have free will, and since we have free will (author doesn't prove that convincingly either) God is fallible etc etc. The author also follows with some insulting remarks too.

Philosophers and other people have done arguments like that or better, far more efficiently and elegantly - e.g. "can God create a rock he cannot lift" and so on. Some have managed to do so without the trolling and insults too.

The universe is more than what we understand so far, and this guy thinks it's so simple?

Explain this (and I mean thoroughly):
http://www.wikipedia.org/wiki/Double -slit_experime nt

And also explain the first observation every scientist makes - self awareness.

What if God lends/puts a bit of himself to/into each human?

Or arranges it so that if you, a glimmer of light, choose accordingly, you end up with constructive interference - light. But if you choose otherwise, you end up in destructive interference and darkness.

It's probably not quite as simple as that either but I'm willing to bet the universe isn't as narrow as those popular philosophical arguments seem to assume.

If it were that simple, why is there a "you" or "I"?