Author: Jorge Pichard
Date: 15:09:52 09/02/02
Simply because Kramnik can prepare with the exact version of Deep Fritz he will
play in the match, while Deep Junior's team is free to make changes up till the
clock is started.
We are a month away from the dueling K vs. Machine matches in the Middle East.
Kasparov leads off against Deep Junior in Jerusalem in a six-game match, and a
few days later Kramnik faces Deep Fritz in Bahrain after a year's worth of
postponements. In my informal e-mail poll the clear consensus was that the
machines are going to be crushed. In fact, there was more contention about
whether or not the computers would win a single game.
This may seem strange at first, considering how well these programs have done
against top GM competition in recent events, but I must agree with these
conclusions for a number of reasons.
1) It's Kramnik and Kasparov. World Champions and the only players ever to break
the 2800 Elo mark, they would be favorites even without the other items on this
list.
2) GM preparation + program availability. A GM's preparation is usually somewhat
canceled out by the other GM's preparation. This is also true in human-machine,
and the programs' GM-prepared opening books will be ready for the K's. But the
machine teams will have a harder time changing how their programs play, and
since both are commercially available we can expect the K's to know these
programs inside and out by match time.
3) Anti-computer chess. This was a relatively unexplored concept back when
Kasparov lost the second match against Deep Blue in 1997. The top players were
used to being able to beat computers without any special strategies and even
today they are a little embarrassed to skip the best move for one that is best
only against a computer. In 1997, Kasparov's idea of anti-computer chess was to
play lousy openings to get Deep Blue out of its book. These days the players
know that if they can blockade the center ("eight-pawn chess") they'll have all
day to set up a smashing attack on the computer's king. This is harder than it
sounds, especially with black, but with enough practice a draw is the worst you
have to fear. In several games in the past few years GMs using this strategy
have made both Fritz and Junior look pathetic. (Yes, the computers have made the
GMs look pathetic on a few occasions too.)
4) Sheer power. Despite the lack of empirical evidence about its play, the fact
remains that Deep Blue was many times faster than these micros. Even with its
parallel architecture and searching up to a billion positions per second Deep
Blue only edged an unrecognizable Kasparov, who resigned once in a drawn
position and lost the final game with a bizarre opening blunder.
Of the two matches, Kasparov-Junior seems to be the more attractive from a chess
perspective, but this is not to Kasparov's credit. It's more attractive only
because I can imagine Kasparov losing! Stylistically he is the ideal opponent
for a computer team to face. The tactical style we mentioned above plays to the
computer's strength and his legendary opening preparation can be partially
nullified by the giant database Junior can access. (Novelties can be even more
effective, however.) Kasparov is also a proponent of OTB chess psychology, which
is quite ineffective against a machine.
If computers have nightmares when in sleep mode, they have nightmares about
playing Vladimir Kramnik. In addition to the obvious world-champion-2800
attributes, he's solid, patient, and practical about doing what it takes to win.
Big Vlad will be content to maneuver until he finds a weakness and he's
sufficiently pragmatic to take a draw if he can't. Plus, he's big enough to
chuck the damn machine out the window if he gets into trouble. It's also
significant that Kramnik can prepare with the exact version of Deep Fritz he
will play in the match, while Deep Junior's team is free to make changes up till
the clock is started.
Kasparov's opponent, Ban & Bushinsky's Junior 7, has a very aggressive and
sacrificial style that is ideal for maximizing winning chances against a human.
The performance of Morsch's Fritz 7 has shown it to be as strong, but more
materialistic and "computer-like" in its play. The keys for both machines are to
keep things open at all costs and complicate at every turn. What we know for
sure is that neither program will miss the smallest tactical opportunity to whip
humanity's collective butt. It will be very difficult for the K's to avoid such
a slip over 14 games (six for Kasparov, eight for Kramnik). In a few days we'll
hear dozens of opinions from readers and experts
This page took 0.01 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.