Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Correspondence Chess

Author: Peter Berger

Date: 08:59:32 05/23/03

Go up one level in this thread


..
"I tried to test this myself, since my rating hovers around ICCF-2500. So in
December 1999, I played four CC games versus the top rated chess engines of that
time, Fritz 6a and Nimzo 7.32. They were assisted by one of the fastest
computers of that day. I played without any computer aid."
..
"Indeed, as of the present time, I think that most human CC players do believe
that the very top CC players, playing without move generating chess engines, are
superior to the best chess engines in CC.
..
Since then, Uri Blass has won the Israeli CC championship using only chess
engines to generate his moves. Uri selected moves generated by a number of
different engines, so there was some indirect human interevention. Still, the
point being that the games were won strictly by chess engine moves. Then,
Gandolf 5.0 challenged 4 very strong CC masters (including Kenneth Frey, rated
around 2600, thus one of the highest rated CC players in the world) to a match
of 1-game apiece. Gandolf isn't even considered to be a top-rated engine. The
result - 4 long draws.
..
Conclusion: While insufficient evidence is available from which to draw
conclusions, I see no reason to believe that the best unaided humans are
demonstrably superior to chess engines in CC."

I agree to your conclusion, but somehow all the examples leave much room for
doubts.

a.) Steve Ham - Computer match

This is the strongest pro-human result from the sample - although the computers
were weakened in the beginning of the match since they were used in multi-PV
mode, the conditions were clearly very much in favour of the computers - a lot
of extra psychological pressure on you, very fast time control etc.
It leaves many doubts if the un-aided human would not have in fact be superior
to the un-aided computers if ICCF rules had been used.
Not only would you have had to face less pressure because of the slower time
control, but you might also have tried to aim against the specific weaknesses of
your opponents. In an ICCF tournament if you suspected your opponents were only
making computermoves you probably could take some advantage of this knowledge,
even if unaided yourself.

b.) Uri Blass - Israeli Championship

This one suffers from that we know very little about to what effect the
opponents used computers themselves. It is possible that everyone acted in a
similar way as Uri, but didn't tell, so it is possible that all used computers
to generate their moves, but Uri simply used them in a more clever way.

c.) The Rebel match

The strongest pro-computer result. Humans could use engines themselves, they
knew exactly what software was used against them, ICCF time control was used,
still they didn't win a single game - quite amazing.

I think that an un-aided human correspondence master will have huge difficulties
if he has to face an engine, but it's hard to tell how common un-aided humans
are in correspondence chess anyway. When the top players use engines themselves
to help them generate their moves (some call it blunder-check I assume ;) )  the
more interesting question is how much they can improve on pure engine play. I
think the setup Uri used is interesting here - he didn't do much more than
choose between suggestions of different engines, but it is likely that this
already can lead to a decent improvement.
Problem is that probably few correspondence players will "tell it like it is" -
if they use engines, and how. So this is mostly guesswork. One hint could be the
amazing draw ratio in recent top correspondence tournaments, but it is hard to
tell. If someone asked three engines to suggest a move, and then he decides to
play something else - didn't the engines help him tremendously to reach this
decision anyway?

Peter



This page took 0 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.