Sixth Harvard Cup UPDATE
Christopher F. Chabris
SIXTH HARVARD CUP HUMAN VERSUS COMPUTER CHESS CHALLENGE
=======================================================
Wednesday-Thursday, 27-28 December 1995
Manhattan Conference Center
Fiterman Hall, 30 West Broadway (14th floor), New York, NY
The Sixth Harvard Cup will feature six of America's top grandmasters against
six of the leading microcomputer-based chess software packages in a team
match format. Each human will play each computer once, at a time control
of game in 25 minutes. There will be 36 games in all (five rounds of four
games on the first day, four rounds of four games on the second
day), beginning at 5:00 PM each day. Demonstration boards and expert
commentary will be provided, software will be exhibited, and T-shirts and
chess books and equipment will be on sale.
Admission for spectators is FREE and includes a copy of the official
program book.
The following teams will participate:
Humans:
Joel Benjamin (two-time defending Harvard Cup Champion)
Boris Gulko (1994 U.S. Champion and world championship candidate)
Ilya Gurevich (former World Junior Champion)
Gregory Kaidanov (in his first Harvard Cup)
Michael Rohde (in his sixth Harvard Cup)
Patrick Wolff (1992 and 1995 U.S. Champion)
Computers:
Chessmaster 5000 (Mindscape)
Junior (Plastronics Interactive Multimedia)
M-Chess Pro (Marty Hirsch; current world microcomputer champion)
Socrates 95 (Kaufman and Dailey; three-time Harvard Cup Champion)
Virtual Chess (I-Motion Interactive)
WChess (David Kittinger; defending Harvard Cup Champion)
All programs will run on the new 150 MHz Intel Pentium Pro processor (in
its international computer chess debut). Last year the computer team
scored 39% on 90 MHz Pentium processors -- will they pass 50% for the
first time this year?
The Harvard Cup Youth Challenge for students ages 5-18 will run each day
from 9:30 AM to 4:30 PM. Participants may register between 9:00 AM and
2:00 PM each day. Activities will include a series of challenge games
against Chessmaster 5000 for prizes, chess lessons from top teachers, and
demonstrations and workshops on how computers play chess.
Complete information is available via the World Wide Web at H3 Online:
http://www.h3.org/h3
Or you may send email to:
harva...@h3.org
Live coverage of the Harvard Cup will be provided by the Internet Chess Club:
http://www.hydra.com/icc
The Sixth Harvard Cup is presented by Mindscape, Inc. and sponsored by
Intel Corporation, I-Motion Interactive, Corel Corporation, IHP Inc.,
M Chess, Plastronics Interactive Multimedia, the Millburn Corporation,
Borough of Manhattan Community College, the U.S. Chess Federation, the
Internet Chess Club, Cybersmith, and The Computer Museum. The Harvard
Cup is produced by H3 in conjunction with the Harvard Chess Club and the
Harvard Computer Society. The Harvard Cup is being held this year in
conjunction with the Pan American Intercollegiate and Scholastic Team
Chess Championships, 26-29 December 1995 at the same location (for more
information, see http://www.redweb.com/panam).
--
======================
Christopher F. Chabris (c...@wjh.harvard.edu)
P.O. Box 382967, Harvard Square Station, Cambridge, MA 02238-2967 USA
Tim Nunnally
that the relatively short time control of G/25 is less than ideal.
The computer processors get faster each year, but the human
brain (overall) doesn't. Thus, although the chess engines may
not improve much, they will perform better simply because they
are running on faster computers. I think it is true that
all computer programs perform progressively worse as the time
control lengthens, which makes sense because the faster time
controls are more condusive to the very fast "thinking" the computer
engages in. But as the time contol lengthens, the computer
doesn't improve as much in its anaysis as does the human.
If the point of the Harvard Cup is to see if humans or computers
are playing better chess (and I know that is a simplification
of what the Cup is all about), then why not use a time control
that is closer to those used in high-level humans-only tournaments
and matches? I'm not really concerned that they (computers) will
one day surpass us ,(Well, I shouldn't include myself --
computers have long surpassed that milestone), but I would'
like to see a more accurate measure of human vs. computer ability
than G/25 seems to provide.
Tim Nunnally, Tuscaloosa, Alabama USA
Halibut
: Computers:
: Chessmaster 5000 (Mindscape)
: Junior (Plastronics Interactive Multimedia)
: M-Chess Pro (Marty Hirsch; current world microcomputer champion)
: Socrates 95 (Kaufman and Dailey; three-time Harvard Cup Champion)
: Virtual Chess (I-Motion Interactive)
: WChess (David Kittinger; defending Harvard Cup Champion)
: All programs will run on the new 150 MHz Intel Pentium Pro processor (in
: its international computer chess debut). Last year the computer team
: scored 39% on 90 MHz Pentium processors -- will they pass 50% for the
: first time this year?
First, it is certainly nice of y'all to provide this chance to
watch strong players competing against computers.
However, since it is a virtual certainty that few of these
programs have been optimized for the P6 "Pentium Pro" and since it has
been clearly shown that the P6 at 150Mhz will actually run regular non-32
bit code >>>>SLOWER<<< than a Pentium 133Mhz, and closer to a Pentium
100MHZ, shouldn't they be running these on Pentium 133's?
(This is a rhetorical question, since obviously the point is to
promote Intel's latest generation, but for people who know anything about
computers and read any of the current magazines [BYTE had an article on
this a month or two ago] this spoils it somewhat)
Dave Gomboc
> However, since it is a virtual certainty that few of these
>programs have been optimized for the P6 "Pentium Pro" and since it has
>been clearly shown that the P6 at 150Mhz will actually run regular non-32
>bit code >>>>SLOWER<<< than a Pentium 133Mhz, and closer to a Pentium
>100MHZ, shouldn't they be running these on Pentium 133's?
>
> (This is a rhetorical question, since obviously the point is to
>promote Intel's latest generation, but for people who know anything about
>computers and read any of the current magazines [BYTE had an article on
>this a month or two ago] this spoils it somewhat)
While they may not be 'optimized' for the Pentium Pro, any code optimized
for the Pentium is likely to be 32-bit already. The C programmer can
click a couple of buttons and recompile. The assembly programmer has
probably been writing 32-bit code for a while already anyway, if not
because it was already quicker, then in anticipation that it would be
when the Pentium Pro was released.
What they should be running them on is the Pentium Pro 200s. :-)
--
Dave Gomboc
drgo...@a.stu.athabasca.ca
Mark Schreiber
: The Sixth Harvard Cup will feature six of America's top grandmasters against
: six of the leading microcomputer-based chess software packages in a team
: match format. Each human will play each computer once, at a time control
: of game in 25 minutes.
Computers playing at these fast time controls have too much of an advantage.
Even Kasparov lost to a computer at these fast time controls.
Longer time controls would be a fairer challenge.
Halibut
: While they may not be 'optimized' for the Pentium Pro, any code optimized
: for the Pentium is likely to be 32-bit already. The C programmer can
: click a couple of buttons and recompile. The assembly programmer has
: probably been writing 32-bit code for a while already anyway, if not
: because it was already quicker, then in anticipation that it would be
: when the Pentium Pro was released.
HEck, given the choice of optimizing them for Pentium, or
spending a little more time bashing the Genius book, which do you think
Marty chose? :)
: What they should be running them on is the Pentium Pro 200s. :-)
This is a good question. A Penmtium Pro 200 should be as fast as
a Pentium 133 on non-optimized code. And they are already commercially
available. Couldn't Intel come up with a few units?
Eric Peterson
The computers scored only 39% in last year's Harvard Cup, so it seems to
be a fair matchup, no?
__________________________________________________________________________
Eric Peterson
USCF Life Master
etpe...@netcom.com
Joe Stella
msch...@cnct.com (Mark Schreiber) writes:
>Computers playing at these fast time controls have too much of an advantage.
>Even Kasparov lost to a computer at these fast time controls.
>Longer time controls would be a fairer challenge.
I tend to think that they use these time controls for many reasons, not
the least of which is that it is easier (and cheaper for the sponsor) to
organize a match that will not last too long.
Everyone who follows computer chess knows that computers have more of an
advantage at fast time controls than they do at slower ones. The
Harvard Cup is useful to see how much the computers have improved,
and how computers fare against humans at *action* chess. No one
(at least no one who knows anything about chess and computers) will
say that computers are better players than humans just because the
computers have won one action chess match.
Joe S.
Brian Karen
>>: of game in 25 minutes.
>>
advantage.
>>Even Kasparov lost to a computer at these fast time controls.
>>Longer time controls would be a fairer challenge.
>
to
>be a fair matchup, no?
>
One of the reasons for having the Harvard Cup is to see how the
computers perform relative to the previous Harvard Cups. Therefore, the
time control must be the same as in previous years.
Naisortep/Brian Karen
Hal Bogner
>Christopher F. Chabris (c...@wjh.harvard.edu) wrote:
>: SIXTH HARVARD CUP HUMAN VERSUS COMPUTER CHESS CHALLENGE
>: The Sixth Harvard Cup will feature six of America's top grandmasters against
>: six of the leading microcomputer-based chess software packages in a team
>: of game in 25 minutes.
>
>Computers playing at these fast time controls have too much of an advantage.
>Even Kasparov lost to a computer at these fast time controls.
>Longer time controls would be a fairer challenge.
Yes. And there is a much more meaningful man vs. machine competition staged
every year in Holland: the Aegon tournament. More games, 2 hours per side,
one game a day, and many more participants on both sides.
The Harvard Cup has several things that are open to criticism:
- As noted above, the games are very fast, which increases the odds of a
computer win, and the GMs usually do worse as the day wears on, as well.
- There is a high fee to enter a computer, so basically this is a promotional
opportunity with a high likelihood of at least one win over a GM being sold as
though it was a definitive test.
-hal bogner
Hal Bogner
>In article <4bdbj8$p...@cnct.com> msch...@cnct.com (Mark Schreiber) writes:
>>Christopher F. Chabris (c...@wjh.harvard.edu) wrote:
>>: SIXTH HARVARD CUP HUMAN VERSUS COMPUTER CHESS CHALLENGE
>>: The Sixth Harvard Cup will feature six of America's top grandmasters against
>>: six of the leading microcomputer-based chess software packages in a team
>>: match format. Each human will play each computer once, at a time control
>>: of game in 25 minutes.
>>
>>Computers playing at these fast time controls have too much of an advantage.
>>Even Kasparov lost to a computer at these fast time controls.
>>Longer time controls would be a fairer challenge.
>
>be a fair matchup, no?
1. Who is to say what is fair?
2. The humans did better last year than the year before. They did not have
to play as many games in a row with very short breaks - an improvement that
reduced the handicap of tiredness somewhat.
Halibut
:
: - There is a high fee to enter a computer, so basically this is a promotional
: though it was a definitive test.
Remember the advertisement in the back of Chess Life about 10
years ago, showing some fish computer's win against "Richard Verber, a
Chicago master", implying that the computer was of master strength at least?
Ariel Mazzarelli
>In article <4bdbj8$p...@cnct.com> msch...@cnct.com (Mark Schreiber) writes:
>>
>>Computers playing at these fast time controls have too much of an advantage.
>>Even Kasparov lost to a computer at these fast time controls.
>>Longer time controls would be a fairer challenge.
>The computers scored only 39% in last year's Harvard Cup, so it seems to
>be a fair matchup, no?
If a matchup were arranged whereby a player like myself could score 39%
against GMs like Gulko, would you thus also assume that it was a fair matchup?
Let's the the ball rolling here, I could use the rating points.
Ariel
David Gertler
: Remember the advertisement in the back of Chess Life about 10
: Chicago master", implying that the computer was of master strength at least?
Seems to me the computer (an early Chess Challenger) lost that game;
they put it there just to show that he hadn't mated it in twelve
moves, I suppose.
By the way: I think that ad was actually 13-14 years ago.
(I was the chess computer specialist at USCF 11 years ago, and
computers were already playing a lot better than that.)
Mark Schreiber
: msch...@cnct.com (Mark Schreiber) writes:
: >Computers playing at these fast time controls have too much of an advantage.
: >Even Kasparov lost to a computer at these fast time controls.
: >Longer time controls would be a fairer challenge.
: I tend to think that they use these time controls for many reasons, not
: the least of which is that it is easier (and cheaper for the sponsor) to
: organize a match that will not last too long.
Easier, cheaper and also a mediocre test of chess playing ability.
Its not hard to have a tournament with longer time controls,
Aegon does it every year.
: Everyone who follows computer chess knows that computers have more of an
: advantage at fast time controls than they do at slower ones. The
: Harvard Cup is useful to see how much the computers have improved,
: and how computers fare against humans at *action* chess.
As a long term barometer,
action chess has just about reached its limit.
In the last Harvard Cup, Wchess had no losses.
It scored higher than any human.
Wchess had 83% vs Joel Benjamin at 81%.
: No one (at least no one who knows anything about chess and computers) will
: say that computers are better players than humans just because the
: computers have won one action chess match.
: Joe S.
But if a computer had won at longer time controls,
then we might say computers play better.
But we won't know unless they play at longer time controls.
Computers do well at longer time controls.
Hiarcs beat Yasser Seirawan in the Aegon tournament.
Mark Schreiber
Mark Schreiber
: In article <4bdbj8$p...@cnct.com> msch...@cnct.com (Mark Schreiber) writes:
: >: SIXTH HARVARD CUP HUMAN VERSUS COMPUTER CHESS CHALLENGE
: >: The Sixth Harvard Cup will feature six of America's top grandmasters against
: >: six of the leading microcomputer-based chess software packages in a team
: >: match format. Each human will play each computer once, at a time control
: >: of game in 25 minutes.
: >Computers playing at these fast time controls have too much of an advantage.
: >Even Kasparov lost to a computer at these fast time controls.
: >Longer time controls would be a fairer challenge.
: The computers scored only 39% in last year's Harvard Cup, so it seems to
: be a fair matchup, no?
: __________________________________________________________________________
: Eric Peterson
: USCF Life Master
: etpe...@netcom.com
No not really, because you are averaging
strong and weak programs together.
The range is wide, over 100 points.
The best individual score in the
94 tournament was Wchess with no losses.
It scored 83% vs Joel Benjamin with 81% score.
So a computer actually was the tournament winner.
The 39% average includes wchess with zarkov.
zarkov only had 1 win with a score of 17%.
zarkov had a tournament performance rating
of 500 points less than wchess.
If you want to know how computers improve in relation to humans,
then compare the best computer performance last year
to previous years best computer performance.
The best computer performance in the 93 tournament was 50%.
So from the 93 tournament to the 94 tournament,
performance of the best computer increased from 50% to 83%.
The tournament performance rating of the best computer
in the 1993 tournament was 2588. In the 1994 it was 2895.
An improvement of 307 points from 1993 to 1994.
The 39% average is not very useful.
Mark Schreiber
Mark Schreiber
: Each human will play each computer once, at a time control
: >>: of game in 25 minutes.
: >>
: >>Computers playing at these fast time controls have too much of an
: advantage.
: >>Even Kasparov lost to a computer at these fast time controls.
: >>Longer time controls would be a fairer challenge.
: >
: >The computers scored only 39% in last year's Harvard Cup, so it seems
: to
: >be a fair matchup, no?
: >
: One of the reasons for having the Harvard Cup is to see how the
: computers perform relative to the previous Harvard Cups. Therefore, the
: time control must be the same as in previous years.
: Naisortep/Brian Karen
Then the Harvard Cup has just about reached
its limit as measure of computer performance.
In the last Harvard Cup, Wchess had no losses.
Wchess scored higher than any human.
It had 83% vs Joel Benjamin with 81%.
Mark Schreiber
Brian Karen
>: computers perform relative to the previous Harvard Cups. Therefore,
the
>: time control must be the same as in previous years.
>
>: Naisortep/Brian
Karen
>Then the Harvard Cup has just about reached
>its limit as measure of computer performance.
>In the last Harvard Cup, Wchess had no losses.
>Wchess scored higher than any human.
>It had 83% vs Joel Benjamin with 81%.
> Mark Schreiber
Well "has just about" is not the same as Has. Perhaps, when the
Harvard Cup has reached its limit as a measure of computer performance
at g/25 they will move on to different time controls.
As it stands now Wchess's result might have been a fluke. Lets see
how it does this year. -Naisortep/Brian Karen
David Flude
>Christopher F. Chabris (c...@wjh.harvard.edu) wrote:
>: SIXTH HARVARD CUP HUMAN VERSUS COMPUTER CHESS CHALLENGE
>: The Sixth Harvard Cup will feature six of America's top grandmasters against
>: six of the leading microcomputer-based chess software packages in a team
>: match format. Each human will play each computer once, at a time control
>: of game in 25 minutes.
>For many years a tournament where there are two teamsa and each member of a team plays
all members of the other team has been called a Scheveningen
tournament after a tournament of this format at Scheveningen. May I
suggest that in future releases that this term be used.
Mark Schreiber
: >: One of the reasons for having the Harvard Cup is to see how the
1 game might be a fluke. 6 games is not a fluke.
If a computer were to win all its games,
we could not measure its performance.
It might have a performance of 3000 or 3500.
Anything over 3000 would achieve this result.
The harvard cup could not measure computer
performance at this point. Mark Schreiber
Ed Seedhouse
>If a computer were to win all its games, we could not measure its performance.
Utter stuff and nonsense. Go read Elo and come back when you actaully
understand something about the rating system and how it works. Until
you do you are only advertising your ignorance.
Ed Seedhouse
President, Victoria Chess Club.
CFC Rating: 2040
Timothy Hanke
I think we should seize the Harvard Cup organizers, lead them out behind
the building, and shoot them all in the back of the head for daring to
mount an event that provides us with less-than-definitive data.
So what if they are my friends. Justice must be served.
Timothy Hanke
Halibut
: A Modest Proposal
:
: I think we should seize the Harvard Cup organizers, lead them out behind
: the building, and shoot them all in the back of the head for daring to
: mount an event that provides us with less-than-definitive data.
But when do we get dinner?
Brian Karen
>: >Then the Harvard Cup has just about reached
>: >its limit as measure of computer performance.
>: >In the last Harvard Cup, Wchess had no losses.
>: >Wchess scored higher than any human.
>: >It had 83% vs Joel Benjamin with 81%.
>: > Mark Schreiber
>
>:
>: Well "has just about" is not the same as Has. Perhaps, when the
>: Harvard Cup has reached its limit as a measure of computer
performance
>: at g/25 they will move on to different time controls.
>:
>: As it stands now Wchess's result might have been a fluke. Lets
see
>: how it does this year. -Naisortep/Brian Karen
>
>1 game might be a fluke. 6 games is not a fluke.
>
> Mark Schreiber
The results are in for the 1995 Harvard Cup:
Wchess score = +1 -5 = 17%
Christopher F. Chabris
>The results are in for the 1995 Harvard Cup:
>Wchess score = +1 -5 = 17%
Just a tiny correction: WChess scored +1 =1 -4, or 1.5/6, or 25%. But the
point remains: six games is a small sample to measure anyone's or
anything's strength.
John McMenamin
: say that computers are better players than humans just because the
: computers have won one action chess match.
Unless you look back on previous action matches and check the results
for the computers over time. Seeing consistant wins over humans and
improving scores (no losses for example) each consecutive tournament
should give you some clue that computers will one day come very close
to beating humans most of the time. :)
---
^^__ Johnny McMenamin -=> Eugene, OR, US _ _ _
/ - \_ http://www.rio.com/~johnnymc .:::. | || || |
<| __< ..... ,.. ::::: |_______|
<| \ ,;' ,;: :: `:::' \__ ___ /
<| \ ;: ::..., ...... .,,,. .,,,. `:' |___|_|
<|______\ ;: :: :: ;: .:: :: '' :: '' : |_|___|
_|____|_ ;: ,;; :: :: ::,::''. ``::, ``::, , |___|_|
(________) ':,,,;' ,;; ,;; `:,,,,:' `;,,,;' `;,,,;' :;: (_______)
/________\ -- If you only look at what is, -- /_______\
-- you might never attain what could be. --
Halibut
: >Wchess score = +1 -5 = 17%
: Just a tiny correction: WChess scored +1 =1 -4, or 1.5/6, or 25%. But the
: point remains: six games is a small sample to measure anyone's or
: anything's strength.
The sample hasn't gotten any smaller since last year.
BTW, what were the programs running on? On 133 Pentiums or on 133
Pentium Pro machines?
Another BTW: Will you be permitting ChessMaster 4000 to advertise
its results from the Harvard Cup?
Tim Mirabile
>etpe...@netcom.com (Eric Peterson) wrote:
>>In article <4bdbj8$p...@cnct.com> msch...@cnct.com (Mark Schreiber) writes:
>>>
>>>Computers playing at these fast time controls have too much of an advantage.
>>>Even Kasparov lost to a computer at these fast time controls.
>>>Longer time controls would be a fairer challenge.
>
>>The computers scored only 39% in last year's Harvard Cup, so it seems to
>>be a fair matchup, no?
>
>If a matchup were arranged whereby a player like myself could score 39%
>against GMs like Gulko, would you thus also assume that it was a fair matchup?
>
>Let's the the ball rolling here, I could use the rating points.
>
>Ariel
Ok, let me know under what conditions could you score 39% against
Gulko in a normal chess match with equal time on both clocks.
+---------------------------------+
| Tim Mirabile <t...@mail.htp.com> |
| PGP Key ID: B7CE30D1 |
+---------------------------------+
Tim Mirabile
>The Harvard Cup has several things that are open to criticism:
>
>- As noted above, the games are very fast, which increases the odds of a
>computer win, and the GMs usually do worse as the day wears on, as well.
>
>- There is a high fee to enter a computer, so basically this is a promotional
>opportunity with a high likelihood of at least one win over a GM being sold as
>though it was a definitive test.
Of course the other side of the coin is that American GM's who play
are provided with a nice income for a few days work, plus additional
money to the highest scorer. Also, it provides the programmers with
something to shoot for. And besides, it's fun!
Ariel Mazzarelli
>maz...@primenet.com (Ariel Mazzarelli) wrote:
>>etpe...@netcom.com (Eric Peterson) wrote:
>>>In article <4bdbj8$p...@cnct.com> msch...@cnct.com (Mark Schreiber) writes:
>>>>Computers playing at these fast time controls have too much of an advantage.
>>>>Even Kasparov lost to a computer at these fast time controls.
>>>>Longer time controls would be a fairer challenge.
>>>The computers scored only 39% in last year's Harvard Cup, so it seems to
>>>be a fair matchup, no?
>>If a matchup were arranged whereby a player like myself could score 39%
>>against GMs like Gulko, would you thus also assume that it was a fair matchup?
>>Let's the the ball rolling here, I could use the rating points.
>Ok, let me know under what conditions could you score 39% against
>Gulko in a normal chess match with equal time on both clocks.
Ok, how about computer conditions. I get a chess database and a buffer board
to move the pieces around. We figure out the time control at which I could do
all this (not too fast, not too slow). Maybe I could improve on the following,
which was 40/2 and no cpu access (N.B.: I was having a good tournament, my
opponent wasn't).
[Event "?"]
[Site "Las Vegas (USA)"]
[Date "1987.??.??"]
[Round "3"]
[White "Gulko Boris F"]
[Black "Mazzarelli Ariel"]
[Result "1-0"]
[ECO "E52"]
1. d4 Nf6 2. c4 e6 3. Nf3 b6 4. Nc3 Bb4 5. e3 Bb7 6. Bd3 O-O 7. O-O d5
8. a3 Bd6 9. cxd5 exd5 10. b4 Re8 11. b5 a6 12. Qc2 Ne4 13. Bb2 h6
14. a4 Nxc3 15. Qxc3 Nd7 16. Ba3 Qe7 17. Bxd6 Qxd6 18. Rfc1 Rec8 19. Bf5
c5 20. bxc6 Rxc6 21. Qb2 Bc8 22. Rxc6 Qxc6 23. Bxd7 Bxd7 24. Ne5 Qd6
25. Nxd7 Qxd7 26. Qxb6 g6 27. a5 Qc8 28. h3 Rb8 29. Qd6 Rb5 30. Qa3 Qc7
31. Rc1 Rxa5 32. Rxc7 Rxa3 33. Rc5 a5 34. Rxd5 Kf8 35. g4 a4 36. Ra5
Ra1+ 37. Kg2 a3 38. Ra7 a2 39. Kf3 Kg7 40. Kf4 g5+ 41. Ke5 Kg6 42. d5
Kg7 43. Ra6 Kg8 44. Kf6 Rd1 45. Ra8+ Kh7 46. Rxa2 Rxd5 47. Kxf7 Rd1 1-0
Even now, looking at this game is a little painful.
Ariel
Tim Mirabile
>Joe Stella (jo...@ultranet.com) wrote:
>: msch...@cnct.com (Mark Schreiber) writes:
>: >Computers playing at these fast time controls have too much of an advantage.
>: >Even Kasparov lost to a computer at these fast time controls.
>: >Longer time controls would be a fairer challenge.
>: I tend to think that they use these time controls for many reasons, not
>: the least of which is that it is easier (and cheaper for the sponsor) to
>: organize a match that will not last too long.
>Easier, cheaper and also a mediocre test of chess playing ability.
>Its not hard to have a tournament with longer time controls,
>Aegon does it every year.
Do all computer/human tournaments need to have the same format?
What's wrong with a little variety.
>: Everyone who follows computer chess knows that computers have more of an
>: advantage at fast time controls than they do at slower ones. The
>: Harvard Cup is useful to see how much the computers have improved,
>: and how computers fare against humans at *action* chess.
>As a long term barometer,
>action chess has just about reached its limit.
>In the last Harvard Cup, Wchess had no losses.
>It scored higher than any human.
>Wchess had 83% vs Joel Benjamin at 81%.
It doesn't make sense to compare the scores of the computers versus
the scores of the players. They didn't play the same group of
opponents. Perhaps you could argue that the computer's 83% is much
more impressive than Joel Benjamin's 81% because Benjamin didn't have
to play the other GM's, but then, WChess didn't have to play the other
computers.
Now this year, WChess only scored 25% on a (supposedly) faster
machine.