ChessMaster 4000
ArturoA433
Joe Stella
artur...@aol.com (ArturoA433) writes:
>Anyone Knows the rating of this program in a Pentium 100
Here we go again!
There is some evidence that the rating is close to 2500 USCF (this is
in Computer Chess Reports, published by ICD 1-800-645-4710). There
are also some people that strenusouly disagree with this number, because
it comes from a lot of computer/computer games and these can tend
to exaggerate the difference in rating between the computers involved.
The upshot of all this is that we don't really know, and aren't going to
really know until CM4000 is played in *many* human tournmaments, which
may never happen.
My personal opinion is that CM4000 is at least 2300 USCF on a pentium 100.
Joe Stella
Doctor SBD
hard to tell from "single case studies." I know how well I do against it,
but I have heard of supposed 2200 players who have minus scores against
CM3000, and therefore estimate its rating at 200-2300.
Joe Stella
doct...@aol.com (Doctor SBD) writes:
>I would estimate 2000-2100 as more likely, although 2300 is possible. It's
But what *system* were you running it on? Was it a pentium 100, as the
original poster was asking? This makes a difference!
The difference between a 486/66dx2 and a pentium 100 is about 85 rating
points for most programs. If (as I suspect) you have an even older system,
then the difference is greater.
Joe Stella
Max Burkett
the program runs, not what's in it. If true then all these claims should
include the time given the program to find a move as well as the
processor. In the endgame, CM4000 is a patzer. In the middlegame, it is
very good. It is fairly easy to beat because it is predictible. The
other Max
ArturoA433
(Packard Bell 8RAM) in which I am playing many games at 6 and 25 minutes
is stronger than Fritz3, both are running in the same time conditions,
but with Fritz3 I am using the Mega hash tables ("x mode"). I tested both
in a Packard Bell Pentium 100 , 60 and in a Gateway2000 Pentium 66 and
CM4000 won the match in any of this 3 computers.
Raymond
Armagnac,Raymond
Doctor SBD
wonder if this would be true over a long period of time as well?
Axel Boldt
Joe> My personal opinion is that CM4000 is at least 2300 USCF on a pentium 100.
Is one of these animals playing on FICS like the brave gnuchess?
--
Axel Boldt * bo...@math.ucsb.edu * http://math-www.uni-paderborn.de/~axel/
Ray Frank
Doctor SBD <doct...@aol.com> wrote:
>I would estimate 2000-2100 as more likely, although 2300 is possible. It's
>but I have heard of supposed 2200 players who have minus scores against
>CM3000, and therefore estimate its rating at 200-2300.
How well do you go against it and what is your rating?
CM4000 has been rated as high as 2500. So if it really is only 2000-2100
then there is something seriously wrong with the ratings formula since
a drop from 2500 to 2100 is a seriously significant difference.
I guess what I'm trying to say is that even if the ratings of computers
only playing computers does not reflect its true rating, it should not
show an error of 300 or 400 rating points.
ray
Doctor SBD
computers, and that (to me anyway) makes the CM rating of 2500 specious
(especially since I have also seen estimated ratings of 2200 for it).
Computer ratings are based on too few games for my liking.
BradleeJ
And probably and too fast a time control. One way to jack up the rating of
a chess program is play it against humans at faster time controls. Where
humans tend to make stupid mistakes in time pressure, computers have
nearly perfect sight and so don't blunder and on the other hand they
pounce on any error made by a human. I suspect if all the top programs on
the market were entered into major competitions at standard time controls,
the ratings wouldn't look so good.
Regards.
-Bradlee Johnson
Axel Boldt
Ray> CM4000 has been rated as high as 2500. So if it really is only 2000-2100
Ray> then there is something seriously wrong with the ratings formula since
Ray> a drop from 2500 to 2100 is a seriously significant difference.
Again, why not just have it play on FICS. Then we would know its
rating.
Christopher Dorr
: Anyone Knows the rating of this program in a Pentium 100
CM4000 went 2 wins, 1 draw, and 3 losses in the 1994 Harvard Cup G/25
tournament, against IGM's rated avg. USCF 2650. I believe that this was
also on a Pentium 90, or possibly 100. From this, I think it's fairly
safe to assume that it is at least 2500 at 40/2 on that hardware.
I find it inconceivable that a player of less than 2500 USCF could score
like that against world-class players. G/25 will spmewhat overestimate
it's strength, but not by that large a margin.
I am rated USCF 2150, and have it running on a 486/80, and feel that it
is well over 2400 (Possibly near 2500) on my hardware. Computer Chess
Reports shows it at 2492 on a 486/66 w/ 8 MB RAM.
There may well be differences of opinion about it's exact rating, but
given it's performances both against other computers, and against humans,
in a variety of settings (on the ICS for example it is rated around 2600
at blitz) I don't think there can be any real question that it is very
strong, and at least US Senior Master strength on anything faster than a
486/50.
Chris
Doctor SBD
convinced when computers play *many* games over 1 or 2 years, not isolated
results.
A new opponent is always more difficult to beat.
Max Burkett
avoids the blunders at moves 16,18,&19 in the following game.
W:Max [USCF 2180/1000g FIDE 2300*/5g](+~50 /Open Thinking Window)
B:Chessmaster [default personality] @ 30moves/hour
1. e4 d6 2. d4 Nf6 3. Nc3 g6 4. f4 Bg7 5. Bd3 Bg4 6. Nf3 Nc6
7. e5 dxe5 8. dxe5 Nd5 9. O-O Nxc3 10. bxc3 O-O 11. h3 Bf5
12. Qe2 Qd7 13. Be3 Bxd3[bad, but in default obk] 14. cxd3 b5
15. d4 Na5 16. Nd2 Rfd8? 17. f5 gxf5 18. Qh5 e6? 19. Rf3 b4?
20. Rg3 Kh8 21. cxb4 f4 22. Rxg7 Kxg7 23. Qg5+ Kh8 24. Qxf4 Nc6
25. Qf6+ Kg8 26. Ne4 Nxd4 27. Rf1 1-0 [eventually]
Belmig
Miguel
Ray Frank
All computers given a rating are playing at legal USCF time controls. I believe
the fasted recognized USCF time controls are 30 moves in 30 minutes. Then there
are the blitz times controls which is a completely different category of chess.
ray
Steve Kelly
wrote:
> All computers given a rating are playing at legal USCF time controls.
> I believe the fasted recognized USCF time controls are 30 moves in 30
> minutes. Then there are the blitz times controls which is a
> completely different category of chess.
When computers are "USCF Certified," however, you will find that the
games are practically all played at 30/30. The computer manufacturers
understand well that computers perform far stronger against humans with
faster time controls, so they almost always opt for the very fastest
time control allowed.
When you see a rating "USCF Certified" rather than a rating established
in rated play, you have to take it with a grain of salt.
---
Steve Kelly sac...@netcom.com
... She offered her honor. He honored her offer!