Chessmaster 6000 Ratings
James Brian Rairden
Chessmaster posts I've seen recently, I couldn't resist. My questions is:
Does anyone know how reliable the CM-6000 ratings are for the personalities?
I only recently began playing against some of them regularly. I defeated
"Vanessa" who is rated 2200 on my Pentium III, 1 GHz, 256 RAM system. I am
curious about whether that 2200 rating is even close to being accurate, or
did I simply beat what is realistically a 1300-or-so personality? Thanks to
anyone who has an answer.
Jim R.
Tom E.
"James Brian Rairden" <rai...@pldi.net> wrote in message
news:u8qmlo2...@corp.supernews.com...
Probably. My rating on Yahoo is usually around 1300, but I am still in
provisional on CM8000 and I'll be lucky to break 1000. Depressing.
Sparing the annotating and chess clocks, CM is about as close to the
real thing as you can get. Yahoo is over rated by at least 300 points
from CM. What does your computer specs have to do with it? -Tom
Quad City Chess
news:u8qnh03...@corp.supernews.com...
> What does your computer specs have to do with it? -Tom
More RAM = faster computer = more calculations per second = longer
variations = better computer opponent.
Regards,
Matt
Sebastian Awatramani
since they are all programmed by the programmers.
-Sebastian
"Quad City Chess" <qcc...@NODAMNSPAMmchsi.com> wrote in message
news:VUfj8.24415$Yv2.10425@rwcrnsc54...
DK
Maybe Venessa let you win because she likes you. Is she hot?
What kind of match was it? Maybe you played an anti-computer strategy
and didn't notice it (i.e. closed game, little tactics requiring a long
range plan to win).
DK
James Brian Rairden
Pentium 1 dinosaur, the personalities were rated differently. Even on my
current system, I have seen small variations in ratings on different days.
I've been told that was caused by the "system resources" under "performance"
in "My Computer/Properties." I don't really know. One day, Vanessa was
rated 2196, and the next day she was 2200. Oh well. Thanks.
Jim R.
"Sebastian Awatramani" <ninonline@(NO_SPAM)onebox.com> wrote in message
news:lvkj8.9442$P4.7...@newsread2.prod.itd.earthlink.net...
james liggett
Chessmaster.com has this to say about CM8000 ratings.
Q: How accurate are the ratings for the personalities?
A: The ratings are mathematically precise, but they are mostly based
on computer vs. computer games. While the human vs. computer games
added much more "real-world" accuracy to the ratings, many of the
Chessmaster personalities have quirks in their playing styles that a
human will be able to exploit but a computer opponent will not be able
to "see". Discovering and then taking advantage of these weaknesses is
part of the fun of Chessmaster, but you might find yourself able to
defeat certain personalities that you don't think you should be able
to based on their ratings.
I have CM8000 on 2 computers one is a PI 200 MHz the other is a AMD
900 MHz.
On the PI 200 CM8K is rated 2727 and vanessa is rated 2112.
While on the AMD 900 CM8k is rated 2900 and vanessa is rated 2255.
So cpu speed will have something to do with its rating. Or at least
it's own way of giving it's self a est rating based on cpu speed.
I use to play vanessa on CM7k and CM8k but moved on because that
personality got to easy to beat and I move to a stronger personality.
Just to see I played vanessa on my PI 200 and checkmated it in 44
moves.
Then played I vanessa on my AMD 900 and it took me 91 moves to
checkmate only after I queened a pasted pawn (the other game vs the PI
didn't go into the endgame) so based on cpu speed I can say the faster
cpu did make vanessa play stronger. Did it make vanessa play 100 pts
better I can't say based on 1 game vs each.
Like CM.com said I can beat some higher rated personality's while at
the same time I have trouble with some weaker ones.
My CM8k rating is 2200 + how accurate is that vs a real otb rating? I
don't know since I have never play in a real rated game. But I
wouldn't put to much in the ratings that anyone gets playing rated
games vs CM personality's unless your beating the ones with 2500+
ratings or the top playing GM personality's playing at top strength of
the program.
I would bet that you can take anywhere from 100 to 200 pts off the
rating to get a est real rating.
The CM.com has some more to say about its ratings.
Q: The Chessmaster personality shows a rating of 2800 (or more). That
certainly can't be "accurate", can it?
A: Strictly based on the math described above, it IS accurate.
However, there is a problem of "diminishing returns" as computers get
faster and faster, especially at shorter time controls. The difference
between a Pentium III-1Ghz processor and a Pentium III-1.2Ghz
processor is not going to be as significant. As the Chessmaster
personality reports higher and higher ratings, this value should be
considered as more of a "benchmark" of your computer's speed, rather
than a "real-world" chess rating.
I have notice that on all of CM versions from 6k to 8k the josh
waitzkins ones all have the same rating no mater how fast the cpu
speed is. Josh at age 12 is 2100 and his rating stays the same.
The only thing I use the "rating" from CM that I have is a way to see
if I am improving over time.
James Brian Rairden
correct. Last night I drew against the CM personality (2785 on my system).
I found one of the flaws they were referring to. I tried the same series of
moves, and sure enough the CM drew against me the exact same way--in a
situation that no human would allow to happen. Now I won't take the program
too seriously except as a training device. Thanks again.
Jim R.
"james liggett" <redzo...@att.net> wrote
John Merlino
> Thank you, sir. I appreciated your reply. You (and the CM folks) are
> correct. Last night I drew against the CM personality (2785 on my system).
> I found one of the flaws they were referring to. I tried the same series of
> moves, and sure enough the CM drew against me the exact same way--in a
> situation that no human would allow to happen. Now I won't take the program
> too seriously except as a training device. Thanks again.
>
> Jim R.
Could you please post the sequence of moves, as well as the time
control and the hardware you were playing on? It might be possible
that this is a "hole" in the opening book that needs to be
investigated.
Many thanks,
jm
John Merlino
There was an earlier (long) post in reply to this that pretty much
hits the mark in all respects. However, there is a BIG difference in
the accuracy of the ratings between Chessmaster 6000 and 8000. For
CM8000 there were several thousand human vs. comp games played against
real rated humans (mostly USCF, although some were Canadian). This
allowed for quite a lot more "real world" accuracy to the ratings
calculations.
There were NO human vs. comp games played until CM8000, so the ratings
for CM6000 and CM7000 are all based on ONE outside source of testing,
that being the SSDF. However, the SSDF ALSO has done very little human
vs. comp testing. Therefore, I would consider ratings in CM6000 and
CM7000 to be only "accurate" within their own world, and could
literally be as much as 300+ points off (usually too high) when
compared to USCF ratings. Most ratings, in CM6000, though, were
adjusted by less than 100 points when they were calculated for CM8000.
CM8000's ratings are, for almost all personalities, believed to be
within 100 points of USCF equivalents. However, some error in the
calculations will creep in thanks to different processor speeds.
Before the 1.0.2 patch of CM8000 (as well as in all previous versions
of CM), for each doubling of CPU speed above the base machine each
personality was given a maximum of 70 points (the actual value was
also based on the personality's strength of play and its maximum
search depth). For the 1.0.2 patch, this value was decreased to a
maximum of 50 points, to try to minimize the "diminishing returns"
effect mentioned in that other post. (BTW, that info came from the
CM8K FAQ, which you can read by going to
www.chessmaster.com/faqs8000.html).
But, like any AI in a computer program, once you find the flaw(s) in
their programming (sometimes intentional and sometimes not ;-), even
the (otherwise) toughest opponents can become much easier to beat.
That's why there are so many personalities in the game, so that there
are always new challenges from personalities that play VERY
differently, requiring you to rethink how you play against each new
opponent.
jm
Chris Kantack
"Sebastian Awatramani" <ninonline@(NO_SPAM)onebox.com> says...
> But that wouldn't affect the computers rating of the CM personalities
> since they are all programmed by the programmers.
>
appropriate ratings. I have several systems ranging from a 133 MHz Pentium
to a new 1.6GHz P4. The ratings very quite a bit between my slowest and
fastest systems.
A 200 MHz Pentium MMX has ChessMaster Elos around 75 points higher than a
133 MHz Pentium. I haven't yet directly compared my new P4 against my
other systems but a brief glance shows a considerable difference in Elo's.
(A couple hundred points higher?)
Chris