SHREDDER wins the WCCCC!!
Rolf Tueschen
Paderborn (Germany) both titles of the Micro Wch and the overall
champion of all machine horsepower!
S H R E D D E R CONGRATULATIONS!
PMG
Rolf Tueschen wrote:
That was a good tournament all the way through to the final playoff round
with Shredder vs Ferret. Shredder wins on points and deserves all
congradulations. No but..but....but..... Shredder is the top dog.
For a download of the tournament in pgn format:
http://www.uni-paderborn.de/~wccc99/gindex.html
Pete
kri...@my-deja.com
was rather modest. But whatever happened in HIARCS-CILKCHESS ???
As per the score sheet after 14 moves why did hiarcs not play
15. N:d7 Q:d7 16 Qa4 winning instead of 15 Ba6 ??? and ultimately
losing. Shocking to see one of the strongest programs miss a win
here.
Is the score wrong ?? Or is there a bug ?
In article <376C07B8...@Chicagonet.net>,
Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.
Rolf Tueschen
> That was a good tournament all the way through to the final playoff round
>with Shredder vs Ferret. Shredder wins on points and deserves all
>congradulations. No but..but....but..... Shredder is the top dog.
If you take the event as a sports event as such this is absolutely true.
It is the top for this event.
"But" it is obvious that "this event" itself has its
prepositions/preconditions. I would NOT say that FRITZ for instance is
now proovenly weaker than SHREDDER. Or JUNIOR weaker than ___.
So, SHREDDER as such, after this event, is not alt all the absolute top
"dog". My opinion and my knowledge about such tournament conditions.
Look at 7 rounds with 30 participants!
Tim Anger
>was rather modest. But whatever happened in HIARCS-CILKCHESS ???
>As per the score sheet after 14 moves why did hiarcs not play
>15. N:d7 Q:d7 16 Qa4 winning instead of 15 Ba6 ??? and ultimately
>losing. Shocking to see one of the strongest programs miss a win
>here.
>Is the score wrong ?? Or is there a bug ?
This game is more than incredible. On my system Hiarcs 7.32
immediately finds 15.Nxd7 and 17.Nxd7, both winning at once. The
Hiarcs who joined WCCC99 was certainly much weaker then the commercial
version.
kris...@hotmail.com
played 13... Be8 instead of 13... Re8.
In article <376d2997...@personalnews.de.uu.net>,
Greg Kennedy
Does anybody know why the world computer championship
does not use equal hardware for each program in order to
compare programs fairly?
- Greg Kennedy
PMG
Greg Kennedy wrote:
That would be considered a restriction, the platform that you choose to
run on can be considered an important part of your work. The platform
that is ideal for one program is not going to be the platform that is
ideal for any of the rest of the programs. It's not too important that
we argue in the newsgroups about what program is _really_ stronger
because of what computer that it did or didn't run on. The fact still
remains that Shredder won that tournament. Stefan deserves to be
congratulated. Going by the guidelines of that tournament Shredder is
the winner, nobody else can claim that. No one wants to hear someone
again say that everyone who participated is a winner, sure ok, but
Shredder won.
Pete
Theo van der Storm
> .....Is the score wrong ?? Or is there a bug ?
>
> This game is more than incredible. On my system Hiarcs 7.32
> immediately finds 15.Nxd7 and 17.Nxd7, both winning at once. The
> Hiarcs who joined WCCC99 was certainly much weaker then the commercial
> version.
Shredder and Ferret were clearly the best two programs.
Notice the strong opponents, e.g. Bucholtz points, Hiarcs had to play.
Cilkchess (4) did not have to play the top 3. This situation would not happen
in an 11 round tournament.
# Name 1 2 3 4 5 6 7 Score BU SB
1 Shredder 30w1 3b1 7w1 8b= 10w= 2b= 5w1 5.5 29.0 22.0
2 Ferret 20b1 14w1 10b0 24w1 12b1 1w= 3w1 5.5 28.0 21.2
3 Fritz 27b1 1w0 26b1 13w1 5b1 10w1 2b0 5.0 27.5 16.5
4 Cilkchess 21w1 24b1 5b0 12w= 11b1 6w= 10b1 5.0 26.5 17.8
5 Junior 15w1 6b1 4w1 10b= 3w0 8w1 1b0 4.5 32.5 20.0
6 DarkThought 12b1 5w0 21b1 11w= 16b1 4b= 7w= 4.5 28.5 17.2
7 Rebel 25b1 11w1 1b0 18w1 8b0 13w1 6b= 4.5 28.0 15.8
8 Nimzo 26b= 22w1 16b1 1w= 7w1 5b0 9w= 4.5 27.5 17.0
9 ChessTiger 10b0 13w0 29b1 22w1 14b1 19w1 8b= 4.5 23.5 13.2
10 Hiarcs 9w1 17b1 2w1 5w= 1b= 3b0 4w0 4.0 33.5! 18.5
11 LambChop 18w1 7b0 15w1 6b= 4w0 24b1 16w= 4.0 27.0 13.5
12 Francesca 6w0 28b1 17w1 4b= 2w0 16w= 20b1 4.0 26.5 12.2
13 Virtual_Chess_X 24w0 9b1 19w1 3b0 26w1 7b0 18w1 4.0 24.5 12.5
14 GromitChess 28w1 2b0 18w0 23b1 9w0 27w1 25b1 4.0 22.0 9.0
15 EUGEN 5b0 29w1 11b0 21w1 24b= 20w= 19b1 4.0 21.0 9.8
16 Zugzwang 19w= 23b1 8w0 17b1 6w0 12b= 11b= 3.5 26.5 12.0
17 MChess 29b1 10w0 12b0 16w0 25b= 26w1 24w1 3.5 19.5 6.8
18 P.ConNerS 11b0 20w1 14b1 7b0 19w0 21w1 13b0 3.0 25.5 10.0
19 Isichess 16b= 26w= 13b0 25w1 18b1 9b0 15w0 3.0 23.5 8.2
20 Diep 2w0 18b0 30w= 28b1 22w1 15b= 12w0 3.0 22.0 7.0
21 Patzer 4b0 27w1 6w0 15b0 30w1 18b0 29w1 3.0 20.5 4.0
22 Mini 23w= 8b0 28w= 9b0 20b0 30w1 27b1 3.0 19.5 5.2
23 Now 22b= 16w0 27b0 14w0 28w= 29b1 26b1 3.0 17.0 5.2
24 SOS 13b1 4w0 25b1 2b0 15w= 11w0 17b0 2.5 28.5 8.5
25 Arthur 7w0 30b1 24w0 19b0 17w= 28b1 14w0 2.5 20.0 4.2
26 Ikarus 8w= 19b= 3w0 27w1 13b0 17b0 23w0 2.0 25.0 5.8
27 Centaur 3w0 21b0 23w1 26b0 29w1 14b0 22w0 2.0 21.0 4.0
28 RuyLopez 14b0 12w0 22b= 20w0 23b= 25w0 30b= 1.5 20.5 3.5
29 XXXX2 17w0 15b0 9w0 30b1 27b0 23w0 21b0 1.0 21.0 1.0
30 Neurologic 1b0 25w0 20b= 29w0 21b0 22b0 28w= 1.0 19.5 2.2
Name 1 2 3 4 5 6 7 8 9101112131415161718192021222324252627282930
Score BU SB
1 Shredder X = 1 1 1 = = 1
5.5 29.0 22.0
2 Ferret = X 1 0 1 1 1 1
5.5 28.0 21.2
3 Fritz 0 0 X 1 1 1 1 1
5.0 27.5 16.5
4 Cilkchess X 0 = 1 1 = 1 1
5.0 26.5 17.8
5 Junior 0 0 1 X 1 1 = 1
4.5 32.5 20.0
6 DarkThought = 0 X = = 1 1 1
4.5 28.5 17.2
7 Rebel 0 = X 0 1 1 1 1
4.5 28.0 15.8
8 Nimzo = 0 1 X = 1 1 =
4.5 27.5 17.0
9 ChessTiger = X 0 0 1 1 1 1
4.5 23.5 13.2
10 Hiarcs = 1 0 0 = 1 X 1
4.0 33.5 18.5
11 LambChop 0 = 0 X 1 = 1 1
4.0 27.0 13.5
12 Francesca 0 = 0 X = 1 1 1
4.0 26.5 12.2
13 Virtual_Chess_X 0 0 1 X 1 1 0 1
4.0 24.5 12.5
14 GromitChess 0 0 X 0 1 1 1 1
4.0 22.0 9.0
15 EUGEN 0 0 X 1 = 1 = 1
4.0 21.0 9.8
16 Zugzwang 0 0 = = X 1 = 1
3.5 26.5 12.0
17 MChess 0 0 0 X 1 = 1 1
3.5 19.5 6.8
18 P.ConNerS 0 0 0 1 X 0 1 1
3.0 25.5 10.0
19 Isichess 0 0 0 = 1 X 1 =
3.0 23.5 8.2
20 Diep 0 0 = 0 X 1 1 =
3.0 22.0 7.0
21 Patzer 0 0 0 0 X 1 1 1
3.0 20.5 4.0
22 Mini 0 0 0 X = 1 = 1
3.0 19.5 5.2
23 Now 0 0 = X 1 0 = 1
3.0 17.0 5.2
24 SOS 0 0 0 1 = 0 X 1
2.5 28.5 8.5
25 Arthur 0 0 = 0 0 X 1 1
2.5 20.0 4.2
26 Ikarus 0 = 0 0 = 0 X 1
2.0 25.0 5.8
27 Centaur 0 0 0 0 1 0 X 1
2.0 21.0 4.0
28 RuyLopez 0 0 0 = = 0 X =
1.5 20.5 3.5
29 XXXX2 0 0 0 0 0 0 X 1
1.0 21.0 1.0
30 Neurologic 0 = 0 0 0 = 0 X
1.0 19.5 2.2
Play-off: Shredder-Ferret 1/2-1/2
See also:
http://ourworld.compuserve.com/homepages/thstorm/
and
http://www.uni-paderborn.de/~wccc99/
Greetings from Amsterdam,
Theo van der Storm
Theo van der Storm
>
> >Is the score wrong ??
> ....
> Other, please reply without insult. I am only assisterating the newbie here.
Anyway, jumpify to
http://supertech.lcs.mit.edu/cgi-bin/chessdb/index?1999-wccc
for original Cilk,
Theo
Greg Kennedy
>> Does anybody know why the world computer championship
>> does not use equal hardware for each program in order to
>> compare programs fairly?
>>
>run on can be considered an important part of your work. The platform
>that is ideal for one program is not going to be the platform that is
>ideal for any of the rest of the programs. It's not too important that
>we argue in the newsgroups about what program is _really_ stronger
>because of what computer that it did or didn't run on. The fact still
>remains that Shredder won that tournament. Stefan deserves to be
>congratulated. Going by the guidelines of that tournament Shredder is
>the winner, nobody else can claim that. No one wants to hear someone
>again say that everyone who participated is a winner, sure ok, but
>Shredder won.
You seem to be reading a lot into my post which isn't there.
The round-by-round results mentioned that Shredder was in
fact, not helped, but HANDICAPPED by the hardware it was
running on relative to the next finishers. Thus, its victory was
quite clear. My question was just a plea for a way to compare
the PROGRAMS and their relative strengths, without the
handicaps which skew the results toward who selected faster
hardware. As far as I know, the selection of faster hardware
requires almost zero skill, but the programming of a stronger
chess engine is a different matter. I'm asking "who is the best
chess programmer" and "how does he do it," not "gee, I wonder
if a ten million dollar Cray is faster than my PC?"
- Greg Kennedy
Rolf Tueschen
<376D86EB...@compuserve.com>:
>Shredder and Ferret were clearly the best two programs.
>Notice the strong opponents, e.g. Bucholtz points,
Please don't forget the restrictions that must be made for a 7-round
Swiss with 30 participants. One single game (loss or win) could decide
about being top or flop. Take for instance the usually strong VIRTUAL.
One loss less and it would have been back in the top ranks.
This doesn't take away something from SHREDDER. At a factual level it
simply won a great sports event. But all other conclusions about being
better or best are simply false.
Note that this question is NOT one of our personal opinions. It is one
of statistics. Like it or not.
PR will be a different question ...
PMG
called IROC where various highly regarded race car drivers from different
car race fields race against eachother in iddentical cars, all prepared by
the IROC organazation. It is the most boring race around, there's no reason
for having it other than to sell advertising on TV.
I'd hate to see WCCC become the IROC of computer Chess. It would be far too
expensive for ICCA to supply all of the programers with a computer to work
with to prepare for the event. This works very well the way it's run. And no
it's not a simple matter of selecting a Cray.
Pete
mclane
>Rolf Tueschen wrote:
>> After he won the Wch in Jakarta in 1996 Stefan Meyer-Kahlen won today in
>> Paderborn (Germany) both titles of the Micro Wch and the overall
>> champion of all machine horsepower!
>>
>> S H R E D D E R CONGRATULATIONS!
> That was a good tournament all the way through to the final playoff round
>with Shredder vs Ferret. Shredder wins on points and deserves all
>congradulations. No but..but....but..... Shredder is the top dog.
>For a download of the tournament in pgn format:
>http://www.uni-paderborn.de/~wccc99/gindex.html
>Pete
Shredder is the top dog. right. because shredder has killed the
preprocessing-paradigm.
no loss.
no special machine.
and less mistakes than the opponnents.
best wishes
wilson spaqi
Paderborn website?
Greg Kennedy
>I'm sorry, I was reading allot into it. We have (or did have) this car
race
>called IROC where various highly regarded race car drivers from different
>car race fields race against eachother in iddentical cars, all prepared by
>the IROC organazation. It is the most boring race around, there's no
reason
>for having it other than to sell advertising on TV.
Boring race eh? Too bad. I suppose a chess tournament
could be looked at the same way, if for example, one player
was going to win by ten points and nobody cared who came
in second.
But my scenario was quite different. I believe it would be
very close if the hardware were the same and only the
programs varied. We would see a one-on-one contest in
each game, where the programmers were battling each
other's chess programming skill on even terms. Far more
interesting a contest than "gee, I wonder if we will get
disconnected from our monster mainframe and forfeit two
Queens ahead, or not? Can you believe this fool is still
using a PC against us?"
>I'd hate to see WCCC become the IROC of computer Chess. It would be far
too
>expensive for ICCA to supply all of the programers with a computer to work
>with to prepare for the event. This works very well the way it's run. And
no
>it's not a simple matter of selecting a Cray.
Who said the ICCA had to supply the computers?
Even if you're right that it works well as-is, that doesn't
mean it couldn't be even better. Let me give an example:
Suppose you and I were going to race from point A to
point B, and I showed up with my pickup truck (the fastest
thing I own). You were on foot (expecting the same from me),
but upon seeing how hopeless and pointless it would be to
race a truck you stroll across the street and rent a Ferrarri.
The race begins, but I don't even start my engine, again
because I see the pointlessness just as you had. Now,
who is faster, you or me? Well, we will never know so long
as we race on unequal terms....
- Greg Kennedy
Robert Sullivan
Any chance you'd be willing to explain what you mean? Thanks.
Regards,
Robert
--
Robert Sullivan, MD
bob4...@earthlink.net
mclane <mcl...@prima.de> wrote in message
news:7kmebp$u3s$1...@steve.prima.de...
mclane
>Hi,
>Any chance you'd be willing to explain what you mean? Thanks.
>Regards,
>Robert
junior and fritz and nimzo are PRE-processing programs.
They evaluate the root position BEFORE they build the search-tree.
and then they only evaluate the moves (not the position !!!!)
relative to this ROOT-evaluation.
But the deeper the tree, the more wrong is the evaluation because
pieces have moved and the relative evaluation cannot get the changes
right.
therefore the deeper they search the less accurate is their
evaluation.
therefore they play like blinds in the dessert trusting their search
depth.
but shredder evaluates the positions IN the tree. therefore shredder
plays more accurate.
this championship the fast-searchers did ANYTHING to have better
hardware to win the championship.
but hardware did not help. you can make a stupid program 1000 times
faster, it will be only 1000 times morte stupid= will make the
mistakes
1000 times faster.
1000 x sero is still sero
:-)))
(exaggerating much of course - but you see what i mean ? :-)))
>--
>Robert Sullivan, MD
>bob4...@earthlink.net
>mclane <mcl...@prima.de> wrote in message
>news:7kmebp$u3s$1...@steve.prima.de...
>> PMG <av...@Chicagonet.net> wrote:
>>
>>
>> Shredder is the top dog. right. because shredder has killed the
>> preprocessing-paradigm.
>>
>> no loss.
>> no special machine.
>> and less mistakes than the opponnents.
>>
>>
>> best wishes
>>
>> mcl...@prima.de
>>
best wishes
Greg Kennedy
>hardware to win the championship.
>but hardware did not help. you can make a stupid program 1000 times
>faster, it will be only 1000 times morte stupid= will make the
>mistakes
>1000 times faster.
>
>1000 x sero is still sero
> :-)))
>
>(exaggerating much of course - but you see what i mean ? :-)))
Maybe so... but DeepBlue whomped Garry with a
mediocre program and some monster hardware.
Perhaps 1,000,000,000 x 0.001 = pretty good!
- Greg Kennedy
mclane
> Maybe so... but DeepBlue whomped Garry with a
>mediocre program and some monster hardware.
>Perhaps 1,000,000,000 x 0.001 = pretty good!
And ?
what is your point ?
We talked about machines vs. machines.
Garry was badly advised.
He played weak.
> - Greg Kennedy
best wishes
Greg Kennedy
>"Greg Kennedy" <ches...@worldnet.att.net> wrote:
>
>
>
>> Maybe so... but DeepBlue whomped Garry with a
>>mediocre program and some monster hardware.
>>Perhaps 1,000,000,000 x 0.001 = pretty good!
>
>And ?
>what is your point ?
>We talked about machines vs. machines.
>
>Garry was badly advised.
>He played weak.
My point was in the part you snipped. I was
responding to a claim in the previous post, which
I quoted. That post stated something like:
infinite speed x zero intelligence program = zero.
My point was obvious: dumb program /= zero
intelligence, but something just a tad better. :-)
Not sure if Garry was badly advised, or if the
decision to play slop as Black was his own-
anybody know for sure?
You're right about his play. Weak. One problem he
had was related to your comments above in that, when
Garry played an aggressive move and grimaced and
squirmed in his chair and made ugly faces during that
match, DeepBlue just responded similarly:
>And?
>what is your point?
- Greg Kennedy
Dan Kirkland
> Maybe so... but DeepBlue whomped Garry with a
>mediocre program and some monster hardware.
>Perhaps 1,000,000,000 x 0.001 = pretty good!
DeepBlue womped Garry?
First, Garry was VERY poorly prepared for this match! He clearly
expected it to play somewhat different than what it did. As such
he played VERY poorly! (When was the last time you saw a World
Champion make three major mistakes in three different games in
a six game match?) Clearly Garry was not playing World Championship
level chess in the match! Yet Deep Blue only beat him by one game!
Basicly, Kasparov gave Deep Blue the match! And to show how much
confidence they had in their computer, IBM REFUSED a rematch! The
people who best knew the program (by far) knew they had no chance
of winning a rematch. And so they retired it!
Mediocre program? Says who?
Well, it doesn't matter who says, because this is absolute nonsense!
By being faster than any other program on the planet, Deep Blue can
also include MUCH more scoring than anybody else! While I don't
agree with Bob Hyatt about the actual strength of Deep Blue (which
we both agree is really an unknown, he thinks it may be good enough
to be top ten, I don't think it could ever EARN a top 100 ranking),
he has written some VERY good comments about the program. Search
for some of his VERY MANY posts on the subject (Deep Blue)!
Deep Blue is surely the most advanced program on the planet EVEN
from a programming point-of-view! But, of course there is no
way to really compare, as there is no hardware to speed up any
other program to Deep Blue's speed, and slowing down Deep Blue
would make much of it's more advanced scoring useless!
[Why is there so many who post what they know nothing about?]
dan
Dan Kirkland
> My point was in the part you snipped. I was
>responding to a claim in the previous post, which
>I quoted. That post stated something like:
>infinite speed x zero intelligence program = zero.
>My point was obvious: dumb program /= zero
>intelligence, but something just a tad better. :-)
And this part of your post was just plain WRONG!
Deep Blue is surely the most intelligent chess program on the
planet! Because Deep Blue is so fast, many (such as yourself)
assume that they sacrificed intelligence for speed. This is
NOT the case! Because of the extra speed they could add MORE
intelligence!
In Deep Blue, IBM had the fastest AND the most intelligent
program ever made! And yet they only beat a VERY poor playing
Kasparov by one game! And they clearly did not have the confidence
to give Kasparov a rematch!
I don't know where you get your info, but there was NOTHING in your
post that was correct (machine vs machine, OR machine vs human, or
anything else!)!
dan
Ben Leostein
I think researchers and programmers are going to have to go in another
direction as you suggested in order to break down chess a little.
Since personal computers are so far from Deep Blue's components and speed, and
since Kasparov wasn't prepared for the match, it's very unlikely that a
personal computer will be able to replicate this.
But... I believe within the next decade perhaps new search methods will be
worked upon to offer much better results and playing strength for computers,
and who knows, maybe even sooner. :-)
Shannon's two methods are extremely old and there is a lot more promosing
research that "can" be done into this, many much more advanced types of
searches that a program can do to eliminate lines, and so forth.
It's just the beginning, and I'm excited. :-)
Regards,
Ben
Wolfram Bernhardt
> In Deep Blue, IBM had the fastest AND the most intelligent
> program ever made!
I don't think so. The fastest - yes, of course.
The most intelligent - no.
It's clearly known, that most of DeepBlue's moves (94 %) can be found by
shredder, fritz & co - in the same time.
So the difference between DeepBlue and todays PC-Chess-programs is not
too great and I don't doubt that PC-program are more sophisticated by
far. By very far.
If Fritz would be implemented on DeepBlue (if possible. Maybe the chips
had to be redesigned) this would be a much more better player.
wolfram
Phil
top speed) ....Why in the game it won .... it made a choice of a DRAWING
LINE after some GM analyses .....
_______________________________________________________________
Now that computers play Chess at Grand-Master Level ...
.... Its about time ... they learn to walk and move the pieces on the board
ALONE....
_______________________________________________________________
Dan Kirkland <kirk...@ee.utah.edu> a écrit dans le message :
7ksub3$5...@ee.utah.edu...
Mig
This ignores the fundamental difference between the two: parallel
processing and the advantages it confers. You can't just plunk Fritz
into a big machine and say go. Deep Blue had hundreds of knowledge
variables that no PC program can afford to look at because DB got them
"for free" processing-wise. If they came up with another variable to
consider they could just toss it in at almost no cost, whereas every
PC program has to worry about slowing to a crawl with every new piece
of knowledge added. DB was incredibly smart as far as total number of
pieces of knowledge factored.
For example, I was terrified to find out that DB considered king
safety in three board areas (q-side, center, k-side) for both sides on
every move until castling! Whereas a PC program could only consider
that when considering the actual castling move in the tree, or
sacrifice huge resources in checking it, DB kept it in mind for both
players all the time. (Just like a human does, really. A master will
play a wing pawn advance to discourage castling into that side,
something DB did, to the amazement of all.)
The question isn't that, it's accuracy of eval. HIARCS or even Fritz,
or any PC program, might just be tuned better and therefore make
better (or similar, re: 94%) decisions in the same time just because
it evaluates the positions better, although with far less total
information. It's knowledge capacity/quantity vs. knowledge
accuracy/quality. DB could consider every single factor they could
think of, while a PC program has to focus on the crucial elements, but
if the eval is out of tune it doesn't matter how many elements are in
it.
Saludos, Mig
NEW EMAIL ADDRESS: mig@fibertelDOTcomDOTar
___________________
CHESS MADNESS!!
More fun than you've ever had with your hand on your mouse.
http://www.chessbase.com NEW JUNE 6! DOS HERMANAS AND SARAJEVO!
-- CHECKMATE 101 IS BACK! Now up at the premiere online
chess training site: ChessWise! http://www.chesswise.com
-- Mig on Chess *DAILY TOURNAMENT COVERAGE OF ALL MAJOR EVENTS!*
http://www.chesscenter.com/twic/twic.html
------------------
Michael "Mig" Greengard * Remove SBLOCK to e-mail me *
Greg Kennedy
>>responding to a claim in the previous post, which
>>I quoted. That post stated something like:
>>infinite speed x zero intelligence program = zero.
>>My point was obvious: dumb program /= zero
>>intelligence, but something just a tad better. :-)
>
>And this part of your post was just plain WRONG!
You have wrongly assumed that I equate "dumb program"
with DeepBlue's program- I don't. I always use the term
"mediocre" to describe DeepBlue's program. My idea of a
dumb program is Boris Diplomat.
>Deep Blue is surely the most intelligent chess program on the
>planet! Because Deep Blue is so fast, many (such as yourself)
>assume that they sacrificed intelligence for speed. This is
>NOT the case! Because of the extra speed they could add MORE
>intelligence!
Wrong again. I assumed nothing- I replayed the games!
DB played some weak moves. On that basis, I labelled the
program "mediocre." I have already pointed out these weak
moves on this forum. One example: 1e3 d5 2c4, and now
came the lemon, ...dc.
>In Deep Blue, IBM had the fastest AND the most intelligent
>program ever made! And yet they only beat a VERY poor playing
>Kasparov by one game! And they clearly did not have the confidence
>to give Kasparov a rematch!
You are making assumptions. I agree that DB was the fastest,
but you are wrong about the program. It was weak in certain
openings (see above) and in spite of all its speed, did not even
try to avoid a draw when clearly ahead in the game where Garry
resigned in a drawn position. It also exhibited the classic horizon
effect in the game Garry won, by voluntarily sac'ing and then, too
late, seeing that its compensation was not there. If Garry had
played alot better, we probably would have seen even more
serious flaws in the program. As it was, we saw enough to
determine that the program needed serious work, or else the
power of the killer hardware would be half wasted.
>I don't know where you get your info, but there was NOTHING in your
>post that was correct (machine vs machine, OR machine vs human, or
>anything else!)!
This is almost impossible, for even a blind squirrel
finds an acorn, now and then. :-)
- Greg Kennedy
Ben Leostein
play very intelligently.
I wonder why IBM didn't try the same approach to Deep Blue, try to give it tons
and tons of chess information to work with (especially cool attacking
information) and with those ice-cold calculations, it could create some really
beautiful games. :-)
Rolf Tueschen
<19990626035425...@ng-fz1.aol.com>:
>You know the chess program CSTAL uses a lot of chess knowledge built into it to
>play very intelligently.
>I wonder why IBM didn't try the same approach to Deep Blue, try to give it tons
>and tons of chess information to work with
You wonder why??
Easy one. Just because, er, because their team doesn't have a chess
player like Chris Whittington. Joel Benjamin came into the business only
after the program 'stood' in principle. He had to deal with the nonsense
of the machine and exclude the lines with the most significent crap.
(Fortunately Kasparov had no chance to train on the DB machine. Would
have been a piece of cake for him to find the holes in Joel's prep.)
Just my two dollars. If we will have another medium for silizium, you
can bet that then the smart approach will suddenly outplay the pure
brute force machine because of its rising momentary smartness (without
all the overheating problem with silizium). The pure depth related BF
approach will never be able to be game with our ordinary human class. A
BF horizonal limitation will always result into dumb or, if it's
preferred, 'mediocre' play ... whether it's at ply 13 or 16 or 20.
Choose any number between 10 and 50. :)
It's so difficult and at the same time so easy. All the BF guys suffer
from a logical decease -- with all respect. They do as if chess had
already been solved. In that case, the BF would always guarantee the
best play. I have read the following statement. Strategy is something
for people who can't calculate deep enough tactically. And again the
logical decease comes into play when it has been said that therefore the
machines don't need smartness but only depth.
What do you think is the reason for the DB retirement?
Easy one again. Even after their "win" against Kasparov, the really
smart guys in the business knew what Greg Kennedy has explained. DB
played a lot of crap. And in a period of say the next 5 years the IBM
team couldn't make it a lot better. It's funny, but a "win" that was
attained with 'psyching out' tricks is not very convincing. Often a
"win" is a loss in reality. Don't forget, chessically DB didn't outplay
Kasparov. Kasparov made all the decisions by himself. (Forfeiting a
drawn game, confusing a forced move order, choosing a passive and bad
opening etc.)
The title of this thread is another proof for the problem. SHREDDER
wasn't the hottest and strongest big iron in the pool! But it still won
the championships.
Is there a bell ringing somewhere? Excuse the minimum of sarcasm being
used here ...
:) [smiley set just in case of self-defense needed] :)
["This is a place where you have to excuse the strength of your ideas."
Anonymous author!]
mclane
>You know the chess program CSTAL uses a lot of chess knowledge built into it to
>play very intelligently.
>I wonder why IBM didn't try the same approach to Deep Blue, try to give it tons
>and tons of chess information to work with (especially cool attacking
>information) and with those ice-cold calculations, it could create some really
>beautiful games. :-)
When i have a car having 320 PS, why should i make considerations
about the price of ONE ltr super-fuel ?
When i have such a fast car, money is not important to me.
Also i am not interested in nature arround me.
When you drive a car with 46PS, that only needs 5ltr/100km
and is nice to the nature arround, THERE it makes sense to put
knowledge into it.
Of course the ones having 320PS could do the same ideas, but why
should
they think about it, when there is no reason to do so.
They build big machines and they think: oh - we don't need tons of
knowledge when we come 4 times or 100 times deeper because of
our hardware, we can spare the knowledge and concentrate on tuning the
hardware.
Shredder runs on a 46PS machine, and tries to make the best out of it.
as you can see: it works.
best wishes
Dan Kirkland
>> In Deep Blue, IBM had the fastest AND the most intelligent
>> program ever made!
>
>The most intelligent - no.
>It's clearly known, that most of DeepBlue's moves (94 %) can be found by
>shredder, fritz & co - in the same time.
Which means what? That Deep Blue only plays a BETTER move about
6% of the time! This is not at all suprising. Keep in mind that
most good chess programs make very good moves MOST of the time!
>So the difference between DeepBlue and todays PC-Chess-programs is not
>too great and I don't doubt that PC-program are more sophisticated by
>far. By very far.
>If Fritz would be implemented on DeepBlue (if possible. Maybe the chips
>had to be redesigned) this would be a much more better player.
Well, this is NOT possible. And, I am sure you are wrong.
Deep Blue is surely MUCH more sophisticated than ANY of today's
PC programs! The fact that it doesn't play perfect chess only
proves it is not perfect. It does NOT prove, in any way, that
it is somehow lessor than the PC based programs!
Again, showing that Deep Blue does NOT play perfect chess only
shows that Deep Blue is NOT perfect! It, IN NO WAY, proves that
Deep Blue is somehow a lessor program than the PC based programs!
What little evidence exists clearly shows that Deep Blue plays
better chess than any other chess program/computer! Ane you have
NO evidence to the contrary! All evidence that ANY of us have
clearly show that Deep Blue is MUCH more sophisticated than ANY
other chess program EVER!
Now, I am NOT saying that Deep Blue is perfect. Far from it!
But for every lessor move that you can find that Deep Blue
makes, at least twice that many can be found in the play of any
other chess program/computer!
Deep Blue is less sophisticated than PC based programs?
YOU ARE DREAMING! GET LIFE and WAKE UP!
dan
Dan Kirkland
>>> My point was in the part you snipped. I was
>>>responding to a claim in the previous post, which
>>>I quoted. That post stated something like:
>>>infinite speed x zero intelligence program = zero.
>>>My point was obvious: dumb program /= zero
>>>intelligence, but something just a tad better. :-)
>>
>>And this part of your post was just plain WRONG!
>
>
> You have wrongly assumed that I equate "dumb program"
>with DeepBlue's program- I don't. I always use the term
>"mediocre" to describe DeepBlue's program. My idea of a
>dumb program is Boris Diplomat.
And you somehow think that there are stronger programs out
there somewhere? PLEASE!
>>Deep Blue is surely the most intelligent chess program on the
>>planet! Because Deep Blue is so fast, many (such as yourself)
>>assume that they sacrificed intelligence for speed. This is
>>NOT the case! Because of the extra speed they could add MORE
>>intelligence!
>
>
> Wrong again. I assumed nothing- I replayed the games!
>DB played some weak moves. On that basis, I labelled the
>program "mediocre." I have already pointed out these weak
>moves on this forum. One example: 1e3 d5 2c4, and now
>came the lemon, ...dc.
Oh gee, Deep Blue does NOT play perfect chess! Wow, what a
suprise! (NOT!!!)
>>In Deep Blue, IBM had the fastest AND the most intelligent
>>program ever made! And yet they only beat a VERY poor playing
>>Kasparov by one game! And they clearly did not have the confidence
>>to give Kasparov a rematch!
>
>
> You are making assumptions. I agree that DB was the fastest,
>but you are wrong about the program. It was weak in certain
>openings (see above) and in spite of all its speed, did not even
>try to avoid a draw when clearly ahead in the game where Garry
>resigned in a drawn position. It also exhibited the classic horizon
>effect in the game Garry won, by voluntarily sac'ing and then, too
>late, seeing that its compensation was not there. If Garry had
>played alot better, we probably would have seen even more
>serious flaws in the program. As it was, we saw enough to
>determine that the program needed serious work, or else the
>power of the killer hardware would be half wasted.
First, a weak opening may be because of a human prepared opening
and have NOTHING to do with the program itself!
Second, ALL chess programs will exhibit the classic horizon effect
now and then! It is unavoidable! (Unless the program can see the
end of the search tree!)
Third, ALL that you point out above only goes to show that Deep
Blue is less than perfect! And yes, surely even a lessor chess
player than Kasparov at his best! But NONE of this shows that
Deep Blue is somehow a lessor chess player than other chess
computer programs.
Even with his very poor play, Kasparov would surely have beat
any OTHER chess program!
If Deep Blue played enough matches, I think enough holes in it's
play would be found so that ANY of the world's top 100 or so
chess players would be able to beat it! But this does NOT mean
it is weaker, or somehow a lessor player than ANY of the other
chess programs!
>>I don't know where you get your info, but there was NOTHING in your
>>post that was correct (machine vs machine, OR machine vs human, or
>>anything else!)!
>
>
> This is almost impossible, for even a blind squirrel
>finds an acorn, now and then. :-)
Yes, well maybe you need to make a few hundred more posts before
you chance to get something right? ;)
Quite frankly, I do not understand the reasoning of the few of
you who think that Deep Blue is somehow a lessor program than
some of the others. (In fact, I have trouble seeing any reasoning
at all!)
For one, the vast majority of chess experts (grand masters and such)
think that Deep Blue played VERY well (much better than any other
chess computer/program has to date).
Also, (at least in the past...) search depth was MUCH more
important in machine vs machine games than in man vs machine
games (although, with shredder beating some much faster machines,
maybe this is also changing in machine vs machine games?).
For example, take a good fast program, put it on two different
computers where one sees two ply or more than the slower one.
The faster one will have a large advantage over the slower one.
But play both against human players, and the advantage is not
so great (unless the slower has a VERY shallow search!).
The same applys, to a lessor degree, with two different programs
of about equal strength.
Add to this that Deep Blue looses some of it's speed advantage
because of a parallel search.
Add to this that Deep Blue has MANY more scoring variables than
ANY other computer chess program...
And so on and so forth...
In the end, the ONLY conclusion a knowledgeable person could
come to is that the reason Deep Blue was able to beat Kasparov
had as much (or more) to do with it's more advance scoring
as it did because of of it's greater search depth!
Any other conclusions are nothing but misunderstandings and/or
silly daydreamings!
In fact, the BEST testimony to Deep Blue's more advanced scoring
comes from Kasparov himself! Where, in game two, he made a move
to a position that he thought that NO computer could possibly
understand! Yet, Deep Blue made the proper reply (much to Kasparov's
suprise) clearly showing it's understanding of the position!
(In fact, Kasparov was so suprised that he said it threw him off
for the rest of the match!)
So, if you think Deep Blue is a somewhat weak or mediocre program,
then maybe you need to do a little more research on the subject!
Or, maybe you just need to quit daydreaming!
dan
Akorps
Hmm, I thought that was an indeterminate
form. Depends on whether the epsilon
of stupidity approaches zero faster than
the speed of the calculation approaches
infinity, etc etc
"despite its internal rottenness, the Byzantine Empire endured a thousand
years, due to a stable military organization made possible by the codification
of the art of war in the manuals 'Strategicon' and 'Tactica'"
-JFC Fuller (Armament and History)
Greg Kennedy
>> You have wrongly assumed that I equate "dumb program"
>>with DeepBlue's program- I don't. I always use the term
>>"mediocre" to describe DeepBlue's program. My idea of a
>>dumb program is Boris Diplomat.
>
>
>And you somehow think that there are stronger programs out
>there somewhere? PLEASE!
I believe that the most finely tuned, strongest programs
for PCs, if given the speed of DB, would be stronger, yes.
But not BorisDiplomat! :-)
>>>Deep Blue is surely the most intelligent chess program on the
>>>planet! Because Deep Blue is so fast, many (such as yourself)
>>>assume that they sacrificed intelligence for speed. This is
>>>NOT the case! Because of the extra speed they could add MORE
>>>intelligence!
>>
>> Wrong again. I assumed nothing- I replayed the games!
>>DB played some weak moves. On that basis, I labelled the
>>program "mediocre." I have already pointed out these weak
>>moves on this forum. One example: 1e3 d5 2c4, and now
>>came the lemon, ...dc.
>
>Oh gee, Deep Blue does NOT play perfect chess! Wow, what a
>suprise! (NOT!!!)
No one expects perfect chess, just mistakes which are
not obvious to patsers like me. Fair enough?
snip
>First, a weak opening may be because of a human prepared opening
>and have NOTHING to do with the program itself!
Unless that was programmed into the computer,
which it was. If Joel told DeepBlue to hang its
queen in the opening, then DeepBlue is weak in
the opening, because it hangs its queen. The origin
of the blunder is human, but it is incorporated into
the program.
>Second, ALL chess programs will exhibit the classic horizon effect
>now and then! It is unavoidable! (Unless the program can see the
>end of the search tree!)
Of course. But the power and speed of DB allowed the
possibility of compensating for this defect by searching ahead
for perps and reps. They didn't even bother. Result: mediocrity.
>Third, ALL that you point out above only goes to show that Deep
>Blue is less than perfect! And yes, surely even a lessor chess
>player than Kasparov at his best! But NONE of this shows that
>Deep Blue is somehow a lessor chess player than other chess
>computer programs.
I cannot prove it- the PC hardware is not available yet.
Just my opinion. Neither can you prove I am wrong, for
similar reasons. Just your opinion.
>Even with his very poor play, Kasparov would surely have beat
>any OTHER chess program!
Maybe. But he was playing much weaker than normal...
>>>I don't know where you get your info, but there was NOTHING in your
>>>post that was correct (machine vs machine, OR machine vs human, or
>>>anything else!)!
>>
>> This is almost impossible, for even a blind squirrel
>>finds an acorn, now and then. :-)
>
>Yes, well maybe you need to make a few hundred more posts before
>you chance to get something right? ;)
"Evacuate, in our moment of triumph?"
"I think you underestimate their chances."
- from STAR WARS
>Add to this that Deep Blue has MANY more scoring variables than
>ANY other computer chess program...
Quantity? What about QUALITY?
>So, if you think Deep Blue is a somewhat weak or mediocre program,
>then maybe you need to do a little more research on the subject!
>Or, maybe you just need to quit daydreaming!
Zzzzzzzzzzzzz- saywha? I mustof dozed off...
Where can I do more research on comparing DB's
program to say, Hiarcs or Rebel or ?
- Greg Kennedy
Rolf Tueschen
>>> In Deep Blue, IBM had the fastest AND the most intelligent
>>> program ever made!
>>
>>I don't think so. The fastest - yes, of course.
>>The most intelligent - no.
>>It's clearly known, that most of DeepBlue's moves (94 %) can be found by
>>shredder, fritz & co - in the same time.
>Which means what? That Deep Blue only plays a BETTER move about
>6% of the time! This is not at all suprising. Keep in mind that
>most good chess programs make very good moves MOST of the time!
>>So the difference between DeepBlue and todays PC-Chess-programs is not
>>too great and I don't doubt that PC-program are more sophisticated by
>>far. By very far.
>>If Fritz would be implemented on DeepBlue (if possible. Maybe the chips
>>had to be redesigned) this would be a much more better player.
>Well, this is NOT possible. And, I am sure you are wrong.
>Deep Blue is surely MUCH more sophisticated than ANY of today's
>PC programs!
On the contrary, I am absolutely sure that you are deadly wrong. Let me
give my explanation. You don't need to react, this is all too clear,
taht it can't be discussed.
And let me add that I don't prove the possible but only that part of the
question that will prove that you are wrong alltogether.
Let's assume that DB is perhaps some factors faster than the micros. Do
you agree?
Now let's take for instance the first game of the 1997 match. Or if you
prefer the second game.
My argument goes like this. In the first game DB played a few terribly
weak, if you prefer 'stupid' moves. I guess that FRITZ would play them
too.
So --- since FRITZ is so much slower, it's obvious that FRITZ is
potentially much smarter.
But honestly I'll doubt that FRITZ would play exactly that crap DB
played in its first game.
I think that the actual micros, not only FRITZ, have a smarter content
than DB. They _must_ have it since they are so much slower.
And finally. I said it already somewhere else.
_IF_ IBM/DBteam hadn't realized that very obvious fallacy, they would
NOT have stopped the DB project - I can tell you that. But look at the
first game and try to give a good reason to invest further money into
that thing ...
In other words. The brute force approach has come to an end. In a decade
or so, other hardware hopefully will allow the breakthrough of smarter
approaches.
...
Ben Leostein
>>Deep Blue is surely MUCH more sophisticated than ANY of today's
>>PC programs!
From what I understand, the program used by IBM in Deep Blue is a very basic
chess program that uses unbelievable powers of calculation.
This is obvious when the engine Wchess (the engine used by Sierra in
Powerchess) was able to draw Deep Blue.
Deep Blue's programming isn't that good, it just simply evaluate's a chess
position but there isn't that much to the program itself, just a "mega"
calculation machine.
Now, give Fritz that same calculation power that Deep Blue has, and you might
have a computer that no human will "ever" defeat. :-)
Ben