Holy cow - have you seen the latest ELO-values for chess engines?

[Skybird]

http://ssdf.bosjo.net/list.htm

Beyond 3200 ELO - that is something. Note that the SSDF already uses the international ELO-calculation that differs form that used in the American chess association (which constantly raises ELOs, especially with computer, by around 200 points). In USCF-ELO, the top list entry maybe would read around 3400-3450 ELO.

for comparison: the strongest human ELO value ever scored was Gary Kasparov with 2851 (the majority of the chess world sees him as the strongest player ever in the history of chess). Beginners rate at around 1000-1300 ELO, regular players at around 1500-1800 ELO, club and tournament players (non-professional up to 2200-2300) - beyond that the official master-ranks begin.

However, the SSDF list may be the most popular and oldest of them all, but even this loist has seen repeated down-corrections in it's history, so the spectacular values should not blind anybody - that they are gong constantly upwards has it'S reason in mathematical implications of the method, and does not necessarily reflect total playing strength comparable to that of the the human's ELO-list. I assume that if you substract 400-600 points, maybe even more (who knows) you probably are closer to the truth. but a chess engine producing a strength of let's say 2750 ELO still would be a total beast to play for any grandmaster in the world. The list of players having an ELO beyond 2700, is short.

Hal, play "Misty" for me...

[baggygreen]

erm... sky?

whats an ELO?

[kranz]

Sky-about machines-do you remember the one which Napoleon played with?

[Skybird]

Originally Posted by baggygreen

erm... sky?

whats an ELO?

Precisely: a man. Practically: a measure unit.

http://en.wikipedia.org/wiki/Elo_rating_system

[Skybird]

Originally Posted by kranz

Sky-about machines-do you remember the one which Napoleon played with?

Yes, why? It was a mechanism operated from inside by a midget. It's called "The Turk". Napoleon lost - different to what people say, by reports from his time he was a lousy chess player.

[kranz]

yeah, a midget inside. Actually the designer claimed that he "taught" the machine how to make moves etc but the fire revealed the truth. I also remember kasparov's games with deep blue. After he lost he said that computer could not have made such a move...

[Skybird]

Originally Posted by kranz

yeah, a midget inside. Actually the designer claimed that he "taught" the machine how to make moves etc but the fire revealed the truth. I also remember kasparov's games with deep blue. After he lost he said that computer could not have made such a move...

Deep Blue is history. Every decent chessprogram today like Rybka, Hiarcs, Shredder, Fritz, plays much stronger, and with far more positional understanding and strategic - instead of just tactical - competence. To have programmed engines so that they can already adapt to coming situations whose complete move variations are beyond their calculation horizon is probably the greatest acchievement of chess programming. IMO it showed to be more decisive than just scaling up the hardware potency. The two names who bet their money on this kind of programming before anyone else were Mark Uniacke from Britain (Hiarcs), and Netherland's Ed Schroeder (the Rebel programs and their many clones). Schroeder has withdrawn form business, but uniacke is still there, and hiarcs is my preferred engine. I can't differ it from a human player. But probabaly is true for all top engines today.

Until the early-90s I was able to beat everything computer-related that you could buy on the public market, wether it be board-computer or PC-amiga-atari-related, if only I took my time and analysed in mail-chess fashion. These times are long over. + As far as players of my level are concerned, computers have become the masters. You can safely buy which you like best by the looks, the playing strength of the engine must not worry you and is no argument in the buying decision anymore - chances are it will slaughter you anyway if you do not use minimal calculation times, handicap levels and artificial brakes. Or you play over hours and days, and anaylse as intensive as in correspondence chess.

[goldorak]

Originally Posted by Skybird

Until the early-90s I was able to beat everything computer-related that you could buy on the public market, wether it be board-computer or PC-amiga-atari-related, if only I took my time and analysed in mail-chess fashion. These times are long over. + As far as players of my level are concerned, computers have become the masters. You can safely buy which you like best by the looks, the playing strength of the engine must not worry you and is no argument in the buying decision anymore - chances are it will slaughter you anyway if you do not use minimal calculation times, handicap levels and artificial brakes. Or you play over hours and days, and anaylse as intensive as in correspondence chess.

I have only Chessmaster X on the pc and can't beat it.
Thats why I prefer to play against dedicated chess computers.
Its way more fun. At least from time to time I can say I won, something I will never be able to do with Chessmaster at the highest level of difficulty.
So for me 1990 dedicated computer elo level is more than enough.

What I would like to see on the other hand, is chess engines becoming more intelligent not only in terms of brute force but in term of strategic thinking. Most of these programs are very powerful because of the immense opening libraries they have. Ever wondered what happens to a chess program that exits its opening library on the first move or doesn't use one? Its a massacre for it. The real goal should be to design a chess engine that doesn't base its knowledge on opening libraries or on endgame tables. Brute force can only take you so far.

[AntEater]

Hey, I've allways told the people who beat me in chess that it is unrealistic and they should try to beat me in Combat Mission 2
Now L'Empereur himself lends credibility to my argument...


Btw, the saying "getürkt" or "einen Türken aufbinden"; lit. "to turk somebody" or "to tie a turk on someone's back", colloquial german for hoaxing or decieving someone does not come from the country Turkey or the turkish people but rather from the mechanical turk.
http://en.wikipedia.org/wiki/Mechanical_Turk

[Skybird]

Originally Posted by goldorak

What I would like to see on the other hand, is chess engines becoming more intelligent not only in terms of brute force but in term of strategic thinking. Most of these programs are very powerful because of the immense opening libraries they have. Ever wondered what happens to a chess program that exits its opening library on the first move or doesn't use one? Its a massacre for it. The real goal should be to design a chess engine that doesn't base its knowledge on opening libraries or on endgame tables. Brute force can only take you so far.

But that'S what I was talking about. And such engines are already there. Just compare the knots/second value of hiarcs 10 (relatively low) with that knots/sec. of an older Fritz 8 several times as high. Both engines are separated by roughly two years i think. Hiarcs is much stronger, but does so with less brute force calculation, and mores smart selective algorithms and better positions evaluation. Engines of the last let'S say 3 years - really play damn strong in the positional and stategic area as well. There is a reason why they score top reslts when being allowed in human tournaments as well, and why they are capable to beat even grandmasters. the real strong horses also come as dedicated "Deep Engine X" vesions, and make optimized use of dual cor and multi-processor systems, which then again gives them brute force power.

Chessmaster does not place in the top of the engine lists. Try one of the contemporary chessbase engines, Rybka, Shredder, Hiarcs, and currently a bit behind, Fritz. You also get the - imo - better looking GUI.

However, the 80s and up to the mid-90s, that was the golden era of board chesscomputers, and I miss that time. I have a small collection of board computers, some of which I bought much later, as a collector, asome real rarities amongst them, and most-wanted ones. Unfoprtunately I have not so much room and thus must store most of them, and the large ones, in the cellar. A Fidelity Avantgarde, and a Sargon 2.5 ARB, are two of the highlights, but both have small technical problems now, unfortunately. most I use a trustworthy Mephisto Exclusive with MM-IV and HG440. It passed the test of time, and imo is the most beautiful chesscomputer desogn ever. the wooden sensor boards of the Exclusice-series, field size 4x4 cm, must be the most successful design ever in board-chesscomputers. For sentimental reasons I also like the Fidelity Sensory Champion very much.

[Skybird]

Originally Posted by AntEater

Hey, I've allways told the people who beat me in chess that it is unrealistic and they should try to beat me in Combat Mission 2
Now L'Empereur himself lends credibility to my argument...


Btw, the saying "getürkt" or "einen Türken aufbinden"; lit. "to turk somebody" or "to tie a turk on someone's back", colloquial german for hoaxing or decieving someone does not come from the country Turkey or the turkish people but rather from the mechanical turk.
http://en.wikipedia.org/wiki/Mechanical_Turk

Also, die Redewendung "jemandem einen Türken aufbinden" habe ich noch nie gehört. Mit Bären kenne ich das nur.

[goldorak]

[QUOTE=Skybird]

Originally Posted by goldorak

However, the 80s and up to the mid-90s, that was the golden era of board chesscomputers, and I miss that time. I have a small collection of board computers, some of which I bought much later, as a collector, asome real rarities amongst them, and most-wanted ones. Unfoprtunately I have not so much room and thus must store most of them, and the large ones, in the cellar. A Fidelity Avantgarde, and a Sargon 2.5 ARB, are two of the highlights, but both have small technical problems now, unfortunately. most I use a trustworthy Mephisto Exclusive with MM-IV and HG440. It passed the test of time, and imo is the most beautiful chesscomputer desogn ever. the wooden sensor boards of the Exclusice-series, field size 4x4 cm, must be the most successful design ever in board-chesscomputers. For sentimental reasons I also like the Fidelity Sensory Champion very much.

I agree with you, the 80-90 were the golden era of wooden chess computers.
I also have some wooden chess computers, a mephisto exclusive with the polgar program and a mephisto munich with the vancouver program.
I was wondering wether to buy the novag citrine but I'm yet undecided.
The one chess computer I would really like to have is the mephisto phantom.