Kaufman on Rybka and Fischer

[Alexander Schmidt]

Komodo 1.0 ?
You should use later versions.

No, I am kinda out of buisnes

Gandalf 6 has special king attack algorithms.

The test positions are not only king attacks, but also endgame positions and strategic piece maneuveres.

[hgm]

Maybe "inspired by Fischer" is not far away from the truth, but this doesn't mean the engine plays like Fischer. So the statement is indeed what some would call "marketing bs".

What statement? The statement I have seen doesn't make a claim that the engine plays like Fischer:

“Since I am mainly responsible for the evaluation function in Rybka, and since Fischer was the No. 1 influence on my own chess career, a lot of Rybka’s evaluation can be said to be second-hand knowledge from Fischer”

(Larry Kaufman, Chess, April 2008).

If players are guided by certain strategical ideas, you can be sure a GM-level player would notice those ideas much better than a patzer would.

[syzygy]

Larry's statement is innocent but meaningless.

Unless Larry has actually discussed chess with Fischer in person or maybe in writing (which I assume not to be case), Fischer's influence on his actual chess knowledge (rather than Fischer just being a motivational factor for studying harder, which clearly does not translate into "second-hand knowledge from Fischer") is limited to Larry's study of Fischer's games.

Larry, like any other strong player, must have studied games of hundreds of GMs and studied the writings of dozens of chess authors. However Larry might subjectively feel about this, his chess skills certainly are in no way mainly borrowed from Fischer.

The statement is innocent and amusing and that's it.

It would be different if Larry had learned the basic rules of chess from reading "Bobby Fischer Teaches Chess".

Btw, the evaluation function of my own patzer engine can be said to be second-hand knowledge from Max Euwe ("Praktische Schaaklessen", not "Oom Jan leert zijn neefje schaken").

[lkaufman]

Larry's statement is innocent but meaningless.

Unless Larry has actually discussed chess with Fischer in person or maybe in writing (which I assume not to be case), Fischer's influence on his actual chess knowledge (rather than Fischer just being a motivational factor for studying harder, which clearly does not translate into "second-hand knowledge from Fischer") is limited to Larry's study of Fischer's games.

Larry, like any other strong player, must have studied games of hundreds of GMs and studied the writings of dozens of chess authors. However Larry might subjectively feel about this, his chess skills certainly are in no way mainly borrowed from Fischer.

The statement is innocent and amusing and that's it.

It would be different if Larry had learned the basic rules of chess from reading "Bobby Fischer Teaches Chess".

Btw, the evaluation function of my own patzer engine can be said to be second-hand knowledge from Max Euwe ("Praktische Schaaklessen", not "Oom Jan leert zijn neefje schaken").

I did discuss chess in person with Fischer when he adjudicated a tournament game I had with Jimmie Sherwin in the mid 1960s, but of course that is insignificant. I watched him play 70+ simul games in 1964 in D.C., following him around to observe. Due to learning chess partly from Chess Life, in which Fischer had a regular column back then, he was the biggest influence on me, followed by Reshevsky. When I first became a master in 1966 I was still heavily influenced by these two. Of course by the time I became a GM in 2008 I had learned from hundreds of GMs as you say, so no one was a dominant influence on me, but one never forgets his childhood influences. I don't think any engine plays very much like any particular GM, but if one had to pick an engine that was influenced more than others by Fischer, it should be one that used an evaluation I was involved with, i.e. Komodo, Rybka, or any of the Ippolit derivatives that were heavily based on Rybka. That leaves only Stockfish among top engines as not based on my eval ideas.

[MikeGL]

Lucas is simply right so why deny it?

This forum is not about being right or wrong. It's about out-trolling each other.

Indeed...

Maybe I can contribute something factual:

https://chessengines.wordpress.com/the-fischer-test/

Let's say it is statistically not verificated that Komodo is more influenced by Fischer than other engines.

Let the shitstorm begin...

Interesting URL link. Gandalf is not in my collection yet,
I will try to download that now.

I remember around 1999 I have downloaded Crafty (the current version at that time) and was excited to learn about its book features. I learned that one can build a crafty book using a PGN file. So I downloaded all of Fischer's games (using dial-up), and also exported all chessbase games of Fischer into a PGN, then filtered all of Fischer's won games with both white and black. Then built my crafty book using those won games of Bobby Fischer. That Crafty collection (version 16 if i remember correctly) played the openning just like Fischer! Of course that crafty book was gone including my collections of crafty sources and binary due to instability (and malwares) of Win95 during those days. If that Win95 machine is alive until now, it will probably perfect this Fischer test in that link.

[AdminX]

Komodo 1.0 ?
You should use later versions.

Gandalf 6 has special king attack algorithms.

I agree with H.G.Muller, the comment was in reference to Rybka. I did test Komodo 9.02 on his Fischer Test and it scored 69%.

[Alexander Schmidt]

I did test Komodo 9.02 on his Fischer Test and it scored 69%.

69% is much but you cannot simply compare the percentage of solved positions under different testing conditions.

Because modern engines search a lot deeper they have a bigger chance to solve a position. So the only way is to use a fixed time without truncation.

[Alexander Schmidt]

What statement? The statement I have seen doesn't make a claim that the engine plays like Fischer:

“Since I am mainly responsible for the evaluation function in Rybka, and since Fischer was the No. 1 influence on my own chess career, a lot of Rybka’s evaluation can be said to be second-hand knowledge from Fischer”

(Larry Kaufman, Chess, April 2008).

Honestly, HGM, you're a beancounter

Im am sure many people understand "second-hand knowledge from Fischer” as a hint that the engine will play somehow like Fischer.

If players are guided by certain strategical ideas, you can be sure a GM-level player would notice those ideas much better than a patzer would.

And I am sure that no GM would pick out Rybka or any other engine from a computer chess database to be a Fischer-Style-Engine.

[hgm]

Well, so many people are wrong. Tell me something I did not know...

[michiguel]

What statement? The statement I have seen doesn't make a claim that the engine plays like Fischer:

“Since I am mainly responsible for the evaluation function in Rybka, and since Fischer was the No. 1 influence on my own chess career, a lot of Rybka’s evaluation can be said to be second-hand knowledge from Fischer”

(Larry Kaufman, Chess, April 2008).

Honestly, HGM, you're a beancounter

Im am sure many people understand "second-hand knowledge from Fischer” as a hint that the engine will play somehow like Fischer.

If players are guided by certain strategical ideas, you can be sure a GM-level player would notice those ideas much better than a patzer would.

And I am sure that no GM would pick out Rybka or any other engine from a computer chess database to be a Fischer-Style-Engine.

Following the concept of this thread, Gaviota has knowledge from O. Panno but it does not play like him. Panno would play like Gaviota if he could search like Gaviota Anyway, this is all a bit of a stretch.

Miguel