Longer Article

[peter]

Hi Anthony!

If I get you right, you say Vasik Rajlich didn't do anything else than tuning the fruitcode translated to bitboard by ultrashort games?
If this is so, Elo measured by rating lists don't say anything else than ultrashort games do, right?
Even if I think so by myself for quite a time, and if we, the unwashed masses, have been played for suckers for several years now just as for Celo (Elo of computerchess) (by the way, you could have told us so a little bit sooner, couldn't have you, knowing it since Mexico? ) wasn't it a simple thing then, to judge from ICGA's side, what to do now?
I mean, if Rybka is just fruit with fine tuned eval (there cannot be any search at such short time controls at all, can there?) it's quite clear, what to do:
The part of rybkacode known by the ICGA must me put under GPL, right?
If there still isn't any rybkacode known by the ICGA, what kind of a judgement would this be anyhow

[Tom Barrister]

So the secret of Rybka's success was hyperbullet testing/tuning.

Other than that, Rybka was simply Fruit, with possibly some Crafty (and whatever other engines) thrown in, then obfuscated in whatever way would throw everybody off as to the engine's origins or for whatever other purpose it was obfuscated..

That sums up the "brilliant programmer" Vasik Rajlich. He got lucky and stumbled onto something that worked.

[acozzie]

Cozzie is [...] suggesting he gave away the biggest secret out of ignorance.

You've said this twice; it may be a misunderstanding.

So I think Vasik Rajlich is simply a good programmer with the chess knowledge of an International Master and no moral issues with plagiarizing the work of others who was using a better tuning method than the rest of us. I'm guessing here, but I don't think he really understood that and instead considered other engine authors incompetent morons (...), which also explains why he was willing to tell me his procedure in Mexico.

I read his guess as: Vas didn't realize his most distinct advantage over his competitors (in terms of writing a strong program very quickly) was his willingness to plagiarize. Thinking instead that his competitors were just incompetent, he saw no risk in sharing his testing procedure. He obviously would have recognized its superiority and importance, since its results quantify themselves.

Did you read it differently? Note: I am paraphrasing, not evaluating.

Yes, I read it quite differently. Cozzie makes it clear he is not alluding to Rybka starting with a bit of Fruit here and there, but the whole thing ("even when you are starting with something decent" i.e. Fruit 2.1), and guesses that Vas considered the others morons, because even though he started with something decent (Fruit 2.1), at a par with other top programs of the day (Shredder, Fritz, and co.), he trounced them in 6 months, something the others were incapable of doing. Why couldn't they? Because they were incompetent. Except that Vas didn't realize that it was not so much his programming skill that achieved this (just optimizations and finetuning) but the hyperbullet testing that was the real secret of his success. Not realizing that, he gave away the actual secret to his success, mistakenly thinking it was his other engine programming skill that had reached it. But not so, the real secret sauce was his hyperbullet testing, that allowed for optimzations and finetuning. The 460 Elo he achieved was purely the result of optimizations and finetuning through the the hyperbullet testing.

"It was simply a highly optimized and well tuned program. (...) I realized the secret sauce: hyperbullet testing. (...) The massive improvement in engine strength from 2005 to 2010 is mostly due to them."

Yes, that is correct Albert, except that computer chess is in fact mostly finetuning and relatively few optimizations. This is also why I believe that computer chess is mostly a solved problem: most of what we do is exactly this fine tuning. Add a pattern here; tweak a value there. 'Oh, maybe I should only reduce 3 ply instead of 4 under case X'. You see 450 elo; I see 'engineering problem'. If you don't believe me, compare the Toga code to Fruit. It's quite a bit stronger in ELO, but with relatively few lines changed.

I admit that this perspective is rather extreme and comes from my background as PhD. For comparison, my current research involves writing programs from English. We have no idea how to do this well currently.

Think about what you say when you talk about his 'programming skill'. Remember a doubling of speed only achieves 60 elo or so. What chess engines really are is a collection of heuristics to estimate solutions to an exp-time problem. So either Vasik has some sort of special talent in selecting and evaluating such heuristics, or he has a better algorithm for selecting them. I find #2 much easier to believe, but that's just me.

Miguel: I agree with the search thing, but the space of search parameters is much smaller. You can afford to do ordinary blitz testing. But when you have 10,000 evaluation parameters you just don't have the CPU time. But hey, I've been wrong before.

P.S. I don't want to imply that Rybka's 5 year domination of computer chess was any less impressive because of this. In computer science you always want to find general solutions. The point of that section is that for most people 'strong engine -> great programmer -> no need/possibility of cloning'.

[Sylwy]

Your argument here isn't even scientific.

Hi Leon !

OK !
And how can you explain -in a scientific way- that all top commercials & top closed source free engines are in big progress after Rybka publication and successive decompilations ? Great jumps and a non -linear progress for all ? A statistical singularity ? Take a look to Lokasoft's Deep Sjeng (1.0-1.6) - Deep Sjeng WC2008- and c't 2010 version ! Or compare -please - Spike 1.2 Turin with Spike 1.4 Leiden ! The "eruption" of Critter from " nothing" to a high level chess engine. OK , agreed that SMK , Mark & Amir preserved the nice style of their engines. But the progress curve is too abrupt.
Take a look here:
============================================
37 .Shredder 9.0 UCI 256MB Athlon 1200 Mz 2811
38 .Shredder 8.0 CB 256MB Athlon 1200 MHz 2801
39 .Shredder 7.04 UCI 256MB Athlon 1200 MH 2795
============================================
Source:SSDF: http://ssdf.bosjo.net/long.txt
============================================

16 (sixteen) ELO points progress in 3 (three) successive publications. Lack of inspiration ? And now ? An unexpected Big-Bang of their programming abilities ?

The Fruit's spectrum is over all top engines.Not only Rybka. All top commercials and top closed source must be verified for a new WCCC.


And to be honests: the commercial computer chess is already discredited.
Facts ? OK , only some :

1.Rybka publication and the suspicion to be a clone;
2.Strelka & RobboLito affairs;
3.the publication of RobboLito source code like free;
4.the publication of Rybka 4 engine before the official release , affair being hushed up in high speed;
5.every new ChessBase GUI being patched by a single impolite "genius" , and ChessBase is silent;
6.a lot of more facts.

Now ,in the world of computer chess is a great chaos. Laws , control , correctness , honesty.......... ? No , it's a primitive world- now- in compuer chess ! Hoping this syndrome is only in computer chess. Hoping for !


Regards,
Silvian

[tomgdrums]

Cozzie is [...] suggesting he gave away the biggest secret out of ignorance.

You've said this twice; it may be a misunderstanding.

So I think Vasik Rajlich is simply a good programmer with the chess knowledge of an International Master and no moral issues with plagiarizing the work of others who was using a better tuning method than the rest of us. I'm guessing here, but I don't think he really understood that and instead considered other engine authors incompetent morons (...), which also explains why he was willing to tell me his procedure in Mexico.

I read his guess as: Vas didn't realize his most distinct advantage over his competitors (in terms of writing a strong program very quickly) was his willingness to plagiarize. Thinking instead that his competitors were just incompetent, he saw no risk in sharing his testing procedure. He obviously would have recognized its superiority and importance, since its results quantify themselves.

Did you read it differently? Note: I am paraphrasing, not evaluating.

Yes, I read it quite differently. Cozzie makes it clear he is not alluding to Rybka starting with a bit of Fruit here and there, but the whole thing ("even when you are starting with something decent" i.e. Fruit 2.1), and guesses that Vas considered the others morons, because even though he started with something decent (Fruit 2.1), at a par with other top programs of the day (Shredder, Fritz, and co.), he trounced them in 6 months, something the others were incapable of doing. Why couldn't they? Because they were incompetent. Except that Vas didn't realize that it was not so much his programming skill that achieved this (just optimizations and finetuning) but the hyperbullet testing that was the real secret of his success. Not realizing that, he gave away the actual secret to his success, mistakenly thinking it was his other engine programming skill that had reached it. But not so, the real secret sauce was his hyperbullet testing, that allowed for optimzations and finetuning. The 460 Elo he achieved was purely the result of optimizations and finetuning through the the hyperbullet testing.

"It was simply a highly optimized and well tuned program. (...) I realized the secret sauce: hyperbullet testing. (...) The massive improvement in engine strength from 2005 to 2010 is mostly due to them."

Yes, that is correct Albert, except that computer chess is in fact mostly finetuning and relatively few optimizations. This is also why I believe that computer chess is mostly a solved problem: most of what we do is exactly this fine tuning. Add a pattern here; tweak a value there. 'Oh, maybe I should only reduce 3 ply instead of 4 under case X'. You see 450 elo; I see 'engineering problem'. If you don't believe me, compare the Toga code to Fruit. It's quite a bit stronger in ELO, but with relatively few lines changed.

I admit that this perspective is rather extreme and comes from my background as PhD. For comparison, my current research involves writing programs from English. We have no idea how to do this well currently.

Think about what you say when you talk about his 'programming skill'. Remember a doubling of speed only achieves 60 elo or so. What chess engines really are is a collection of heuristics to estimate solutions to an exp-time problem. So either Vasik has some sort of special talent in selecting and evaluating such heuristics, or he has a better algorithm for selecting them. I find #2 much easier to believe, but that's just me.

Miguel: I agree with the search thing, but the space of search parameters is much smaller. You can afford to do ordinary blitz testing. But when you have 10,000 evaluation parameters you just don't have the CPU time. But hey, I've been wrong before.

P.S. I don't want to imply that Rybka's 5 year domination of computer chess was any less impressive because of this. In computer science you always want to find general solutions. The point of that section is that for most people 'strong engine -> great programmer -> no need/possibility of cloning'.

Is your new reply yet another example of computer chess "wasting your time"??

[jmartus]

tom was that necessary to say??

[tomgdrums]

tom was that necessary to say??

Sure! Why not? Was it necessary for Anthony to say it in his original post? He made some money off of his engine from the many people who care about chess and computer chess. People gladly supported his engine and endeavors. People who cared about the fruits of his labor.

And for him to take such a condescending swipe at those people, as if he is somehow above it all, is rather childish to me.

So yeah I think it was necessary for me to say.

[Tom Barrister]

tom was that necessary to say??

In case you meant me, I toned down what I wanted to say.

[MooImAFish]

LoL.......

[h1a8]

Your argument here isn't even scientific.

Hi Leon !

OK !
And how can you explain -in a scientific way- that all top commercials & top closed source free engines are in big progress after Rybka publication and successive decompilations ?



I agree you mostly in concept (I believe several engines benefited from those source, although not necessarily copied code though). But you mentioned Hiarcs and showed data that was inconsistent. I don't believe Hiarcs benefited well (if any) from those sources, otherwise there wouldn't be so many free engines over 100 elo stronger than it. Hiarcs is still 32bit (most programs have 64bit capability).

Now it's possible that Hiarcs did benefit somewhat from those sources. But there is no way of knowing since Hiarcs' elo jumped greatly even before those sources were released, thus proving Mark is competent enough to improve his engine without the help of those sources. Plus you still have the 32bit thing (and we all know that 64bit alone can increase the elo greatly).