Whether Rybka 4 Should Be Allowed To Play At the WCCC?

[Dann Corbit]

Part of this thread was split due to a inappropriate personal attack plus a like response. Sorry to any others affected, but this can happen when posting responses to personal attacks.

Good Call Graham

actually i am quite surprised by the voting
pretty close
i would have thought the second option would be winning going away

i guess if this vote were held on one of the commercial sites the second one would be far in the lead

if held in the OPEN forum the first option would be in the lead

but here..on the CCC...we have a split decision

showing once again that the place for FAIR AND BALANCED CC information is the

CCC Regards
Steve


Somewhat Startled Regards
Steve

The poll questions are very bad, Steve. As a consequence, it is not clear what people have voted for.

My feeling is that, had a proper poll been offered, those with negative feelings about Rybka 4 would have been in the majority.

My own feelings are that there is strong enough evidence for a civil court to rule that Rybka 1 is a clone. If it is, then some of Rajlich's statements are less than totally truthful. As those statements have been made since Rybka 3 was released, then Rajlich's comments and values may be untrustworthy and the onus should therefore be on him to provide evidence from an impartial expert that Rybka 3 and 4 are clean. Until he does that, his programs should be banned.

I do not think that there is a strong enough evidence to rule that rybka1 is a clone.

Some facts:
Rybka1 has clearly different evaluation then fruit.
Fruit has knowledge that rybka1 does not have about endgame(and Rybka1 does not know simple endgames)

The evaluation is not the same and even the people who claim that rybka1 is derived from fruit admit that the evaluation weight are different
in almost every factor.

The similiar parts can be considered as using ideas from fruit.

Uri

I would simplify the statement to this:

Proven fact: Rybka contains fruit ideas. There is simply no doubt about it.
Not a proven fact: Rybka is a modified version of fruit.

I think the fact that people do not like what Vas did makes them assume that he took the path they consider the worst. While it is possible that he did it is also possible that he did not. I also demonstrated in a previous post that somewhat different looking C algorithms {written independently} can boil down to *extremely* similar assembly (the chief differences being the choices of registers). I posted four independently written string reversal algorithms, collected from the newsgroup news:comp.lang.c

I think that the majority of programmers have no idea what this means:
Algorithms cannot be copyrighted (though they can be patented).
The only thing a programmer owns {via copyright protection} is his or her implementation of the algorithm.

[K I Hyams]

The similiar parts can be considered as using ideas from fruit.
Uri

Professor Hyatt states that at least one of the tables that are in early versions of Rybka are identical to that which appears in Fruit. If that is the case, it is not simply ideas that have been used. You appear to be saying either that Professor Hyatt is lying about identical tables in both engines or that he is not capable of seeing differences which are obvious to you. Which one is it?

Civil court judgment in the UK hinges on the criterion of "balance of probabilities". Clear, unequivocal evidence on the existence of those tables from Professor Hyatt, one of the most experienced experts in chess programming on the planet and also a man who looks for plagiarism as part of his day-job, in addition to the evidence of Zach Wegner would be sufficient to obtain a judgment on the "balance of probabilities" criterion.

[yanquis1972]

wow, these poll results are ludicrous. i guess it's once again anti-vas sentiment/people trying to make an oblique point re IPPO, but...sheesh. i'm guessing there was some kind of invasion but barring that these results are a bit embarrassing coming from a/the 'respected' (heh) computer chess forum...

[gerold]

wow, these poll results are ludicrous. i guess it's once again anti-vas sentiment/people trying to make an oblique point re IPPO, but...sheesh. i'm guessing there was some kind of invasion but barring that these results are a bit embarrassing coming from a/the 'respected' (heh) computer chess forum...

Was a little surprising for me also.
Guess you can't fool all the people all the time
The main reports on trying to decide if Rybka is a clone has not
really shown much for most people. The First by Z shows bits
and pieces but also leaves ifs and buts here and there.
The last report by BB who would not put a name to the
report does not show much.
Without the code of Rybka ..................
Guess work.

Best,
Gerold.

[Dann Corbit]

The similiar parts can be considered as using ideas from fruit.
Uri

Professor Hyatt states that at least one of the tables that are in early versions of Rybka are identical to that which appears in Fruit. If that is the case, it is not simply ideas that have been used. You appear to be saying either that Professor Hyatt is lying about identical tables in both engines or that he is not capable of seeing differences which are obvious to you. Which one is it?

Civil court judgment in the UK hinges on the criterion of "balance of probabilities". Clear, unequivocal evidence on the existence of those tables from Professor Hyatt, one of the most experienced experts in chess programming on the planet and also a man who looks for plagiarism as part of his day-job, in addition to the evidence of Zach Wegner would be sufficient to obtain a judgment on the "balance of probabilities" criterion.

If this is the criteria, then all chess engines are clones.
All chess engines use Alpha-Beta (at least 99% anyway).
Nearly all chess engines use piece square tables. Most of the tables are extremely similar.
Nearly all chess engines use SEE. The operations that take place in SEE are extremely similar for all engines.
Nearly all chess engines use hash tables. Nearly all hash tables are Zobrist based. Many chess engines use similar or identical Zobrist random tables.

In general, if you examine carefully chess programs you will find that literally all of them follow the same outline. EACH AND EVERY chess program is a sequnce of ideas BORROWED FROM OTHER PEOPLE.

An algorithm does not have copyright status. It must be patented for protection. An algorithm is:
A well defined sequence of steps for solving a problem.

That sequence of steps is NOT protected.

Is the chess programming field growing petty and bitter? I am not sure.

But I do think that the main problem is competition. Competition in chess programming has gotten out of hand so that people no longer want to share ideas and are so concerned about winning that anything they feel may disturb the competition not in their favor is wrong and bad.

The algorithms and ideas that THEY borrowed were simply fair use. But the algorithms and ideas that OTHER people borrow are criminal acts.

That is what competition has done. If there were not these fierce battles for who is best, we would not see these problems.

I could be wrong, of course. And I do not deny that it is possible that Vas has done something legally wrong -- it is possible. But the armies of vigilanties with pitchforks and torches have gone off half cocked in my opinion.

[K I Hyams]

The similiar parts can be considered as using ideas from fruit.
Uri

Professor Hyatt states that at least one of the tables that are in early versions of Rybka are identical to that which appears in Fruit. If that is the case, it is not simply ideas that have been used. You appear to be saying either that Professor Hyatt is lying about identical tables in both engines or that he is not capable of seeing differences which are obvious to you. Which one is it?

Civil court judgment in the UK hinges on the criterion of "balance of probabilities". Clear, unequivocal evidence on the existence of those tables from Professor Hyatt, one of the most experienced experts in chess programming on the planet and also a man who looks for plagiarism as part of his day-job, in addition to the evidence of Zach Wegner would be sufficient to obtain a judgment on the "balance of probabilities" criterion.

If this is the criteria, then all chess engines are clones.
All chess engines use Alpha-Beta (at least 99% anyway).
Nearly all chess engines use piece square tables. Most of the tables are extremely similar.
Nearly all chess engines use SEE. The operations that take place in SEE are extremely similar for all engines.
Nearly all chess engines use hash tables. Nearly all hash tables are Zobrist based. Many chess engines use similar or identical Zobrist random tables.

Well Dann, I think that I may rate Professor Hyatt slightly more highly than you rate him. I base that assumption on the fact that, unlike you, I am willing to give him the benefit of the doubt on certain common sense issues. Whereas I believe that he realises that most if not all chess programs use devices such as hash tables, zobrist keys and piece square tables and therefore bases his claims about identical tables on items that are more unique, you don't appear to do that.

Although you don't appear to do that, I am however also prepared to give you the benefit of the doubt; I think that you also realise that he would not deliberately make specious claims, that he would not claim that generic tables had undue significance and so I am at a loss to see the point of your post.

[Dann Corbit]

The similiar parts can be considered as using ideas from fruit.
Uri

Professor Hyatt states that at least one of the tables that are in early versions of Rybka are identical to that which appears in Fruit. If that is the case, it is not simply ideas that have been used. You appear to be saying either that Professor Hyatt is lying about identical tables in both engines or that he is not capable of seeing differences which are obvious to you. Which one is it?

Civil court judgment in the UK hinges on the criterion of "balance of probabilities". Clear, unequivocal evidence on the existence of those tables from Professor Hyatt, one of the most experienced experts in chess programming on the planet and also a man who looks for plagiarism as part of his day-job, in addition to the evidence of Zach Wegner would be sufficient to obtain a judgment on the "balance of probabilities" criterion.

If this is the criteria, then all chess engines are clones.
All chess engines use Alpha-Beta (at least 99% anyway).
Nearly all chess engines use piece square tables. Most of the tables are extremely similar.
Nearly all chess engines use SEE. The operations that take place in SEE are extremely similar for all engines.
Nearly all chess engines use hash tables. Nearly all hash tables are Zobrist based. Many chess engines use similar or identical Zobrist random tables.

Well Dann, I think that I may rate Professor Hyatt slightly more highly than you rate him. I base that assumption on the fact that, unlike you, I am willing to give him the benefit of the doubt on certain common sense issues.

Whereas I believe that he realises that most if not all chess programs use devices such as hash tables, zobrist keys and piece square tables and therefore bases his claims about identical tables on items that are more unique, you don't appear to do that.

Although you don't appear to do that, I am however also prepared to give you the benefit of the doubt; I think that you also realise that he would not be deliberately mendacious, that he would not claim that generic tables had undue significance and so I am at a loss to see the point of your post.

I think very highly of Dr. Hyatt and I also respect his opinion.
However, I disagree that we have proof that Vas has broken the law.

[Dann Corbit]

The similiar parts can be considered as using ideas from fruit.
Uri

Professor Hyatt states that at least one of the tables that are in early versions of Rybka are identical to that which appears in Fruit. If that is the case, it is not simply ideas that have been used. You appear to be saying either that Professor Hyatt is lying about identical tables in both engines or that he is not capable of seeing differences which are obvious to you. Which one is it?

Civil court judgment in the UK hinges on the criterion of "balance of probabilities". Clear, unequivocal evidence on the existence of those tables from Professor Hyatt, one of the most experienced experts in chess programming on the planet and also a man who looks for plagiarism as part of his day-job, in addition to the evidence of Zach Wegner would be sufficient to obtain a judgment on the "balance of probabilities" criterion.

If this is the criteria, then all chess engines are clones.
All chess engines use Alpha-Beta (at least 99% anyway).
Nearly all chess engines use piece square tables. Most of the tables are extremely similar.
Nearly all chess engines use SEE. The operations that take place in SEE are extremely similar for all engines.
Nearly all chess engines use hash tables. Nearly all hash tables are Zobrist based. Many chess engines use similar or identical Zobrist random tables.

Well Dann, I think that I may rate Professor Hyatt slightly more highly than you rate him. I base that assumption on the fact that, unlike you, I am willing to give him the benefit of the doubt on certain common sense issues.

Whereas I believe that he realises that most if not all chess programs use devices such as hash tables, zobrist keys and piece square tables and therefore bases his claims about identical tables on items that are more unique, you don't appear to do that.

Although you don't appear to do that, I am however also prepared to give you the benefit of the doubt; I think that you also realise that he would not be deliberately mendacious, that he would not claim that generic tables had undue significance and so I am at a loss to see the point of your post.

I think very highly of Dr. Hyatt and I also respect his opinion.
However, I disagree that we have proof that Vas has broken the law.

A caveat:
There is enough similarity in some algorithms of Rybka 1.0 that I think it would not be unreasonable for any governing body that chooses to consider Rybka 1.0 a derivative of Fruit and ban it from certain competitions. That would, of course, be up to their judgement.

I do make a clear distinction between borrowing of ideas (which has clearly and without a shadow of a doubt happened with Rybka 1.0 and Fruit) and proof of wrongdoing.

We do have definite proof that Rybka 1.0 uses Fruit ideas. Whether or not this makes Rybka 1.0 a Fruit derivative is another matter and up to sanctioning bodies to make a decision.

On the other hand, the evidence available is (to me, your opinion may vary) not at all conclusive that Vas has broken the law.

I do think that many chess programmers consider Vas' use of Fruit ideas in Rybka repulsive. That is neither here nor there.

[Sean Evans]

A caveat:
There is enough similarity in some algorithms of Rybka 1.0 that I think it would not be unreasonable for any governing body that chooses to consider Rybka 1.0 a derivative of Fruit and ban it from certain competitions. That would, of course, be up to their judgement.

Dan why would you be against the WCCC judges from performing a derivative test on Rybka 4? It is right in the rules that they are allowed! I mean a real derivative test, not just does R4 make the identical moves as Fruit.

Cordially,

Sean

[Harvey Williamson]

A caveat:
There is enough similarity in some algorithms of Rybka 1.0 that I think it would not be unreasonable for any governing body that chooses to consider Rybka 1.0 a derivative of Fruit and ban it from certain competitions. That would, of course, be up to their judgement.

Dan why would you be against the WCCC judges from performing a derivative test on Rybka 4? It is right in the rules that they are allowed! I mean a real derivative test, not just does R4 make the identical moves as Fruit.

Cordially,

Sean

If any other competitor was going to make an official complaint to the ICGA about Rybka it probably should have been done when Rybka 1st entered the WCCC. I doubt they would consider a complaint several years/versions later. Fabien has never complained and Fruit and Rybka have both competed in the same event.

I also doubt that any commercial author would submit their source if asked to.