Is Belka a Rybka?

[matejst]

Ryan,

I don't want this issue to die. It's very interesting.

You misunderstood my point. In my view, it is not a question of cloning, but a question of ethics and standards in the world of computer chess. We have to admit that using not only the ideas, but also the solutions of others is already widely spread. And that it is done not only by an unknown Russian who, in a naïve way, admitted it, but probably also by the leading figure(s) in the field. It is not nice, but c’est dans l’ordre des choses. And most of the time, it is perfectly legal.

On the other side, without the heavy use of F. L. code by a variety of programmers – V. R. among them - we wouldn’t have so many strong engines. Now that part of Rybka’s secrets have been disclosed we can expect a new jump in strength on the engines’ scene. A new race has begun. It’s good for the users. Why should we complain?

So when everybody is yelling about Strelka being a clone, I don’t care. Nobody today speaks about Rybka being a clone of Fruit, although Osipov eventually gave the proofs for that (just go on the Kasparovchess forum). I want the same treatment for everybody. If we condemn Osipov, we shall condemn the behavior of many others. If we stick to what is legal, then let the man in peace.

Just one more word: I doubt very much in Rajlich sincerity (we know for certain that he refuses to review Strelka's code). He has too much money to lose in this story.

Kind regards.

BS

[Uri Blass]

Ryan,

I don't want this issue to die. It's very interesting.

You misunderstood my point. In my view, it is not a question of cloning, but a question of ethics and standards in the world of computer chess. We have to admit that using not only the ideas, but also the solutions of others is already widely spread. And that it is done not only by an unknown Russian who, in a naïve way, admitted it, but probably also by the leading figure(s) in the field. It is not nice, but c’est dans l’ordre des choses. And most of the time, it is perfectly legal.

On the other side, without the heavy use of F. L. code by a variety of programmers – V. R. among them - we wouldn’t have so many strong engines. Now that part of Rybka’s secrets have been disclosed we can expect a new jump in strength on the engines’ scene. A new race has begun. It’s good for the users. Why should we complain?

So when everybody is yelling about Strelka being a clone, I don’t care. Nobody today speaks about Rybka being a clone of Fruit, although Osipov eventually gave the proofs for that (just go on the Kasparovchess forum). I want the same treatment for everybody. If we condemn Osipov, we shall condemn the behavior of many others. If we stick to what is legal, then let the man in peace.

Just one more word: I doubt very much in Rajlich sincerity (we know for certain that he refuses to review Strelka's code). He has too much money to lose in this story.

Kind regards.

BS

Osipov did not prove that rybka is a clone of fruit.
Strelka is written in a clearly different way than fruit and
I do not believe that Osipov used copy and paste from fruit.

It is possible that osipov understood fruit's evaluation and used part of fruit's evaluation in a different structure but I do not consider it to be cloning.

Similiarity between fruit and rybka is clearly smaller than similiarity between rybka and strelka.

Note that I do not give an opinion if Strelka is a clone of rybka and only say that the situation of strelka and rybka is different than the situation of fruit and rybka.

Uri

[Edsel Apostol]

Hi Uri,

Note that strelka has piece square table in the following array
static __int16 PieceSquareValue[12][64][2]

It does not use
static const int BishopLine to generate piece square table.

Just my opinion, maybe PieceSquareValue[12][64][2] has been precomputed from BishopLine, and in Belka the author of WildCat added the capability to change this values by providing a function to calculate this piece square tables based on BishopLine. Maybe it is just like what you are doing now with Strelka, you are trying to replace the precomputed stuff in Strelka with an equivalent function. Please look at the source code of Sloppy, it has a precomputed piece square tables that are derived also from those exact values from the BishopLine.

About the clone issue:

I have observed about the analysis of Strelka 1.8, Fruit 2.1, Toga 1.3.1, and Rybka 1.0beta in a certain position where my program has blundered that Strelka, Fruit and Toga has produced an odd behavior (maybe a bug) where they have a similar PV and score output although in varying depths, and Rybka has a completely different PV and score.

About the motives of the Strelka author:
Based on my observation, the Strelka author might want the whole computer chess community to know that Rybka wouldn't be that strong without borrowing ideas from Fruit. Maybe he has found out how Rybka works, that it is similar to Fruit, and to prove his point he decided to create an engine that is very similar to both engines, Fruit and Rybka, and name it deliberately to sound like Rybka to create a scandal to get the attention of the whole computer chess community.

Maybe he is against the idea of using/(borrowing ideas) from other peoples code for commercial purposes, thats why he has done this to Rybka. He may thought that if he would be called a cloner, by creating an engine that is based on Fruit and behaves like Rybka, the author of Rybka will be called a cloner as well.

I'm not saying that the author of Rybka is a cloner, I'm just trying to explain my theory of what might be the motivation of the Strelka author. I may be right and I may be completely wrong as well.

In my own program, I used a lot of ideas from Fruit, and from Glaurung. No copy/paste, just ideas based on how I understand them.

What I have to say is that knowing that the most efficient design for a wheel is a circle, why would somebody still continue using one that is octagon.

[Rolf]

Ryan,

I don't want this issue to die. It's very interesting.

You misunderstood my point. In my view, it is not a question of cloning, but a question of ethics and standards in the world of computer chess. We have to admit that using not only the ideas, but also the solutions of others is already widely spread. And that it is done not only by an unknown Russian who, in a naïve way, admitted it, but probably also by the leading figure(s) in the field. It is not nice, but c’est dans l’ordre des choses. And most of the time, it is perfectly legal.

On the other side, without the heavy use of F. L. code by a variety of programmers – V. R. among them - we wouldn’t have so many strong engines. Now that part of Rybka’s secrets have been disclosed we can expect a new jump in strength on the engines’ scene. A new race has begun. It’s good for the users. Why should we complain?

So when everybody is yelling about Strelka being a clone, I don’t care. Nobody today speaks about Rybka being a clone of Fruit, although Osipov eventually gave the proofs for that (just go on the Kasparovchess forum). I want the same treatment for everybody. If we condemn Osipov, we shall condemn the behavior of many others. If we stick to what is legal, then let the man in peace.

Just one more word: I doubt very much in Rajlich sincerity (we know for certain that he refuses to review Strelka's code). He has too much money to lose in this story.

Kind regards.

BS

To add something basic: Rajlich is a real name who works under the Law. Everything from Russia is NOT real and legal or standing under the Law of almost whole Europe, so that "Osipov" could do many things. But to insinuate that then also Rajlich stood under the same or similar suspicion - that is evil and this can have a single direction of origin: those who economically are disadvantaged by the Rybka rating leader.

Therefore it is wrong what Boban insinuates. It is NOT reasonable to say, either we control or check Rajlich, or if we dont, then we should also let in peace Osipov. Because Osipov is profitting from the atrange copyright system in Russia. China would even be stranger. But Rajlich isnt profitting from strangeness of copyright laws in Russia but from his own talents.

I mean am I the only one who can look through this scheme? Rybka is better than all the known commercial programs. From all the logic I know of Ryan is correct with "Vas and the computer chess community are both better off without the continued stress of this subject." That then Boban from Serbia (?) likes to continue is understandable but only so far as Serbia does also not stand under the European copyright laws, like Russia too.

Chrilly Donninger comes from Austria and he therefore stopped his reverse so and so of Rybka rather quickly two years ago. Is it now honest if we would argue in Osipov's favor that it's interesting what he has found out with extended reverse so and so? Dont we recognize the pattern behind such attempts? Do we really need to have the threat from somewhere that someone would really like to publish the sourcecode of one "Strelka" who is based on the reverse so and so of Rybka potentially and possibly? Am I the only one who has realised that form of blackmail?

What has been the fate of the few interested who have sold FRUIT? And who couldnt develop further that remarkable engine, while Rajlich knew it?

I think this was important material for American readers who live under slightly different copyright laws than we in Europe. Perhaps the very indecent allegation of Boban is understood who wants to combine Vas with anything illegal or evil. Because Vas might refuse to analyse the source of Strelka (sic) he has "lost his sincerity" (Boban version of sincerity loss).

Since when someone honest can lose something by not contacting people who might be engaged in illegal stuff? This is postede under my real name and it stands under the European Law. What people who are anonymous are posting is irrelevant! And what people in Russia are doing is legally potentially illegal after the European Law.

[matejst]

Uri,

I found this allegation on kasparovchess forum. I'm trying to find it once more, but I need time...

You could try here http://kasparovchess.crestbook.com/view ... =1869&p=26.

I don't see why I should not believe this.

Kind regards!

BS

[matejst]

Serbia has long ago ratified all the international conventions about copyright. Just as Germany did, and Russia too.

You are just revealing what I am talking about from the beginning. You have prejudices if somebody is from Serbia, Russia, Romania or Scotland. I am asking for the same standards. The same standards for Osipov and Rajlich.

Osipov is not hiding, you know. You can find his email. A lot of persons know him. He is perfectly convinced that what he has done is legal. If Convekta sues him, they will know where to find him.

V. R. contacted Y. O. asking about reverse engineering (check on the kasparovchess forum).

About the sincerity loss: V. R. was never sincere. He very well knows that there is no place for sincerity in business.

Boban Stanojevic is my real name. I am a university teacher and a writer. You can find my name also as a chess player. I am not anonymous.

And I am not Boban for you: please use my surname.

Kind regards!

BS

[Andrej Sidorov]

correlation table

# Pair Ponder hit Moves counted
1 Rybka 1.0 64-bit – Strelka 1.8 71.4 1875
2 Rybka 2.3.2a 64-bit – Fritz 11 68.0 1374

What about Fritz 11? Do anybody wants to call it 'Fritzka' and remove from all tournaments and rating lists?

[matejst]

In my view, Osipov was telling the truth about how Strelka - and therefore Rybka 1.0 - was made. For me, it is legal (he rewrote the code) but not moral. If we agree that cloning is the use of original code, it is not a clone. If we think that the output is relevant, then, it is one. [But... the output (the evaluation, precisly) of Fritz 11 and Rybka 2.3.2 is very similar. Nobody is debating about this. (Sorry, Andrej Sidorov just did it.)]

Clone or not, his disassembling of Rybka 1.0 beta is against my conception of honesty.

It is always the scenario of an evil Russian and a good American. But here, we have something that is not black or white, but rather gray. Why has V. R. refused to inspect the code of Strelka? Why has Osipov offered that code if he wasn't certain he is innocent? There are many questions to think about.

This is a forum, a debate club. There is no reason not to speak about whatever that interests us. And, to complete my previous post: I am not insinuating. I am telling things openly.

Kind regards!

bs

[Uri Blass]

correlation table

# Pair Ponder hit Moves counted
1 Rybka 1.0 64-bit – Strelka 1.8 71.4 1875
2 Rybka 2.3.2a 64-bit – Fritz 11 68.0 1374

What about Fritz 11? Do anybody wants to call it 'Fritzka' and remove from all tournaments and rating lists?

You selection include only part of the engines

When I look at all engines it seems that we cannot take statistics about ponder hit seriously because micromax has the highest ponder hit and it is not logical.
It seems that there is some bug in the program that calculates ponder hit because I do not believe that micromax is so similiar to other engines.

http://computerchess.org.uk/ccrl/4040/c ... es+only%29

Uri

[Paul Bedrey]

correlation table

# Pair Ponder hit Moves counted
1 Rybka 1.0 64-bit – Strelka 1.8 71.4 1875
2 Rybka 2.3.2a 64-bit – Fritz 11 68.0 1374

What about Fritz 11? Do anybody wants to call it 'Fritzka' and remove from all tournaments and rating lists?

You are right of course. Ponder hit does not tell the whole story. I've been following this debate and find it perplexing. If anyone has a 2 core machine and Arena Chess all you have to do is load out both programs and analyze some positions simultaneously. The difference is most notable for positional testsuites. But even when they agree the whole lines usually differ. Of course I've been comparing Strelka 1.8 and I believe I heard that it was different from the initial Strelka. Perhaps someone who has the original could do the comparison and post it.