TEST position TCEC5- Houdini 1.03a-DRybka4 1-0

[Tom Barrister]

So this thread has, like many others I've read here and there, transformed into a "Houdini/etc. is a clone of Rybka is a clone of Fruit" pole-waving match.

(NOTE: Those significantly familiar with the whole clone of Rybka which is a clone of Fruit controversies and related arguments might wish to save a few minutes of their lives by not bothering to read the rest, as it's mainly a rehash of what's been said already)

During the 1990's, the Othello/Reversi programmers didn't have this problem, since Othello has never had a wide following outside of Japan. Except for Louis Geoffrey (Brutus), nobody that I know of sold their Othello program. For those who don't know, Fabien Letouzey had one of the top 3 or 4 Othello programs then: Turtle. After the game was more or less "solved" as drawish, with slight winning chances for the second player (White) due to parity, a 16-piece random variant was introduced. When this ran its course, a lot of interest was lost, and most of the programmers moved on to other ventures. Several of the programmers, Mr. Letouzey included, released their Othello programs for the general use.

Mr. Letouzey, a relatively strong chess player, eventually turned his attention to chess programming. As most of you know, he is/was the primary author of Fruit. (If ever a program's name was a reflection of its author, Fruit certainly qualifies, but that's my opinion, and sorry for the digression.) Having taken that to the heights of the computer chess programming world, he probably grew tired of it (and of those who cloned and/or stole from it for their own programs---I'm not pointing a finger at anybody in particular) and moved on. Mr. Letouzey undoubtedly had/has a day job, and the bit of income that he made from Fruit probably didn't change his life significantly, nor would it ever do so.

Enter Vas Rajlich, who apparently decided to make Rybka a major commercial product. Whether or not he took much of Fruit's open-source (or decompiled commercial) code for Rybka is, frankly, a moot point, since Mr. Letouzey apparently never made a big issue of it. While (my name withstanding) I'm not a lawyer, I would imagine that it would be too late for a Letouzey vs. Rajlich lawsuit, were it provable that the original Rybka was a direct clone. Therefore, while the source to Rybka may or may not be Mr. Rajlich's legal property, he certainly has "earned" the right to use it, either by it being original or through lack of pursuit on Mr. Letouzey's part. Mr. Letouzey, if he cared at all, undoubtedly felt that such litigation would be (pardon the pun) fruitless.

Now several chess engines have sprung up (and most of you know all the names and don't need them repeated here) and have been accused of deriving much of their code from Rybka. Mr. Rajlich has accused several of them of doing this, has banned their discussion on his forum (which is his privilege), and has managed to get them banned from several computer tournaments.

It has been suggested that the easiest way to solve the problem would be to release the source code of one of the early versions of Rybka. That would help answer two questions: 1) Was Rybka a derivative of Fruit? and 2) Are the currently-accused engines derivatives of Rybka? So far, Mr. Rajlich has refused to do this, citing (I suppose) business concerns associated with releasing the source of his program.

Most of the above is old news, but it brings me to my own conclusions. It's apparent to me that if Mr. Rajlich were to bring a lawsuit against the clones, he would need to produce Rybka's source to make it stick; I wouldn't imagine that similar evaluations would hold up in a court of law, while line-for-line duplications of Rybka source code, said source being sufficiently original and not a duplicate of previous open-source code, would probably be interpreted as an infringement. It's possible that Mr. Rajlich believes that such litigation would be useless (and I agree). It's also possible that releasing the source of an early Rybka would prove that it is, indeed, largely based on Fruit. Such a revelation would, of course, not be good for Mr. Rajlich. I don't personally buy into the argument that releasing the source would hurt his business, as I believe that's already happened. Anybody with half a clue can use Google/Bing/Yahoo/etc. and find several engines and GUIs that are, if not as good as Rybka, good enough for most purposes.

If, as a hypothesis, releasing the source of an early Rybka would not hurt Rybka's business, and if such source were not a derivative of Fruit, then why hasn't Mr. Rajlich released it? Surely this would go a long way to discrediting the alleged clones, while also clearing the air that Rybka is mostly-original work and not a clone of Fruit.

[Laskos]

Release the source code ... Is this really what you are suggesting him to do ?

Yes, I think I was clear.

Then ask Vas to release his best source. More thant that, you are OUGHT to ask him, as he is making money out of this. Your behaviour is a little weird, you are asking a lot from a person not asking anything for the best engine out there, but you do not ask something from people literally profiting from every available info.

Now, how about Rybka 1.0 beta VAST knowledge, Vas "theories" about what "knowledge" is, the lack of only endgame knowledge (LOL), where its lack was evident, and the final conclusion that his engine is an obfuscated fast Fruitish searcher. He plainly lied, reducing depth and nps, increasing the size of the binary (a lot of freaking knowledge probably), plain and dirty lie.

Ask Vas for his best source, not Robert.

Kai

[notyetagm]

...


Most of the above is old news, but it brings me to my own conclusions. It's apparent to me that if Mr. Rajlich were to bring a lawsuit against the clones, he would need to produce Rybka's source to make it stick; I wouldn't imagine that similar evaluations would hold up in a court of law, while line-for-line duplications of Rybka source code, said source being sufficiently original and not a duplicate of previous open-source code, would probably be interpreted as an infringement. It's possible that Mr. Rajlich believes that such litigation would be useless (and I agree). It's also possible that releasing the source of an early Rybka would prove that it is, indeed, largely based on Fruit. Such a revelation would, of course, not be good for Mr. Rajlich. I don't personally buy into the argument that releasing the source would hurt his business, as I believe that's already happened. Anybody with half a clue can use Google/Bing/Yahoo/etc. and find several engines and GUIs that are, if not as good as Rybka, good enough for most purposes.

If, as a hypothesis, releasing the source of an early Rybka would not hurt Rybka's business, and if such source were not a derivative of Fruit, then why hasn't Mr. Rajlich released it? Surely this would go a long way to discrediting the alleged clones, while also clearing the air that Rybka is mostly-original work and not a clone of Fruit.

Nice to see someone make some sense on this topic.

[Gino Figlio]

+1

[Roger Brown]

So this thread has, like many others I've read here and there, transformed into a "Houdini/etc. is a clone of Rybka is a clone of Fruit" pole-waving match.

[SNIP]

If, as a hypothesis, releasing the source of an early Rybka would not hurt Rybka's business, and if such source were not a derivative of Fruit, then why hasn't Mr. Rajlich released it? Surely this would go a long way to discrediting the alleged clones, while also clearing the air that Rybka is mostly-original work and not a clone of Fruit.

Hello Tom Barrister,

This is really beautiful. It almost brought tears to my eyes.

Thanks for a really good post.

Later.

[zullil]

Release the source code ... Is this really what you are suggesting him to do ?

Yes, I think I was clear.

Your behaviour is a little weird, you are asking a lot from a person not asking anything for the best engine out there, but you do not ask something from people literally profiting from every available info.

Behaving weirdly doesn't bother me. And I have no interest in Rybka, or its source code. I sense that Mr. Houdart is worthy of redemption, and in my small and perhaps annoying way I am encouraging him to seek it.

[Graham Banks]

Arasan should be well outclassed, but at least it's original.

What is original? How do you know that? Are you a programmer?
Who is to say which techniques, ideas, lines of code, belong to particular programmers? Who decides who owns particular bits of intellectual property and for how long? Who decides what is in public domain and what isn't?

I would trust Jon Dart over the Ippo family scavengers any day.

[Albert Silver]

It's apparent to me that if Mr. Rajlich were to bring a lawsuit against the clones, he would need to

...identify who he was suing.

[notyetagm]

It's apparent to me that if Mr. Rajlich were to bring a lawsuit against the clones, he would need to

...identify who he was suing.

Wow, is this controversy *ever* going to end?

Is it not clear that Rybka 1 was heavily dependent on the Fruit source code? And that Houdini is partially based on the Ippolit family, as the author clearly states on his website?

[Albert Silver]

It's apparent to me that if Mr. Rajlich were to bring a lawsuit against the clones, he would need to

...identify who he was suing.

Wow, is this controversy *ever* going to end?

Is it not clear that Rybka 1 was heavily dependent on the Fruit source code? And that Houdini is partially based on the Ippolit family, as the author clearly states on his website?

What does that have to do with identifying the authors of Ippolit?