Previous World Champion Engine Authors Speak Out...

[Tord Romstad]

For instance, "The name of the rose" (book and movie) by Umberto Ecco is heavily inspired by the "Library of Babel" of Borges. Borges was blind, and a librarian at one point in his life. There is a blind librarian character in the story named "Burgos".

Ultan, the blind librarian in The Book of the New Sun, is obviously also supposed to remind the reader of Borges. Some readers take this further and believe that the city of Nessus must be a future Buenos Aires, but I personally don't see much evidence of this in the text.

[SuneF]

In programming, especially a chess program, adopting significantly large and critical blocks of very specific symbiotic logic and factored weights verbatim is only permissible in an acknowledged code fork. It certainly couldn't pass the smell test as an original work. And the distance between subsequent versions to the plagiarized starting point will hardly matter (to most people) if a programmer is shown to be so dishonest.

Ordinarily yes. What makes the Rybka situation special is that Vas couldn't have adopted "significantly large and critical blocks of very specific symbiotic logic and factored weights verbatim" as Rybka is bitboard and Fruit is mailbox. At least for the evaluation and move generation code it would have to be coded completely different - that is a fact. We also know that Rybka has new important material terms in the evaluation and the search seems original as well.
What they need to prove then, is either that some other parts of the engine contains large parts of copied code (for instance the hashing or protocol) or that large parts of the bitboard code is functionally equivalent to the mailbox code. The latter is a bit problematic however, as most engines are known to contain certain large parts of functionally equivalent codes, like for instance the SEE, Qsearch, Nullmove and so forth. No one would claim Fruit was a derivative of Crafty just because it had sections of nullmove code that was functionally equivalent to Crafty for instance.

What annoys me most about this whole issue is that we do not have objective means and rules to specify what is legal to copy and what is not. It's always a judgement call in every single case. This makes the whole process too ad hoc and un-scientific IMHO.
The process should be reversed, first we need to establish what is legal and what is not, then we need to decide how to punish the rule breakers and only then does it make sense to begin considering the individual engines. Otherwise we are having a trial and jury with no accepted laws to judge by.

[Dann Corbit]

In programming, especially a chess program, adopting significantly large and critical blocks of very specific symbiotic logic and factored weights verbatim is only permissible in an acknowledged code fork. It certainly couldn't pass the smell test as an original work. And the distance between subsequent versions to the plagiarized starting point will hardly matter (to most people) if a programmer is shown to be so dishonest.

Ordinarily yes. What makes the Rybka situation special is that Vas couldn't have adopted "significantly large and critical blocks of very specific symbiotic logic and factored weights verbatim" as Rybka is bitboard and Fruit is mailbox. At least for the evaluation and move generation code it would have to be coded completely different - that is a fact. We also know that Rybka has new important material terms in the evaluation and the search seems original as well.
What they need to prove then, is either that some other parts of the engine contains large parts of copied code (for instance the hashing or protocol) or that large parts of the bitboard code is functionally equivalent to the mailbox code. The latter is a bit problematic however, as most engines are known to contain certain large parts of functionally equivalent codes, like for instance the SEE, Qsearch, Nullmove and so forth. No one would claim Fruit was a derivative of Crafty just because it had sections of nullmove code that was functionally equivalent to Crafty for instance.

What annoys me most about this whole issue is that we do not have objective means and rules to specify what is legal to copy and what is not. It's always a judgement call in every single case. This makes the whole process too ad hoc and un-scientific IMHO.
The process should be reversed, first we need to establish what is legal and what is not, then we need to decide how to punish the rule breakers and only then does it make sense to begin considering the individual engines. Otherwise we are having a trial and jury with no accepted laws to judge by.

Something that I think is overlooked is that algorithms are protected only by patent.

The EXACT sequence of steps is simply not protected (that is, after all, the algorithm). If the implementation is not identical, then it is not protected (unless the process includes copy and modification which I would not know how to determine).

The whole thing turns my stomache. Besides legal issues, there is the more fundamental issue of right and wrong, fair verses unfair.

I am not certain that legal wrongdoing has occurred. It does seem quite clear that distasteful things have occurred. But I am a dinosaur and long for the old days when people would publish their algorithms in the ACM and then *hope* that other people would use them to their benefit.

All I can say is, "Too bad."

[mhull]

Ordinarily yes. What makes the Rybka situation special is that Vas couldn't have adopted "significantly large and critical blocks of very specific symbiotic logic and factored weights verbatim" as Rybka is bitboard and Fruit is mailbox....What they need to prove then, is either that some other parts of the engine contains large parts of copied code (for instance the hashing or protocol) or that large parts of the bitboard code is functionally equivalent to the mailbox code.

The issue that cuts to the heart of the case is the fact that correct tuning is a needle in a haystack of possibilities. The only way to Vas could improve upon Fruit right away would be to functionally effect the exact formulas in exact juxtaposition of tuned terms as fruit had done, and only after that add his improvements. And in fact, these exact critical tuned factors are seen in his code, even though he converted to bit-board. The evidence seems to show that these functional Fruit equivalents were dropped into the shell of his existing program in bulk, and were then built upon to surpass the original.

Others didn't do this, but simply took the principles observed in Fruit and applied them in the vernacular of their own designs rather than exactly mimic Fruit functionality. The only other person who seems to have tried (and eventually abandoned) something like Vas' method was Uri Blass, though he described his process publicly.

[Dann Corbit]

Ordinarily yes. What makes the Rybka situation special is that Vas couldn't have adopted "significantly large and critical blocks of very specific symbiotic logic and factored weights verbatim" as Rybka is bitboard and Fruit is mailbox....What they need to prove then, is either that some other parts of the engine contains large parts of copied code (for instance the hashing or protocol) or that large parts of the bitboard code is functionally equivalent to the mailbox code.

The issue that cuts to the heart of the case is the fact that correct tuning is a needle in a haystack of possibilities. The only way to Vas could improve upon Fruit right away would be to functionally effect the exact formulas in exact juxtaposition of tuned terms as fruit had done, and only after that add his improvements. And in fact, these exact critical tuned factors are seen in his code, even though he converted to bit-board. The evidence seems to show that these functional Fruit equivalents were dropped into the shell of his existing program in bulk, and were then built upon to surpass the original.

Others didn't do this, but simply took the principles observed in Fruit and applied them in the vernacular of their own designs rather than exactly mimic Fruit functionality. The only other person who seems to have tried (and eventually abandoned) something like Vas' method was Uri Blass, though he described his process publicly.

Uri's process was something like this:
Study exactly what Fruit was doing, then try the idea in his program.
He wanted to understand why the ideas worked and how they worked and so he dissected the ideas into their fundamental components.

Sometimes the fruit ideas did not work in his program the same as they did in fruit.

[playjunior]

I agree that Hyatt being in the committee seems wrong.

Hyatt to me is one of the main accusers, he has systematically gathered, structured and argued for the evidence that Rybka is a Fruit clone.

People who have such direct involvement (on any side) cannot be considered "impartial judges".

The best would be if they have some scientists/authors from related fields, like go/checkers/whatever who can fully comprehend the evidence presented but do not have any previous involvement in the issue, whatsoever.

This is what has been wrong with this process from the beginning. You are assuming facts not in evidence. I am not a judge. I will not decide innocence or guilt. I will not hand down a sentence. I am simply supposed to keep the investigation on the topic of "Did Vas copy parts of Fruit (or other programs) verbatim?" We are starting with Rybka 1 beta, but that is not the only version that will be addressed. We may go back to earlier versions, or to later versions (thru Rybka 4) since all have competed.

The panel that will do this is quite large and consists of a large number of programmers. The three "secretariat" members are simply there to try to keep the discussion limited to the specific topic of the investigation, and prevent all the other noise that always shows up here on CCC from obfuscating the technical merits of the arguments being made.

But nobody is paying attention to the document David sent out describing our function. They are making up their own incorrect assumptions and then complaining that those are not fair. However, they are also not real.

In U.S. where you are from to the best of my knowledge, the judge does not decide innocence or guilt, the jury does. The judge is largely responsible for the procedure, as you are in this case.

If you were to stand a trial, would you agree to have as the person responsible for moderating a fair and impartial process someone who has actively gathered, structured and argued on the prosecution side?

Surely there are many specialists who haven't had such a direct involvement in accusing Rajlich, and who could be a moderator? Many wouldn't have your credentials in the field but is that so crucial for moderation? So the question here is: why not someone else as a judge? After years of finger-pointing there at last is a hope for a credible process; by sitting as a judge you are giving the other party the opportunity to call this a witch hunt and dismiss the whole process, on the objective basis that one of the main accusers is sitting as a judge!

[Uri Blass]

No matter how much it morphed from version 1, it still stands upon the shoulders of Fruit 2.1. Without the Fruit fundamentals it would not be where it is.

And that is probably true of most strong engines now. I think few have not borrowed some algorithm or trick from Fruit. But only one is being accused of cloning.

--Jon

There's a difference between borrowing an idea (a "trick" as you say), and using the verbatim engine as a starting point, as it's been alleged that Mr. Rajlich has done. The question for engines as a whole is: where is the line drawn, e.g. what constitutes a violation?

The claim is not that Rajlich used the full fruit engine as a starting point but that part of the code of Rybka is copied from fruit.

I do not express opinion about that claim but it is impossible to prove that Rybka used fruit as a starting point when it is clear that most of the code is clearly different(move generator for example is clearly different than fruit and many other parts of the code are also different because they are dependent on the structure of the move generator).

I express no opinion if significant part is the same but even in this case this significant part is not most of the code.

[geots]

And Sign The OPEN letter to the ICGA President

http://hiarcs.net/forums/viewtopic.php?t=4038

excluding the Engine Author under examination .. we have all but one World Champion engine author since 1992 signing the letter
it should be remembered that these men were fierce competitors for many years ..so seeing them all agree to the notion that Rybka is an unauthorized Fruit derivative is quite astonishing and remarkable

Of course many other highly respected and talented Engine authors also signed the letter..some of whom might have been World Champions themselves if not losing to the engine in question or deciding not to even compete entirely in the ICGA World Championships given the participation of the allegedly illegally derived engine

Bob Hyatt ..a World Champion author himself of course did not sign as he is on the panel invested with making a determination on the derivative issue

i think this is unprecedented in the history of competitive sport Regards
Steve

Bob Hyatt ..a World Champion author himself of course did not sign as he is on the panel invested with making a determination on the derivative issue


I cant believe you state this in passing, because I cant believe you dont see anything wrong with him being on this panel. He has stated he knows Vas is guilty, and even if he is found innocent he is guilty. And you dont call this lynch mob mentality. How would you like to be the one Bob is making a decision about? This whole deal is about as slimy as computer chess can get.

I have clearly said that based on what I have seen with my own eyes, there is no doubt copying has occurred. It is _possible_ that Vas might somehow explain the identical parts of the code satisfactorily. I can't imagine what that might be, but it is possible.

My responsibility here is going to be to let the "prosecution" show the existing evidence everyone has seen, plus some other very significant evidence hardly anyone has seen, and then let Vas respond, if he chooses to. He can take the evidence point by point and explain why he believes it does not represent a problem of plagiarism or GPL violation. Then the "prosecution" will get to respond to his comments. And we will repeat this until nothing new is being shown. At that point, we will write a report giving the evidence and counter-evidence and send this to the ICGA. What they might do with it is unknown. It could involve stripping titles won with improper code, it could involve a "bad boy, don't do it again", or it could result in nothing at all.

We are not the jury. The three of us are a 3-judge panel to make sure that the discussions stay on topic, no flame wars, no personal conflicts. Just show the evidence and address it as needed until nothing more can be added...

There's no lynching. But I will bet you that once you see _everything_ you will change your opinion about what has happened. At least if you are willing to look at the evidence for yourself. There is a lot to be told here. Including a few shocks along the way...

This whole mess is packed with programmers who have a vested interest in getting Rybka off the scene. As I have said more than once- the most all this can do is keep Vas and Rybka out of tournaments. And that is what they are after. It wont stop him from selling the program. A court of "real law", and not second-place whiners, will have to do that. And who will be paying the legal costs then. I have a mental picture of you writing the biggest check.

All I see is programmers whose programs I have tested are either merely crap, or at best second-rate, trying to get together and basically saying "we cant beat him, so lets get rid of him." And their names dont impress me. Before Rybka was on the scene, I payed one commercial programmer around 50 bucks for his next version that he promised was way stronger than the previous one. And it turned out to be 8 elo stronger. I guess you think Im stupid enough to believe that came as a shock to him. And I see him right in the middle of all this slimy mess. You have said that even if found innocent, he is still guilty in your eyes. I will go you one better. If he is found guilty- he is still innocent in my eyes. Only the victim of being railroaded by a bunch of second place finishers. And dont even try to sell me the idea that all these programmers wont affect the outcome. I aint buying.

[Laskos]

All I see is programmers whose programs I have tested are either merely crap, or at best second-rate, trying to get together and basically saying "we cant beat him, so lets get rid of him." And their names dont impress me. Before Rybka was on the scene, I payed one commercial programmer around 50 bucks for his next version that he promised was way stronger than the previous one. And it turned out to be 8 elo stronger. I guess you think Im stupid enough to believe that came as a shock to him. And I see him right in the middle of all this slimy mess. You have said that even if found innocent, he is still guilty in your eyes. I will go you one better. If he is found guilty- he is still innocent in my eyes. Only the victim of being railroaded by a bunch of second place finishers. And dont even try to sell me the idea that all these programmers wont affect the outcome. I aint buying.

An opinionated basement tester. Give me the LOS of +53 =96 -32 result, I guess you have it in your tables. This seems to become a forum of opinionated, impertinent ignorants.

Kai

[Tom Barrister]

This whole mess is packed with programmers who have a vested interest in getting Rybka off the scene. As I have said more than once- the most all this can do is keep Vas and Rybka out of tournaments. And that is what they are after. It wont stop him from selling the program. A court of "real law", and not second-place whiners, will have to do that. And who will be paying the legal costs then. I have a mental picture of you writing the biggest check.

This program is brought to you by Rybka: "The engine made from scratch.™"

"Second place whiners" lol.

It isn't a matter of Rybka being the top engine. The honest chess programmers have a problem with those who aren't. That's part of the reason that the Ippolit/Robbolito series got banned, although Mr. Rajlich's influence probably had something to do with it. Now Rybka may be soon to follow. Or it may not be.

Rybka isn't even in first place anyway, and it might not be in the top five by the end of the year if Mr. Rajlich doesn't get off his butt and make some improvements to it.