Programmers Open Letter to ICGA on Rybka/Fruit

[hyatt]

How about if I send you some of mine?



(hate mail, that is...)

[Olivier Deville]

How about if I send you some of mine?



(hate mail, that is...)

Wow Bob, are you getting hate mails at the moment ? It is rather unpleasant indeed...

Olivier

[hyatt]

How about if I send you some of mine?



(hate mail, that is...)

Wow Bob, are you getting hate mails at the moment ? It is rather unpleasant indeed...

Olivier

A few hate mails. Lots of hate posts. Comes with the territory.

[BB+]

It makes sense for him [Rajlich] not publicly comment or respond to ICGA inquiries especially if there are improprieties that could result in legal matters. That said, does any foresee a scenario where a private and public apology and acknowledgment of misconduct revolves the long debate?

As with many things, it depends...

*) Should the ICGA process proceed, the first "charge" to be considered seems to be plagiarism, which would likely need to be shown beyond a reasonable doubt (Levy and the ICGA board would make any decision here). If Rajlich were to be found guilty, I would guess a sufficient remedy for part of this might simply be to change the relevant victories to read "Rybka/Fruit" rather than "Rybka", or maybe just to append Letouzey's name to the authors (incidentally, Kaufman does not appear as an author on the ICGA website for later Rybka victories), though the ICGA might disagree and/or dole out additional penalties. Whether the ICGA tries to investigate past that (or whether Letouzey would ask them to do so), I can't say. [I might also add that my PDF is not a "statement of the prosecution" in any sense, but rather just one evidentiary document, which presumably could be used by either side].

*) Any civil action would be mainly about copyright infringement, and the standard would be "balance of probabilities" -- my own opinion is that the defense would have a difficult time here. I expand on this, largely as posters in other places have suggested a "court trial" as a way of "settling" the issue.

First, a definite prior history of Rajlich infringing copyright with Crafty exists [by definite I mean, for instance, that one should be able to take the relevant compiler and Crafty's EvaluateWinner() and produce the same ASM code as in Rybka 1.6.1]. There also seems to be various deceptions and hiding of this fact (for instance, the ELChinito author admitted guilt in a similar Crafty-copying case and suffered almost no punishment -- Rajlich seems aware of this issue and the reconciliation [he commented in the thread albeit on a side issue], but then a few months later when he could have opted to remove Rybka from Chess War, he chose instead to enter a new version Rybka 1.6.1).

Next, there exist fairly explicit examples of Fruit code copied into Rybka 1.0 Beta, for instance in the iterative deepening code and the UCI parsing (see Appendix A of the PDF in particular -- these are perhaps "minor" elements for some purposes, but nevertheless show a pattern of copying). This can be then combined with the fact that of the ~20 evaluation features in Fruit, all of them appear in Rybka 1.0 Beta (and vice-versa except for 3 new additions by my count: material imbalance, lazy eval, and tempo bonus), most times with exact commonality of side conditions [a back-of-the-envelope comparison with various engines found none with an overlap much more than half this]. For more on "standards" for such things (and one can apply the same rationale to the root search operations, the PST structuring, the hash structure similarity, etc.), see the opinion in the linked PDF in this post, in particular: As to the "copying" element, the plaintiff rarely can obtain direct evidence that the defendant copied the work. Rather, the plaintiff generally puts in circumstantial evidence of the defendant's copying. [The quotation in the post itself is also interesting, concerning the "substantial similarity" between Romeo and Juliet and West Side Story]. While something as mechanical as "Occam's razor" is not a legal standard, on the standard of "balance of probabilities", my own (perhaps biased) opinion is that things do not look well for the defense on the prima facie level.

If Rajlich were to try make an explanation of all this, I expect a decent lawyer could (rather easily IMO) undermine his credibility/believability by bringing up (say) the issue of node counting [whether or not this is "proper" is a different matter -- at this point I speak merely in a sense of winning/losing any trial], among other things (e.g., a programmer who previously worked in industry not having any backups or old source code snapshots -- at all?!). However, even if Rajlich were to be found guilty of some amount of copyright infringement, the penalty stage would likely be a massive headache at the very least, with both actual and punitive damages coming into play.

Finally, trials cause a lot of mental/emotional stress for both sides (not to mention court costs and lawyer's fees, though with the latter Letouzey might get some backing from the FSF -- and maybe Rajlich from ChessOK/ChessBase?), and there's no guarantee of anything -- even that the arguments would finally end.

[BB+]

Thanks Mark. Actually I have not seen the thread where my actions have been criticized, but I don't mind much anyway.

I think it was Alan Sassler who originally mentioned this, but there is now an explicit ex post facto example.

Very true if we consider some people illegally acquire an early non public Rybka version, decompile it and publish parts of it.

As explicated above, "illegal" is not likely to be true in the given case. "Publishing parts" of it can hardly be considered scrofulous when said parts are themselves copied from Crafty.

[hyatt]

Of course. Posting evidence of illegal activity should be considered illegal. I'm sure all that commit criminal acts believe that...

[kingliveson]

Thanks Mark. Actually I have not seen the thread where my actions have been criticized, but I don't mind much anyway.

I think it was Alan Sassler who originally mentioned this, but there is now an explicit ex post facto example.

Very true if we consider some people illegally acquire an early non public Rybka version, decompile it and publish parts of it.

As explicated above, "illegal" is not likely to be true in the given case. "Publishing parts" of it can hardly be considered scrofulous when said parts are themselves copied from Crafty.

I want to agree with Lukas, but given that the individual who provided the binary said there was reason to believe impropriety, and might just be right. So I would have to say it is fair -- legal? Am not a lawyer.

[Jeremy Bernstein]

Thanks Mark. Actually I have not seen the thread where my actions have been criticized, but I don't mind much anyway.

I think it was Alan Sassler who originally mentioned this, but there is now an explicit ex post facto example.

Very true if we consider some people illegally acquire an early non public Rybka version, decompile it and publish parts of it.

As explicated above, "illegal" is not likely to be true in the given case. "Publishing parts" of it can hardly be considered scrofulous when said parts are themselves copied from Crafty.

I want to agree with Lukas, but given that the individual who provided the binary said there was reason to believe impropriety, and might just be right. So I would have to say it is fair -- legal? Am not a lawyer.

It was in all cases legal. NDAs are contracts and an email exchange doesn't cut it, whatever the Rybka peanut gallery purports to know.

Jeremy

[Harvey Williamson]

Thanks Mark. Actually I have not seen the thread where my actions have been criticized, but I don't mind much anyway.

I think it was Alan Sassler who originally mentioned this, but there is now an explicit ex post facto example.

Very true if we consider some people illegally acquire an early non public Rybka version, decompile it and publish parts of it.

As explicated above, "illegal" is not likely to be true in the given case. "Publishing parts" of it can hardly be considered scrofulous when said parts are themselves copied from Crafty.

I want to agree with Lukas, but given that the individual who provided the binary said there was reason to believe impropriety, and might just be right. So I would have to say it is fair -- legal? Am not a lawyer.

It was in all cases legal. NDAs are contracts and an email exchange doesn't cut it, whatever the Rybka peanut gallery purports to know.

Jeremy

I think I have now agreed with Jeremy at least twice this week

[kingliveson]

Thanks Mark. Actually I have not seen the thread where my actions have been criticized, but I don't mind much anyway.

I think it was Alan Sassler who originally mentioned this, but there is now an explicit ex post facto example.

Very true if we consider some people illegally acquire an early non public Rybka version, decompile it and publish parts of it.

As explicated above, "illegal" is not likely to be true in the given case. "Publishing parts" of it can hardly be considered scrofulous when said parts are themselves copied from Crafty.

I want to agree with Lukas, but given that the individual who provided the binary said there was reason to believe impropriety, and might just be right. So I would have to say it is fair -- legal? Am not a lawyer.

It was in all cases legal. NDAs are contracts and an email exchange doesn't cut it, whatever the Rybka peanut gallery purports to know.

Jeremy

I think I have now agreed with Jeremy at least twice this week

Say am trying to put on some muscles and a friend hands me a bottle of what he says was creatine, but rather than gaining muscles am becoming more mentally unstable. So being suspicious that it might be really cocaine, I send the bottle to my good friends Charlie Sheen and Lindsay Lohan (they being experts on the matter) to help verify. Have I committed a crime? Silly analogy, I admit.

But seriously, I don’t know because it is always gray with software.