garybelton wrote:This seems fair, yes, thanks for the explanation.
After reading the reports on your wiki it would seem that if the Rybka author had communicated with you then probably only the 2007 and 2008 titles would have been stripped, as from Rybka 3 onwards no offending codes were found?
Not quite. We did not look at Rybka 3. This is a time-consuming process. You start with a binary that was produced by a very good optimizing compiler, then you strip all symbols so that nobody can tell what function does what, what variable means what,, and then you have to convert that binary back to assembly language, and then study the code to see what it does. not easy. Then, once you uderstand the code, you start to work backward from the assembly language to C, to make the comparison possible. We found that the first rybka versions (including 1.6.1) were pure crafty ripoffs, then, starting with 1.0 beta, we see fruity search and evaluation converted to bitboards, but with the same scores, the same things evaluated, in the same order, in the same way (except translating mailbox to bitboard). And then 2.3.2a shows the same evaluation with a few minor additions.
At some point in an investigation like this, the burden of proof shifts to the accused. There is no need to do more of this time-consuming work to check rybka 3 or 4 (not even sure 4 has competed). There's already plenty of indications that Rybka is "bad". If Vas wants to contest anything with respect to Rybka 3, he can produce source code to prove originality. I don't see any reason to check every version he has done, since so many of them have been proven to be bad. It is not unreasonable to assume the rest are bad until he finally steps up and offers proof.
So is it possible he is being punished by losing all four and a lifetime ban for being uncooperative/uncommunicative?
Only David and the ICGA board can answer that. But it would certainly seem reasonable, that after all the prompting David did to try to get him to respond to either the panel's deliberations, or to the panel's report, or to the board after the panel's report was submitted, and he supplied _nothing_ in return. I'd think that in any sort of legal trial, the judge would not be too impressed on such a lack of cooperation. In the US, in federal courts, this is actually a component of the sentencing formula the sentencing recommendations contain. There are points added for various things like prior convictions, did the defendent show remorse and accept responsibility for his actions, and such. And the more points you get, the longer the sentence. Not cooperating, not accepting responsibility can absolutely land you with a stiffer sentence in US Federal court..
Also, do you have any insight into why all the hyperbole in the press? Why not have a balanced announcement saying "We took these actions because we found X and our rules do not permit it. We will say however that Rybka and ideas from it have helped all engines improve by significant amounts"? Why a lot of the use of the word "cheat"? From a chess engine fan perspective it looks over the top.
There is no "good side" to this story. Cheating is the right word. From Webster's:
Verb: Act dishonestly or unfairly in order to gain an advantage, esp. in a game or examination.
Pretty well described exactly what happened...