I think Soren was the best for the job for a Chessbase article about it,
As with most
ex post facto commentators, Dr. Riis appears to assume that he has some great insight, in particular one that the Panel did not. And as with most such revisionists, I'd say he's simply wrong. I also question whether he has sufficient technical competence (his speciality is
theoretical computer science, and I don't know if he has ever really programmed).
He states that:
It is clear that Rybka is an original program by any reasonable standard. Based on the evidence I will present, a person can form a very credible alternative conclusion: that the implementation of similar evaluation concepts and algorithms in a computer chess program will inherently lead to code similarities even if no code is copied. But as of yet, I see little if any evidence for either of these claims. Perhaps Part 2 will eludicate?
He made two main points, segwayed into a long spiel about Rybka's impressiveness, and then switched into pedantic lawyer mode to ask
exactly what the rules were, who can possibly interpret them, etc. Notably, on both of his main points, he is simply wrong. His grand attempt to re-position the issue as
"Rybka implements concepts and algorithms learned from Fruit” appears to be the most correct and accurate formulation was already essentially considered (and rejected) in the ICGA process. Furthermore, as of yet, he has given no basis for this "correct and accurate formulation" (other than to give a general pointer to Mr. Schröder's website, and perhaps some unpublished notes of Mr. Schüle), whereas the contrary has voluminous material backing it. In particular, in my recent recapitulative
PDF, on page 5, I state (as a synopsis of the R1/Fruit evidence, which is outlined in a later 14 pages):
The Panel concluded from EVAL_COMP, in part when viewed alongside other evidence, that it was quite clear that more than just "ideas" from Fruit 2.1 re-appeared in Rybka 1.0 Beta, but rather quite specific creative choices. Any individual element could be declared to be simply "Fruit influence", but the picture as a whole stretched much beyond that. There was a general consensus that the Fruit/Rybka situation was much beyond the "standard" amount of engine sharing that was typically permitted in author-based computer chess.
His second main contention, that all computer chess programs are much the same, seems only to be backed by a quotation claimed to be from Rajlich (where "95% algorithmically" is claimed). [Riis exaggerates his position so much as to state that
one can start by acknowledging the truth contained within Rajlich’s remark on this topic, perhaps in the hope that no one will challenge it]. Again I can't imagine that most programmers would agree with it, at least in the manner used by Dr. Riis. For instance, Rajlich himself claimed that Rybka had a
very original search and evaluation framework, hardly in line with the idea that chess programs are mostly all the same. Furthermore, the ICGA process particularly considered this angle, and took pains to measure how much "expected" overlap one might expect between programs. Again the consensus was much against the claim of Riis (or Rajlich), no matter much he hectors us to "acknowledg[e] the truth" of it.
