On June 28th 2011 IGCA president David Levy shattered the (computer) chess world with the announcement that programmer Vasik Rajlich, world champion computer chess in 2007, 2008, 2009 and 2010 with Rybka has been found guilty to have plagiarized the work of another chess program called: Fruit.
As a result Vasik Rajlich was banned for life, his 4 world champion titles were taken away, an ICGA press release followed that was picked up by the mainstream media containing all kind of juicy extensions and half truths.
This page is meant to put things back into perspective and its main focus is to refute the technical evidence that was used againt Rajlich, see page Contra Investigation on programmer level in the left menu.
Relevant less technical issues in the "External links" section among it a refutation (contributed to Chessvibes Part one, Part two) on an ICGA document of 2013 rehashing its (old) arguments of 2011.
The technical case in layman terms
The ICGA has made a case against Rybka because they found (too) many similarities between Rybka and Fruit hence the verdict was plagiarism. Our own investigation (which is the work of 1˝ year by various chess programmers) points out that after inspection many of those alleged similarities weren't similarities at all, but differences. This is demonstrated throughout the whole document in detail and summarized in Chapter V.
Statements of dissenting (critical) chess programmers
Chris Whittington games programmer and entrepreneur:
There may be some anecdotal evidence for your model, but, as far as I can tell there is no good proof, just a groupthink belief. Basically I think your argument is: Vas started with Fruit because he is a thief. Vas is a thief because he started with Fruit. Or, in other words, your argument is circular: the conclusion is the premise and the premise is the conclusion. More...
Marcel van Kervinck a Dutch software engineer and author of the chess program Rookie:
Second, I would also like to inform you that with the knowledge I have today, I would have voted differently in the investigation process.
This is a manufactured justification. A lot more people didn't follow the forum rules and that was just let go. Chris was excluded because he would undermine the purpose of the panel: to provide a case against Rybka. One member announced preliminary findings in a public forum. One member leaked discussions on a public forum. Half of the panel was discussing things behind the backs of the rest, despite the rules forbidding this. In retrospect, and from my point of view, that panel was just setup to lend credibility to the desired outcome, and nothing else. More...
Ed Schröder producer of the REBEL series and 2-times ICGA world champion:
After 15 months of intensive research it's my final conclusion the accusers who investigated the Rybka chess program researched an original program. Strength is one aspect of originality and the way Rybka ruled (on equal hardware) was unprecedented in computer chess history. The ICGA verdict and ban are uncalled for and the event is a regretable moment in the history of computer chess. More...
Sven Schüle a German computer scientist and author of the chess program Surprise:
This kind of statement has now been repeated by Bob approximately some thousands of times. So frequently that most people seem to have accepted this as a matter of fact, even many programmers who had doubts about it before. I also believe that Zach has done a great work by no doubt. I just do not share his final result, which is most important here. More...
Miguel A. Ballicora an Argentinean Biochemist and Associate Professor and author of Gaviota:
Your question is a very good question. Taking ideas from one source? many? is it the same? I do not know, and probably there is no simple answer for ICGA purposes, but the question is relevant. The problem is, rule #2 is terribly worded and was designed eons ago with only cut&paste clones in mind. The part that says "e.g. programs that play nearly the same moves" (or something like that) is extremely naive. Let's suppose that I write a program following Ed's document about how rebels plays. That would probably be a program that is 80-90% identical. There was zero code copy because I did not even see it. Is that ok? I believe yes... what would be the difference if I saw the code? More...
Dann Corbit an American programmer and computer chess expert, tester and advisor:
My main complaint is with the process of fault-finding itself. I think that the design of the process was flawed and with a flawed process flawed decisions are far more probable than via correct process. I do not expect the majority of chess programmers to agree with me or with my analysis. My opinion is mine alone but I wanted to say something since it appears to me that a bad process has been used to reach a final decision that affects a man's career in a significant way.
I don't think that we can call non-literal copying provable because it was never defined. Without a definition for something, how can you prove that it happened? Clearly, Vas did use many features of the evaluation in fruit. However, as Ed, Miguel, Sven and others have shown, Ryba's eval is not a copy of the eval in Fruit. The thing that disturbs me most about this is that all programs do this sort of thing. As Anthony Cozzie aptly pointed out, the program that wins is the program that collects the biggest bag of tricks invented by other people. Since everyone does it the rule cannot mean what it actually says and still be sensible. The intention was (obviously) to prevent things like Voyager and Bionic from competing in the contests but that is not what happened here. More…
Uri Blass an Israeli mathematician and author of the chess program Movei:
Rybka is not designed to be a copy of fruit and you cannot show me a big similiarity in the analysis between rybka and fruit that is a strange similiarity between different engines. More...
José C. Martínez Galán a Spanish computer chess programmer and author of Averno:
People is not condemned for doing unsual things. To condemn someone, to be able to say you have proven guilt, you must refute even the most unusual possibilities. Otherwise, the "in dubio pro reo" must be applied. More...
Charles Roberson an American computer scientist and chess programmer author of the chess engines NoonianChess, Telepath and Ares and as ICGA panelist: - Some things I stated seemed ignored - common inspiration being the main one. Some weren't, but they took effort. For instance, I managed to get the group to drop consideration of Rybka 1. I did this by supplying release dates for Rybka 2. That combined with the increased rating of Rybka 2 suggests that Rybka 1 didn't enter any ICGA events. That alone ends all of Zach's conclusions about Rybka 1.0 beta. And during the official voting: I think the key to the Fruit eval issue is whether or not they both could have been inspired by other previous open source code and open technical discussions. On this issue, I've found some of Zach's paper to be incorrect as to the originality of Fruit's eval.
Sergei Markoff a Russian chess programmer and author of SmarThink:
As the Rybka source was never published, the only possible thing to do with disassembly is to say that it contains a code that compiles at the same instuction sequences. But it's not a prove of direct copying of the code. More
Now 2 years after the ICGA verdict
where do we stand ?
1. An untruth first.
From the ICGA press release
During the course of the investigation and upon presentation of the Secretariat's report Vasik Rajlich did not offer, despite repeated invitations from the ICGA to do so, any kind of defence to the allegations, or to the evidence, or to the Secretariat's report
This is a discrepancy as to facts as the ICGA fails to mention that Vasik Rajlich was denied proper preparation time to defend himself against the charges. From the correspondence
From: Vasik Rajlich
Subject: Re: Open letter to the ICGA about the Rybka-Fruit issue
To: "David Levy"
Cc: "Larry Kaufman"
Date: Wednesday, March 2, 2011, 7:21 AM
I prefer to plead my case with the final arbiter rather than with the accusers. (1a)
Re. the schedule - could we have the final hearings in October or November? I'd like to prepare properly, and my schedule is very full until the first week in October. (1b)
1a. Rajlich signed for the rules of the ICGA (in the 2006-2010 period) to show the source code in a trusted environment in case of doubt. He did not sign for the new 2011 rule of a public trial in order (using the words of David Levy) "who will be named and shamed on the Internet". Reminding the ICGA to their own rules apparently did not impress David Levy (in fact the ICGA never asked Rajlich for his source code) and the investigation started without Rajlich.
1b. The complexity of this case is demonstrated by the two Rybka investigators who needed 1-1˝ years to compile the charges against Rybka and the contra investigation that took about the same time. Given the complexity and importance of getting things right, it was very reasonable to grant Rajlich a couple of months of his spare time to prepare and defend himself to the charges. However the right of a proper defense was (also) denied to Rajlich and the investigation started without him.
2. Vasik Rajlich author of Rybka filed a complaint to the FIDE Ethics Commission.
3. Rajlich asked the ICGA for an appeal which was denied to him.
4. In return Rajlich contributed a supplement to FIDE regarding the ICGA denial of his rightful appeal.
5. Creation of 2 polls, one strictly for chess programmers and an user poll regarding the ICGA denial of an appeal.
6. Refutation contributed to Chessvibes Part one, Part two on an ICGA document of 2013 rehashing its (old) arguments.
[ next subject ]