ChessBase: A Gross Miscarriage of Justice in Computer Chess

[BB+]

If ICGA would like to have their decision validated they should present their evidence to an independent group for review. It seems to me that the 'defense' has appeared ready to present their case so the situation and resulting decision would be more balanced.

I suggested something (vaguely) similar a few months ago. The main problem is that I doubt the 'defense' would be willing to have such a decision be definitive (and if it is not, why bother with such a "validation"?) -- they seem to prefer to argue endlessly in their forum. Rajlich skipped the ICGA process [Riis implies VR found it "biased"], and I don't see any reason to think he would submit to judgement elsewhere.

I had suggested Jonathan Schaeffer as a suitable person to head such a review (as Ken Thompson has already weighed in), though maybe you want to go further afield. For instance, having the independent group be some slashdot boffins should suffice, as they would understand the coding (and open source) issues. But as above, I doubt that VR and Rybka Forum would go for it, probably disputing the "independent" nature of such groups. Also, if one can't find a gratis review panel (unlikely once you leave the CC circle), it seems fair for the loser to shoulder the great majority of the associated costs.

[hyatt]

Please find ONE quote where I said the Rybka PST VALUES and the Fruit PST VALUES matched EXACTLY. Just one will do. You won't find such a quote, however, except in the archives of your mind, where reality diverges from what you want it to be.

[Prima]

Part four: http://www.chessbase.com/newsdetail.asp?newsid=7813


OK, regardless of where you stand on this issue, we can all agree that Bob needs a break from Rybka forum.

I've always appreciated Bob's witnessing of the facts, as demonstrated by copious [Fruit-Rybka] data comparisons & candidness. Personally, If I were in Bob's shoes, I'd reply and confute EVERY false posts ubiquitous only in the Rybka forum, appertaining to this issue. People need to know the truth. Bob IS doing a good job at that.

These are clearly 'cheap shots' & straw-man's attempts, by soren & the rest of Rybka moderators, to silence Bob by complaining of the quantity of his posts/day. Whoa!.....I guess that really proves Vas didn't copy-codes and is innocent
Like the IPPOLIT-censor era, they attempt to censor the quantity of posts by individuals proving Vas copied-CODES & lied about it, and/or disfavoring Rybka's dubious origins.

Their arguments (soren & Rybka moderators and advocates) are so faulty on so many levels, it's hilarious. And to think these folks actually fathered children! Is it any wonder why the world is so mentality-warped as it is today, or where such prevalence of mental illness stems from?

[hyatt]

were it not for Alan, Derrière, Chris, Ed and Jeroen's extremely rude goading, Bob wouldn't have posted as much as he did.

So bob can't be made responsible of his own actions?

Something like:

"It was not my fault that I posted so much about the case, it was Alan, Derrière, Chris, Ed and Jeroen's fault!"

IS a lame excuse.

How do I critique it, except to take issue with his targeting of Bob Hyatt's posting patterns in an irrelevant attempt to make Rybka's accusers seem freaky and desperate?

Well, bob does seem freaky and desperate, what with telling people they should suicide and all that.

So YOU want to perpetrate that lie as well? NO big surprise there, I don't suppose.

Never been "freaky or desparate" in my entire life, for the record.

[hyatt]

adding to the procedural issueshttp://www.chessvibes.com/reports/progr ... uit-issue/
http://www.chessvibes.com/plaatjes/rybk ... gation.pdf

Also Opinions of Panel Members

Since so much attention has been around obscuring the truth I would like to know the details of the proceedings.
Especially considering the panel consisted of the following members:
Open letter to the ICGA about the Rybka-Fruit issue signers
"We believe as an unauthorized Fruit derivative Rybka's entry into ICGA events has been contrary to the ICGA rules and the rules of fair play."
Fabien Letouzey, Zach Wegner, Mark Uniacke, Stefan Meyer-Kahlen, Ed Schröder, Don Dailey, Christophe Theron, Richard Pijl, Amir Ban, Anthony Cozzie, Tord Romstad, Ralf Schäfer, Gerd Isenberg, Johannes Zwanzger
It is also clear that both
RYBKA - FRUIT PANEL
Albert Silver, Amir Ban, Charles Roberson, Christophe Theron, Dariusz Czechowski ,
Don Dailey ,Fabien Letousky ,Frederic Friedel ,Gerd Isenberg ,Gyula Horvath ,Ingo Bauer ,Jan Krabbenbos ,Kai Himstedt , Ken Thompson ,Marcel van Kervinck ,Maciej Szmit ,Mark Watkins ,Mark Uniacke, Mincho Georgiev ,Olivier Deville, Omid David,Peter Skinner ,Ralf Schäfer,Richard Vida ,Richard Pijl ,Stefan Meyer-Kahlen,Thomas Mayer,Tord Romstad,Tom Pronk,Vladan Vuckovic,Wylie Garvin,Yngvi Björnsson,Zach Wegner

Also 1.3 Investigation Procedure. This investigation is peer reviewed, and led by a Secretariat of three members (including one past World Champion) appointed by the ICGA President and Board.

Shouldn't the members that had already made up their minds about the investigation have recused themselves from the proceedings if they were meant to be a fair and impartial investigation:
Purpose: To investigate claims that the chess playing program Rybka is a derivative of the chess programs Fruit and Crafty and violated International Computer Games Association (ICGA) Tournament rules. Rybka is a program by Vasik Rajlich.
Fruit was written by Fabien Letouzey. Crafty was written by Robert Hyatt


I again am not taking sides but want the truth to be as untainted as possible by bais. Can the proceedings and decisions by this group really be considered unbais?
Now that other programmers with equally substantial credentials are coming forward to raise concerns about the validity of the decision by ICGA I have to examine how the decision was made in addition to the presented evidence.

The ICGA asked a group of _programmers_ to analyze the evidence and prepare a written report to the ICGA board. We invited Vas to participate DURING the evidence gathering phase of the process. He refused. Once the report was finished, the ICGA then sent a copy of the report to Vas and asked for any defense he wanted to supply. He refused to supply anything. The ICGA board, which also DOES have former programmers (David himself worked on a program a long while back, as did Jaap at least) then looked at the report and decided it proved beyond any doubt that ICGA rule 2 had been broken. There were some well-known people that have not been active in competing for over 20 years, including Ken Thompson. Not one person that read the report, including Ken, thought there was any doubt. So the "panel" had no dissenting votes whatsoever. The panel's "vote-taking" however, was solely for our own benefit, as early votes showed that there were some missing pieces of information that would make the case stronger. By the time we finished, everyone that was active agreed. Then it was turned over to the ICGA for analysis and they reached their decision.

That not everyone voted formally was not unexpected. This was a NEW process. And when we formed the panel, we simply verified that we were dealing with "real people" and that they had some sort of expertise in computer chess (not all were programmers). But we did not try to impose any sort of "you must participate and vote on the final report" policy. Several that joined realized that the process was pretty technical, and pretty lengthy, and they decided they were unwilling/unable/uninterested in getting that deep into the investigation. It does not imply that they did not agree with the report. It implied that they were not willing to vote for whatever reason(s) they reservations about.

If we do this again, we will probably refine the "panel admission process" a bit to try to include those that will actually participate, but that's for the next time around. The process will be refined as it is used.

[lmader]

Part of the problem is that a lot of people are confused about the issues.

In an attempt to boil down all of the discussion into something easily understood, I think that we need some kind of Rybka Code Copying for Dummies FAQ, so that less technically oriented people can understand the discussion. Something like this, maybe (It seems like even the author of the Chessbase articles might benefit):

0) What is code copying in this context?
The goal of the ICGA investigation into Rybka was to determine if code copying occurred. To clarify, this does not mean "going through Fruit and using the ideas", which would be fine. This means actually cutting and pasting sections of code, or perhaps even starting with the Fruit code and making modifications.

1) But isn't some of the allegedly copied code irrelevant?
It does not matter if the evidence for code copying shows code that doesn't seem important to the strength of the engine, because the only point is to see if _any_ code copying happened. If any code copying happened, then VR has to explain himself.

3) Why does this matter?
It matters because it is a violation of Fruit's GPL to use _any_ code, and a violation of ICGA's rules to enter an engine of this nature.

2) How is the Fruit/Rybka code compared?
The only way to make the comparison is to disassemble the Rybka binary into a c code representation, and look for constructs that match the Fruit source code. The original Rybka source code, as written by the original programmer, is not recreateable in exact form by disassembling the binary. i.e. The resulting code produced from disassembly is c code, but you can't compare it word for word with the Fruit source code because c code can be structured in a variety of ways. What you can do is look for what is known as semantic equivalence - the code does the same thing even though it may be structured slightly differently. That is the purpose of the side-by-side comparisons in the submitted documents. i.e. The disassembled code is not "imaginary code", it is a functional representation of the binary and does indeed contain information that can be used for comparison.

3) The code comparison documents are long, tl;dr. What do these code comparisons show?
The documents show beyond any shadow of a doubt there are long sections of code that match semantically. This fact cannot be reasonably disputed.

4) What does all that mean?
The real question is whether or not all of this semantic equivalence constitutes enough evidence to prove that Rybka copied code. It is indeed possible for two programmers to implement a feature or algorithm such that their compiled form will look semantically equivalent. However the odds of this happening go down exponentially with the length of the code sample and number of occurrences. Thus the question could be restated like this: Is the amount of semantic equivalence shown in the documents far enough beyond what can be explained by normal occurrences to conclude that code was copied?

5) Once the above is understood, one can make a more informed conclusion. And it's pretty obvious that there is indeed a level of semantic similarity that is tough to dismiss. Programmers familiar with code comparison consider it impossible to explain any other way (code was copied).

(This discussion ignores other smoking guns like the use of unusual constants, etc.)

[Prima]

were it not for Alan, Derrière, Chris, Ed and Jeroen's extremely rude goading, Bob wouldn't have posted as much as he did.

How do I critique it, except to take issue with his targeting of Bob Hyatt's posting patterns in an irrelevant attempt to make Rybka's accusers seem freaky and desperate?

So YOU want to perpetrate that lie as well? NO big surprise there, I don't suppose.

Never been "freaky or desparate" in my entire life, for the record.

The desperation is actually from the Rybka advocates. Apparent by their need to post profuse, untruthful, and apparently uninformed, posts.

It is only in THEIR world where it is okay for anyone to post 5,000 fabrications and/or insult-posts per day and not considered excessive poster. Someone else post 1 factual/truthful post a day, THAT is considered "too much"?.... or does too much posts now translate to be a "lie"?...because so far, I haven't seen any distorted facts presented from your end, BB+'s, Zach's etc. Nor does anyone posting prolifically somehow changes the fact Vas was desperate enough, to begin with, to plagiarize codes and claim it 100% original as his work.

It's got to be the persistent raw truth that's too "copious" and "offensive" to them to handle.

[Rebel]

Please find ONE quote where I said the Rybka PST VALUES and the Fruit PST VALUES matched EXACTLY. Just one will do. You won't find such a quote, however, except in the archives of your mind, where reality diverges from what you want it to be.

By manipulative postings like this:

The evidence is _not_ based on "conjecture". It is based on specific analysis of Rybka and Crafty or Rybka and Fruit. There is no "interpretation" required. Have you actually _read_ Zach's and Mark's report? People keep saying "show me side by side comparisons." First page of Zach's report has _exactly_ that. Two columns. The comparison goes on for pages and pages. Side by side. Piece by piece...

It forced me (even when I was VIG at that time) to expose your debating techniques starting a new thread. You misused the innocence of your readers.

And from the same thread about PST's:

No, there was no code copying. however, there was "Data Copying" which is exactly the same thing in the given context. Not allowable.

The ICGA point of view was always the CODE copying of Fruit's PST initialization code.

Now try to wriggle yourself out of this.

Mission impossible but I am sure you will try.

Never mind.

But the irony of it all is that by your manipulative debating techniques I began to realize something was very rotten in Denmark and mainly thanks to you I had a change of heart in the case. If you had kept your mouth shut in Rybka forum most of this all had happened, no defending web-page, no Chess Base article. I am sure that after the CB article I am the ICGA most wanted

The relevant links then:

http://rybkaforum.net/cgi-bin/rybkaforu ... #pid356476

http://rybkaforum.net/cgi-bin/rybkaforu ... 22522;pg=1

[BTO7]

Only at Open Chess can one follow along the whole story with out a bunch of forum deletes or hiding the place to see. Like all the other chess forums do. As a chess enthusiast I am glad this site opened up. Txs Jeremy for the readers sake.

BT

[hyatt]

Please find ONE quote where I said the Rybka PST VALUES and the Fruit PST VALUES matched EXACTLY. Just one will do. You won't find such a quote, however, except in the archives of your mind, where reality diverges from what you want it to be.

By manipulative postings like this:

The evidence is _not_ based on "conjecture". It is based on specific analysis of Rybka and Crafty or Rybka and Fruit. There is no "interpretation" required. Have you actually _read_ Zach's and Mark's report? People keep saying "show me side by side comparisons." First page of Zach's report has _exactly_ that. Two columns. The comparison goes on for pages and pages. Side by side. Piece by piece...

It forced me (even when I was VIG at that time) to expose your debating techniques starting a new thread. You misused the innocence of your readers.

And from the same thread about PST's:

No, there was no code copying. however, there was "Data Copying" which is exactly the same thing in the given context. Not allowable.

The ICGA point of view was always the CODE copying of Fruit's PST initialization code.

Now try to wriggle yourself out of this.

Mission impossible but I am sure you will try.

Never mind.

But the irony of it all is that by your manipulative debating techniques I began to realize something was very rotten in Denmark and mainly thanks to you I had a change of heart in the case. If you had kept your mouth shut in Rybka forum most of this all had happened, no defending web-page, no Chess Base article. I am sure that after the CB article I am the ICGA most wanted

The relevant links then:

http://rybkaforum.net/cgi-bin/rybkaforu ... #pid356476

http://rybkaforum.net/cgi-bin/rybkaforu ... 22522;pg=1

Either dense or dishonest. Doesn't matter.

What we CLAIM happened is that Vas copied the Fruit PST initialization code, just as he copied other fruit code AND other crafty code. He then tweaked the base constants in said code until he was happy. Just as I did. Then he decided to get rid of the initialization code which would make it easier to do final tweaking. Again, just as I did. So he DID originally copy the fruit code. He modified it very slightly (changed the constants Zach pointed out) to produce his final values.

The proof is in Zach giving the original fruit code side by side with the fruit code that was modified by Zach to EXACTLY produce the Rybka PST values. To believe that Vas hit upon this EXACT formulation for each and every PST is a wild stretch of one's imagination. So the purpose of the PST section was to show the copied (but now not used) fruit pst code that produces the current statically initialized PST arrays.

That is one point of the report out of N. The report CLEARLY says that the code Zach gives is NOT in current Rybka. He is showing how little the Fruit code needs to be changed to produce Rybka's values. You certainly are not in the "most wanted" category. Perhaps "most dishonest" or something to that effect. You KNOW what the PST data is about. EVERYBODY knows what it is about. You and your meathead side-kick Chris keep claiming that Zach has presented "made up code, claiming it is in Rybka." Zach claimed no such thing, and in the first paragraph of the PST section one finds this:

Piece Square Tables
Piece square tables are a very simple technique used for basic evaluation. For every piece type and square, PSTs
have a value for that piece being on that square. Fruit uses a clear and simple but effective way of calculating the
tables. Looking at Rybka's PSTs, we will see that they are calculated using these exact same constants except with
different weights. Also, note that here too that the PST values are hardcoded into the Rybka executable file, they are
not calculated at startup like Fruit's. The code shown here is simply the functional equivalent; it calculates the Rybka
PSTs.

So no dishonesty on Zach's part, just on yours and Chris' because that statement has ALWAYS been at the front of the PST data, yet you want to claim that Zach says anything but that "bolded" part. The only one that is dishonest is you for making such blatantly and obviously false statements. Anyone can find Zach's PDF and verify the above is present and was when the report was originally made public.

If it takes comments by me to make you think, you have MAJOR issues. I tend to think for myself, whenever I want. You should try it. And you should try it on a cause that is not hopeless. Because this one is beyond that. Way too much evidence. Way too little defense by Vas, for obvious reasons.