Fwd: Open letter to the CSVN

[Terry McCracken]

I'm very disappointed with you Ed.

[wgarvin]

A summarizing of what is wrong with the evidence:

1. PST's -> debunked

Only in your mind. Occam's Razor dictates that VR copied the PST along with the rest of Fruit's eval, and then tuned the constants (like he did for the rest of the eval). Presumably he copied it to save himself time, although maybe it was because he knew all of the eval including PST worked together pretty well in Fruit. Yes its circumstantial, and we can only speculate about the reason why VR copied the PST, but there's no doubt at all in my mind that he did copy it. The whole structure is the same, and that was not found to be true for any other contemporary programs that were examined. Nothing you or Chris W. or Miguel has brought up about the PST comes even close to "debunking" it.

2. Order in EVAL -> missing

Not very significant. What features are present is more significant. The evidence showed a much closer correspondence in features between Fruit 2.1 and Rybka 1.0 than any other pair of programs examined.

3. EVAL -> going through them one by one, so far all debunked.

This is entirely your wishful thinking at work. Nothing has been debunked.

3. The panel never checked the BB and Zach documents on an assembler level for correctness.

You are essentially blaming us for not independently repeating the work Zach and Mark had already done (painstakingly, over months). Some of us did read the disassemblies, though. The Rybka 1.0 Beta evaluation features are not too hard to interpret in the disassembly. IIRC, the eval was very nearly unchanged up to about 2.3.0. By 2.3.2a, it was starting to see some larger changes, however, many "telltale" scorepair constants were still present in the disassembly, showing that it was still a derivative.

What Zach, Mark Watkins researched was the Rybka chess program. With the "Vas is guilty" already branded in their minds they found zillions of traces of "a" chess program doing all the same things other (good) programs also do. Unfortunately they called all these traces Fruit. Classic case of a tunnel vision.

I'm not sure they're the ones with tunnel vision here.

I am not going to defend myself here on CCC, I am at Rybka forum in case any programmer is interested in the debate, just wanted to correct Bob's misinformation.

Ed

Good. Refuting your nonsense on this little cell phone keypad is time consuming. By the way, the Rybka forum is not exactly unbiased in this matter.

If anyone wants to dispute the statements I've made in this post, I invite them to do so either here at talkchess, or over at open-chess.org. I don't read the Rybka forum though, so don't write your rebuttal there unless you actually prefer that I don't get to see it.

[geots]

Ed and Chris lack a great deal in terms of honesty. Ed started questioning the data by saying "I still believe Vas to be guilty." Now he is not questioning the data, he is questioning the motives of the panel, ICGA and programmers. They've found NOTHING wrong with the evidence, to date. Right now it is more personal than factual...

You should not call people dishonest and then start being dishonest yourself by saying we found nothing wrong, no good logic

A summarizing of what is wrong with the evidence:

1. PST's -> debunked
2. Order in EVAL -> missing
3. EVAL -> going through them one by one, so far all debunked.
3. The panel never checked the BB and Zach documents on an assembler level for correctness.

What Zach, Mark Watkins researched was the Rybka chess program. With the "Vas is guilty" already branded in their minds they found zillions of traces of "a" chess program doing all the same things other (good) programs also do. Unfortunately they called all these traces Fruit. Classic case of a tunnel vision.

I am not going to defend myself here on CCC, I am at Rybka forum in case any programmer is interested in the debate, just wanted to correct Bob's misinformation.

Ed

Ed, you could have saved a lot of time and bandwidth by listing the things the panel did RIGHT.

Bob's misinformation!!?? Jeez, you got a horrible case of political correctness. I hope you are taking something for it.

[bob]

NO, to late!

all sources from the OPEN LETTERS programmers are now to check. This group of programmers can not start such things without to give here own sources for a check.

THIS IS absolutlely clear!
Persons which say A have to say B too and in this case they have to say C and D.

Please no burning of witches again.
But perhaps different computer chess people like the middle eges.

We had a CCT with > 50 entrants. WHO is going to compare all those sources with each other, plus compare them against all the other open-source engines that are out there? Who is going to pay them? Who has that much time. It is fine to make suggestions, so long as you realize that the suggestion is beyond ridiculous because of the time required to actually do the job.

How many open source programs are there? 200? 300? Crafty is 50,000 lines of code. You are going to compare Crafty against 299 other programs, that is only comparing 50,000 x 299 = 15M lines of code compared. Now you know Crafty is unique. Do this for the other 50 entries? We are getting close to one billion comparisons. Can you compare two lines of source in 1 second? We only need 1 billion seconds, or 11,400 DAYS, non-stop. 32 years. We need 32 people working 24 hours a day to do this in one year. Can we REALLY compare 1 line per second?

Obviously what you suggest is not just impractical, but IMPOSSIBLE...

[bob]

Sorry, Stefan Meyer-Kahlen I believe and Theo van der Storm checked all important parts of Ruffian sources before we win the tournament in Leiden. Absolutly clear opinion, no clone and nothing to do with Crafty or other free sources. Theo public his opinion in these times. Furthermore, PerOla sent this parts to two other persons after Vincent comes with such a story. I know, you and Vincent discuss about in TalkChess. No sorry from Vincent later. Sources from Ruffian are checked and clean ... bad Bob ... without any proof and after the programs was checked you comes today again which such a story!

Robert, you have really such information?
Your informations because Bionic, what you wrote yesterday, are also wrong. You are meaning Bionic Impact, the version, better complete other program, which played in Leiden with Crafty SMP code without your permission. Bionic is an other engine! Bionic and Bionic Impact isn't the same. Later Hans Secelle help with Ant. Ant by Tom Vijlbrief and Hans is for sure also a clone of Crafty.

You have a big problem Bob.
Honest: Each alpha/beta engine is a clone!
That is the truth and about his one you should thinking about it.

You have to start the chain of causation on the right position and not on a position you like! I have a very high opinion about your persons and all what you do but more and more, with new discuss to this topic you lost it. From yourself I await a bit more as to give other programmer the information that they have to blocked the best organize tourney we have.

I await an sorry!

What did I say that was "bad"? Didn't MENTION ruffian whatsoever...

[bob]

Hi Thomas,
I consider Vas to be a programmer and not a hacker / cloner.

I think that there are enough original parts in Rybka to consider Vas to be a programmer(same for Robert Houdart).

I think that the case of Rybka is clearly in the grey area that means that there is no agreement between chess programmers if Vas is quilty.

I did not investigate the evidence but I understand that some programmers like Ed of Rebel and Chris of CStal read the evidence and it did not convince them that Vas is quilty.

It is not something clear like other cases(for example all agree that the similiarity between Toga and Fruit is bigger than the similiarity between Rybka and Fruit).

Ed and Chris lack a great deal in terms of honesty. Ed started questioning the data by saying "I still believe Vas to be guilty." Now he is not questioning the data, he is questioning the motives of the panel, ICGA and programmers. They've found NOTHING wrong with the evidence, to date. Right now it is more personal than factual...

Be careful. It would be in your best interests not to get into a character debate with them- and you should be really careful about bringing honesty into any discussion.

Let me add that generally if more than a few people are involved in an issue, you end up with a train wreck. That was proven recently. So that is the reason I have been in touch with Cock numerous times lately. I knew nothing of their group, only that the programmers who signed the open-letter to them were critical of their decisions concerning the Rybka issue. And also critical of the way they have handled past issues.

If the "open-letter guys" are that critical of them, it's a given they are making some good decisions and doing a hell of a good job at whatever they do. That peaked my interest.

I'm willing to debate any point they choose. However, debate is not possible when dealing with two people as dishonest as they are.

An example from the Rybka forum. Ed stated categorically that I could not reverse engineer anything, in spite of programming in assembly language for 43+ years now, in spite of having written more than one compiler from scratch. And in spite of helping debug some of the gcc long long code when I started to use that compiler for Crafty in 1994. He picked out a piece of code from Zach's report and asked me to "identify the offset into the rybka binary where this code is found." Not only did I do that, I broke the assembly language code down, after locating it, and matched it up line for line with the C code given in Zach's report. Ed then claimed the test was no good, that I had cheated because the code had 4 if-tests and I just looked for 4 test instructions and "hoped it matched." Even though I had matched the asm line for line with the C to SHOW that it matched, EXACTLY.

That is what I call "dishonesty". It wasn't about whether I could really interpret assembly language or not. It was just to cast aspersions on me, on the panel, on the process, and on the evidence. That's all he is good for, it seems...

a "lie-machine". He accused me of copying ip/robolito code. I challenged him to prove this since my source is open. Zero. Zilch. Ran and hid. As always. Same here. Zero. Zilch.

[bob]

Ed and Chris lack a great deal in terms of honesty. Ed started questioning the data by saying "I still believe Vas to be guilty." Now he is not questioning the data, he is questioning the motives of the panel, ICGA and programmers. They've found NOTHING wrong with the evidence, to date. Right now it is more personal than factual...

You should not call people dishonest and then start being dishonest yourself by saying we found nothing wrong, no good logic

A summarizing of what is wrong with the evidence:

1. PST's -> debunked
2. Order in EVAL -> missing
3. EVAL -> going through them one by one, so far all debunked.
3. The panel never checked the BB and Zach documents on an assembler level for correctness.

PST's -> NOT debunked. Still as solid today as when the report came out.
Order in eval -> you have no clue
eval -> all debunked? In your shallow mind, perhaps. I posted the FIRST actual piece of rybka code compared to fruit code, on the rybka forum. You claim THAT was "debunked". Liar... And I don't like to use that term indiscriminately, but it absolutely fits you in this case. NOTHING was debunked except for the nonsense you and/or chris post. THAT has been debunked dozens of times now.

the last statement is false. I told you what I did. You said I could not do that. You gave me a silly test. I did far more than you asked. You then imply your test was no good. Dishonestimus maximus - from the roadrunner cartoon...

What Zach, Mark Watkins researched was the Rybka chess program. With the "Vas is guilty" already branded in their minds they found zillions of traces of "a" chess program doing all the same things other (good) programs also do. Unfortunately they called all these traces Fruit. Classic case of a tunnel vision.

I am not going to defend myself here on CCC, I am at Rybka forum in case any programmer is interested in the debate, just wanted to correct Bob's misinformation.

Ed

"correct bob's misinformation". Bob has been correcting yours for a couple of months now, for anyone that wants to run over to the Rybka Cesspool and look...

[bob]

A summarizing of what is wrong with the evidence:

1. PST's -> debunked

Only in your mind. Occam's Razor dictates that VR copied the PST along with the rest of Fruit's eval, and then tuned the constants (like he did for the rest of the eval). Presumably he copied it to save himself time, although maybe it was because he knew all of the eval including PST worked together pretty well in Fruit. Yes its circumstantial, and we can only speculate about the reason why VR copied the PST, but there's no doubt at all in my mind that he did copy it. The whole structure is the same, and that was not found to be true for any other contemporary programs that were examined. Nothing you or Chris W. or Miguel has brought up about the PST comes even close to "debunking" it.

2. Order in EVAL -> missing

Not very significant. What features are present is more significant. The evidence showed a much closer correspondence in features between Fruit 2.1 and Rybka 1.0 than any other pair of programs examined.

3. EVAL -> going through them one by one, so far all debunked.

This is entirely your wishful thinking at work. Nothing has been debunked.

3. The panel never checked the BB and Zach documents on an assembler level for correctness.

You are essentially blaming us for not independently repeating the work Zach and Mark had already done (painstakingly, over months). Some of us did read the disassemblies, though. The Rybka 1.0 Beta evaluation features are not too hard to interpret in the disassembly. IIRC, the eval was very nearly unchanged up to about 2.3.0. By 2.3.2a, it was starting to see some larger changes, however, many "telltale" scorepair constants were still present in the disassembly, showing that it was still a derivative.

What Zach, Mark Watkins researched was the Rybka chess program. With the "Vas is guilty" already branded in their minds they found zillions of traces of "a" chess program doing all the same things other (good) programs also do. Unfortunately they called all these traces Fruit. Classic case of a tunnel vision.

I'm not sure they're the ones with tunnel vision here.

I am not going to defend myself here on CCC, I am at Rybka forum in case any programmer is interested in the debate, just wanted to correct Bob's misinformation.

Ed

Good. Refuting your nonsense on this little cell phone keypad is time consuming. By the way, the Rybka forum is not exactly unbiased in this matter.

If anyone wants to dispute the statements I've made in this post, I invite them to do so either here at talkchess, or over at open-chess.org. I don't read the Rybka forum though, so don't write your rebuttal there unless you actually prefer that I don't get to see it.

You ought to visit the RF even if you don't sign up, just to see what nonsense goes on. When this "nobody verified Mark's/Zach's work came up, I pointed out that I had done some spot-checking. I was not about to check everything when every spot-check I did showed a perfect match to what Zach or Mark had reported. Ed claimed I could not RE, even with 43+ years of asm and compiler writing experience. He gave me a simple "test". I did about 10x more than what he wanted. He then found fault with it by "I suspect you did this, or I suspect you assumed that, I think you were lucky to spot this..." I took the C from Zach's report, found where it was (it was not in the commented code Mark had published), and matched it up line for line with the assembly language in the pawn eval code. That sort of shows what is going on over there. NO PROOF will be enough. Somehow, Ed thinks he and Chris have debunked "everything". Perhaps in their minds. Not in terms of technical proof...

What a pair.

BTW I did mention that during the investigation, you had gone thru the binary and figured out where every procedure was located, address-wise and giving them sensible names. Of course that doesn't mean anything either. Their denial is intense...

[wgarvin]

BTW I did mention that during the investigation, you had gone thru the binary and figured out where every procedure was located, address-wise and giving them sensible names. Of course that doesn't mean anything either. Their denial is intense...

I was using the function names from Strelka 2.0, whose source turned out to be a very close match for most of Rybka 1.0 Beta. It does have a few modifications (underpromotions for example) so you can't just blindly assume that the two are the same, but if anyone wants to study disassemblies of Rybka 1.0 Beta further, the Strelka code is a useful roadmap.

[Terry McCracken]

BTW I did mention that during the investigation, you had gone thru the binary and figured out where every procedure was located, address-wise and giving them sensible names. Of course that doesn't mean anything either. Their denial is intense...

I was using the function names from Strelka 2.0, whose source turned out to be a very close match for most of Rybka 1.0 Beta. It does have a few modifications (underpromotions for example) so you can't just blindly assume that the two are the same, but if anyone wants to study disassemblies of Rybka 1.0 Beta further, the Strelka code is a useful roadmap.

The very code that caused Vas to let the proverbial Cat out of the Bag back in '07, complaining Strelka was a Rybka clone.