Rybka 1.0 vs. Strelka

[Zach Wegner]

1)I think that loop list may be a derivative of fruit.

The only evidence that I have is the correlation table from the ccrl
that shows higher correlation between fruit and list relative to other pair of programs.
I do not know know about people who tried to reverse engineer loop list
and report their results.

I remember that Fritz made a significant improvement at the fruit time from Fritz8 to Fritz9 so I suggest to check Fritz9 and later Fritz versions.

I think that doing "drug test" only to rybka is simply unfair.

I do agree that Loop should be examined at least in the interest of fairness. However, I don't have it, and I have no interest in buying it or obtaining it illegally.

I also will note that reverse engineering is very hard. Completely dissecting a program would take a very long time, and I'm afraid I don't have the skill. I was able to find what I did in Rybka in large part due to Strelka and the disassembly posted by Rick Fadden.

I was also under the impression from somewhere that Tord Romstad had seen the source. I can't remember why though, so I could very well be wrong.

2)I also think that all the rules about GPL are unfair if rybka can be considered to be a derivative of fruit based on code that is not relevant to geneate moves like input_available() and they were part of the functions that were used as evidence.

Like it or not, those are the rules of the GPL. I agree that it's kind of silly to get in a rut over some 10 line function, but we're not just discussing the input_available function. The very first post in this thread doesn't mention any of that--only things that are directly related to choosing moves. Looking at the input_available function is just another drop of water in the pond. Though it is true that very many engines will have similar functions (even ZCT looks pretty similar), the probability is not 100%. So each little piece that we show to be the same between Fruit and Rybka 1.0 is just reducing the probability that Rybka is not a derivative.

Really though, this thread is getting pretty ridiculous. The issue isn't just a few lines of unimportant code. The original post showed areas where Rybka 1.0 has semantically identical code in the search logic. How many engines do you know limit the depth to 4 when there is only one legal move? How many do you know that mark a move as easy when at a certain depth it is more than 150 cp better than the next move? There are many more questions I could ask, and I will continue to until I get answers.

[swami]

Therefore if the charter states 'no allegation' then isn't it impossible to discuss issues of cloning in this forum in any form?

Cheers,
Andy

Theoretically I guess so, not that I necessarily agree with that.
"Discussion is fine, libel is not" is the stance that the current moderation team has taken, and I think that is fair to everybody.

Cheers, Graham.

Given the weight of material produced, I would encourage readers to report cases of potential libel using the ! icon.

Yes, I was going to reiterate this.

It would very much help the moderators if members start reporting posts they feel offensive.

Most of us have job that takes up nearly 10 hours so most probably we wouldn't read every single posts except those who are engaging in discussion about serious matters.

[bob]

I think that the discussions to this point would pose no problem for me as a moderator, I would leave them untouched. Perhaps here and there someone might get a bit insulting, although I don't recall such. But "the quest for truth" ought to outweigh "the fear of false accusations". Yes, blatant false claims need to be dealt with. But there is quite a bit of evidence here that should be discussed since more is being presented each day.

I was not particularly interested, and still am not, other than to correct some false information about detecting copied code. Faculty members deal with this on a regular basis and the issues are not exactly new.

Either new information will continue to come from the discussion, or it will die of its own accord...

Thanks for your opinion, Bob. very much appreciated.

Looks like we are in complete agreement on this. So I guess Chrisw's sticky notice "Discussion is fine, libel not" worked very well and many people have shown restraint and have kept their posts under control.

There is no way that these discussions won't be offensive in some way, because in the clone thread, the underlying premise is that someone has copied something that should not be copied. But perhaps a legal opinion might be informative here. A lawyer friend of mine had called me about something and I chatted a few minutes about the ongoing discussions. And he pointed out that _somewhere_ such accusations have to be made. I asked "OK, if they come out in a courtroom, what happens, particularly if they are proven false?" He said "that is privileged discussion and you can't be sued for something you say under oath in court. You can be prosecuted if it is proven that you perjured yourself, but that is all." And in thinking about that, that should be the way this is conducted. If accusations can not be made, they can not be discussed, proven or disproven, and the situation does not get improved as a result.

[chrisw]

1)I think that loop list may be a derivative of fruit.

The only evidence that I have is the correlation table from the ccrl
that shows higher correlation between fruit and list relative to other pair of programs.
I do not know know about people who tried to reverse engineer loop list
and report their results.

I remember that Fritz made a significant improvement at the fruit time from Fritz8 to Fritz9 so I suggest to check Fritz9 and later Fritz versions.

I think that doing "drug test" only to rybka is simply unfair.

I do agree that Loop should be examined at least in the interest of fairness. However, I don't have it, and I have no interest in buying it or obtaining it illegally.

I also will note that reverse engineering is very hard. Completely dissecting a program would take a very long time, and I'm afraid I don't have the skill. I was able to find what I did in Rybka in large part due to Strelka and the disassembly posted by Rick Fadden.

I was also under the impression from somewhere that Tord Romstad had seen the source. I can't remember why though, so I could very well be wrong.

2)I also think that all the rules about GPL are unfair if rybka can be considered to be a derivative of fruit based on code that is not relevant to geneate moves like input_available() and they were part of the functions that were used as evidence.

Like it or not, those are the rules of the GPL. I agree that it's kind of silly to get in a rut over some 10 line function, but we're not just discussing the input_available function. The very first post in this thread doesn't mention any of that--only things that are directly related to choosing moves. Looking at the input_available function is just another drop of water in the pond. Though it is true that very many engines will have similar functions (even ZCT looks pretty similar), the probability is not 100%. So each little piece that we show to be the same between Fruit and Rybka 1.0 is just reducing the probability that Rybka is not a derivative.

Really though, this thread is getting pretty ridiculous. The issue isn't just a few lines of unimportant code. The original post showed areas where Rybka 1.0 has semantically identical code in the search logic. How many engines do you know limit the depth to 4 when there is only one legal move? How many do you know that mark a move as easy when at a certain depth it is more than 150 cp better than the next move? There are many more questions I could ask, and I will continue to until I get answers.

Semantically identical?! You mean containing similar ideas? Ideas are fine.

Suppose hypethetically we're basing our engine version 1 beta on ideas from another program(s).

We write the code and we steal, beg or borrow some parameters from elsewhere. Null move depth from Crafty = 2. "Easy" move from Fruit = 150. Limit search when only one move from Fruit (or Rebel, doesn't Ed do something with 4?) = 4. etc. Or knight = 300 from everybody. Or double pawns = 25 from somewhere else etc.

Exactly what licence does this breach?

[Zach Wegner]

Chris, please read the original post in this thread. Everything that was posted was taken directly from Fruit. We obviously can't look at the sources of Rybka 1.0, but we can take a look at the disassembly. There isn't just a parameter here, a parameter there. It's an entire block of code that is the same. And really, that's just the tip of the iceberg. What has been presented is very preliminary, I can only do so much disassembly. But from what I have seen, the odds are becoming _very_ small that Vasik only copied "ideas".

[chrisw]

Chris, please read the original post in this thread. Everything that was posted was taken directly from Fruit. We obviously can't look at the sources of Rybka 1.0, but we can take a look at the disassembly. There isn't just a parameter here, a parameter there. It's an entire block of code that is the same. And really, that's just the tip of the iceberg. What has been presented is very preliminary, I can only do so much disassembly. But from what I have seen, the odds are becoming _very_ small that Vasik only copied "ideas".

what did you mean by "semantically identical", because it sounds like "identical in an ideas-ey sort of way", similar meaning, but different implementation.

[Zach Wegner]

It was just a reference to what Bob said earlier. The meaning is functionally identical, but possibly slightly modified on a source level. They could be identical in the source too, but we can't tell...

[Rolf]

1)I think that loop list may be a derivative of fruit.

The only evidence that I have is the correlation table from the ccrl
that shows higher correlation between fruit and list relative to other pair of programs.
I do not know know about people who tried to reverse engineer loop list
and report their results.

I remember that Fritz made a significant improvement at the fruit time from Fritz8 to Fritz9 so I suggest to check Fritz9 and later Fritz versions.

I think that doing "drug test" only to rybka is simply unfair.

I do agree that Loop should be examined at least in the interest of fairness. However, I don't have it, and I have no interest in buying it or obtaining it illegally.

I also will note that reverse engineering is very hard. Completely dissecting a program would take a very long time, and I'm afraid I don't have the skill. I was able to find what I did in Rybka in large part due to Strelka and the disassembly posted by Rick Fadden.

I was also under the impression from somewhere that Tord Romstad had seen the source. I can't remember why though, so I could very well be wrong.

2)I also think that all the rules about GPL are unfair if rybka can be considered to be a derivative of fruit based on code that is not relevant to geneate moves like input_available() and they were part of the functions that were used as evidence.

Like it or not, those are the rules of the GPL. I agree that it's kind of silly to get in a rut over some 10 line function, but we're not just discussing the input_available function. The very first post in this thread doesn't mention any of that--only things that are directly related to choosing moves. Looking at the input_available function is just another drop of water in the pond. Though it is true that very many engines will have similar functions (even ZCT looks pretty similar), the probability is not 100%. So each little piece that we show to be the same between Fruit and Rybka 1.0 is just reducing the probability that Rybka is not a derivative.

Really though, this thread is getting pretty ridiculous. The issue isn't just a few lines of unimportant code. The original post showed areas where Rybka 1.0 has semantically identical code in the search logic. How many engines do you know limit the depth to 4 when there is only one legal move? How many do you know that mark a move as easy when at a certain depth it is more than 150 cp better than the next move? There are many more questions I could ask, and I will continue to until I get answers.

But you in special wont. Or on what experience you base your expectations? If I were Vas I wouldnt go to China. So easy. The Junior will be Wch and all are happy. The whole World knows that this is a hoax. The strongest ever is Rybka. No matter what you asked. Another point is that you ask here people who cant give you answers. But that doesnt interest you. Well, so you show where the people are who want to see such constant shower like auditing.

[chrisw]

It was just a reference to what Bob said earlier. The meaning is functionally identical, but possibly slightly modified on a source level. They could be identical in the source too, but we can't tell...

Bob was talking earlier about detection of students who produced modified variants of code for assignments. Even if massively modified, Bob was actually after algorithmic theft with rewritten or semi-rewritten code. If detected they failed or whatever. Not unreasonable, students are expected to come up with their own ideas.

However, you're talking about possible violations on a legal or contractual level. Of using source code protected under GPL.

I thought we established in an earlier argument that if all the source was re-written - ie. the rewriter read and understood the earlier source, then rewrote the code, that that was ok under GPL, albeit not ok at Uni Alabama.

[Rolf]

There is no way that these discussions won't be offensive in some way, because in the clone thread, the underlying premise is that someone has copied something that should not be copied. But perhaps a legal opinion might be informative here. A lawyer friend of mine had called me about something and I chatted a few minutes about the ongoing discussions. And he pointed out that _somewhere_ such accusations have to be made. I asked "OK, if they come out in a courtroom, what happens, particularly if they are proven false?" He said "that is privileged discussion and you can't be sued for something you say under oath in court. You can be prosecuted if it is proven that you perjured yourself, but that is all." And in thinking about that, that should be the way this is conducted. If accusations can not be made, they can not be discussed, proven or disproven, and the situation does not get improved as a result.

And Bob, just this to my many others: there is no such thing called friendship and respect in this scene, no? And Vas has gone commercial and that is what happens but what you can not estimate. True? And did you also ask your law expert what happens in court if a campaign like man hunt could be proven? Is that also allowed in the Law in the USA? In such a public forum? Where only writing affords registration??? Are there no internet laws in the States? And man hunt in the style of virtual witch hunt is tolerated? Not to speak of business damages... Why not seeking a sort of peace? But ok with your team that couldnt be the goal. ..

What you wrote above in boald letters from my side, means that Zach and others are actively commiting witch-hunt nothing else. Because this has absolutely no relevance when he said but I will continue to ask as long as I get answers. Only he wont get some. So this is a fine forum for Zach. For how long this will bew supported? Forever? <lol>