Robbolito 0.09 New Edition VS Rybka 3

[Rolf]

Larry Kaufmann posted the following in the Rybka forum:

Well, to be precise a new program that was clearly made by decompiling Rybka 3 and putting it back together with changes to the evaluation function published its source code. So not the entire R3 code is public, but most of it other than the exact evaluation function is. The "author" does not publicly admit that his program was taken from R3, but Vas has said so and I can tell you that the similarities are so overwhelming that any claim to the contrary is a joke.

But that is excellent news! Just show me some of those overwhelming similarities. I'd be totally satisfied knowing Ippolit is a Rybka clone, I find the current situation awkward.

PS. I am a bit sceptic about Larry's words. How can the similarities be overwhelming for him when all he did was listen to what Vas said to him.

Where is the original message fro SzG?? Where Graham wrote his message?? All I know is that I quoted Larry. And to the intriguing question if Larry did only listen to what Vas said, this is untrue because Larry spoke about his __own__ code that was stolen and used in rolito. Understood SzG?

[Dr.Wael Deeb]

It is only legally questionable when legitimate proof of the clone claim is provided. That would _not_ be hard to do. If it _is_ a clone, how would it reveal any Rybka secrets since Robo* apparently has them inside already. So exactly what justification can be given for Vas _not_ providing some clear-cut examples of identical code? I can't think of a single valid explanation, unless Robo is not really a clone.

There is a difference because how else for you it had any importance that Vas claimed Strelka? Only then you began your argument although the Osipov figure had said the same before and after the claim from Vas. I am not alawyer but perhaps for you the difference is that without the claim you would do something which you shouldnt do but with the claim you can exploit it. Sorry it's more a question from my side, not at all a critic or anything like this against you. But you must know the difference. Please explain it.

I don't know what you are talking about. Vas claimed Strelka to be his code. Someone compared this to fruit and said "aha, lots of copied/similar code". That started the discussion. Later others went directly to Rybka 1 binary and compared to fruit. End of that story.

Here, Vas has only claimed Robo* is a clone. No proof. No nothing. Makes the claim absolutely meaningless.

You misunderstand. THe point is if data from Vas would have any impact. Strelka showed a different result, he was even more accused. Same now. What all are asking you now is look at the code (without allowance from Vas) and join Don and PLEASE then help to get over the mess, Bob. Save computerchess now. Please.

Of course, if he has even more to "hide" then you could be correct. But then that would not exactly be flattering for him, would it? You can't have it all ways. This either falls on one side of the line or the other. We know what happened with Rybka 1 and fruit. And no amount of hand-waving and such is going to change that at all. We don't yet know what happened with Robo* and Rybka 3. And I suspect we are not going to know. What that implies I don't know, and don't care.

Well, to be precise a new program that was clearly made by decompiling Rybka 3 and putting it back together with changes to the evaluation function published its source code. So not the entire R3 code is public, but most of it other than the exact evaluation function is. The "author" does not publicly admit that his program was taken from R3, but Vas has said so and I can tell you that the similarities are so overwhelming that any claim to the contrary is a joke.

Does this help you or not?

Not one iota.

Sure you need details so that they could be utilized. But the text is from the Honorable GM Larry Kaufman, and you will perhaps treat it with respect please. Larry knows what he's talking about. But again unless the whole original source code of R isnt sent to you you will call Vas a liar. The story is well known. Tolerating anon dirt and scapegoating Brightness. -> Witchhunting.

Make that two liars when counting you in....

[Dr.Wael Deeb]

Larry Kaufmann posted the following in the Rybka forum:

Well, to be precise a new program that was clearly made by decompiling Rybka 3 and putting it back together with changes to the evaluation function published its source code. So not the entire R3 code is public, but most of it other than the exact evaluation function is. The "author" does not publicly admit that his program was taken from R3, but Vas has said so and I can tell you that the similarities are so overwhelming that any claim to the contrary is a joke.

But that is excellent news! Just show me some of those overwhelming similarities. I'd be totally satisfied knowing Ippolit is a Rybka clone, I find the current situation awkward.

PS. I am a bit sceptic about Larry's words. How can the similarities be overwhelming for him when all he did was listen to what Vas said to him.

Where is the original message fro SzG?? Where Graham wrote his message?? All I know is that I quoted Larry. And to the intriguing question if Larry did only listen to what Vas said, this is untrue because Larry spoke about his __own__ code that was stolen and used in rolito. Understood SzG?

Aha,now Larry's code is stolen....hmmm,I hope that they didn't steal his pants also because that would be a great shame for the whole Rybka team

[kingliveson]

I think there is a good way to show that it is a Rybka 3 clone, even though I have absolutely no doubt. By looking at a lot of games between the two engines I have yet to see a position where both engines do not agree, except of course with some of the known bugs of Rybka 3 ... mostly the EGTB hash bug and the blind bishop bug.

Of all the statements that have been made, this has to be the silliest. if both engines agree on the same position like Rybka vs Rybka would do, how do you explain Rybka getting trashed? I mean this engine on a single core is as strong or stronger than Rybka on multiple cores. Try again...

If there was a MV mode maybe we could look at the output of the best moves and correlate that to see how both engines assess the position. Obviously there would have to be a speed correction factor since the clone runs the executable much faster. I would not be surprised if the equivalent move selection would be by far higher than would be expected from a non clone. Maybe there is a good reason there is no MV mode as things would be too obvious. Some knowledge in Robbo is missing but that knowledge seems to be compensated for by speed. I have yet to see a game where Rybka 3 was outplayed by evaluation and not by faster speed.

More fallacy...I don't believe you are a programmer or know what you are talking about. Zappa Mexico II is 1.38 MB and Rybka 3 is 2.80 MB. Zappa of all top engines is probably the worst at playing blitz. Yes, there is a good reason Robbo* does not have some of these other features; it is an engine under development.

Rybka 3 has some weaknesses, of course the bugs are well documented, but it also has a weakness in King safety and in bishop pair evaluation ... if you look at games with Rybka 3 against DS 12 and Naum 4 and Stockfish ... you will see that even R3 on 8 cores will lose games due to mis evaluation of the positions being mentioned against single core programs of those versions. This does not happen with R3 8 cores against Robbo 1 core ... which is consistent with what I think a clone's results would be. I think if we set some positions and factor in whatever speed advantage Robbolito has over R3 single core, that the play would be virtually identical.

Same ridiculous argument from the first paragraph. Search through the forums, you will see different time controls where Rybka is punished repeatedly. Give me a time control and conditions. I am willing to carry out the test and post the results. Come back when you have evidence...

[kingliveson]

I can't imagine how adults can be so gullible and follow by blind devotion without pausing and thinking for a second. It is almost getting embarrassing when you read some of the comments. Bogus and ridiculous claims are made without backing them up. If you merely express reasonable doubt, you are labeled anti xxxxx. When someone hints that you have done wrong, and you come out and say you have not done wrong, who should describe what happened and provide proof?

[bob]

It is only legally questionable when legitimate proof of the clone claim is provided. That would _not_ be hard to do. If it _is_ a clone, how would it reveal any Rybka secrets since Robo* apparently has them inside already. So exactly what justification can be given for Vas _not_ providing some clear-cut examples of identical code? I can't think of a single valid explanation, unless Robo is not really a clone.

There is a difference because how else for you it had any importance that Vas claimed Strelka? Only then you began your argument although the Osipov figure had said the same before and after the claim from Vas. I am not alawyer but perhaps for you the difference is that without the claim you would do something which you shouldnt do but with the claim you can exploit it. Sorry it's more a question from my side, not at all a critic or anything like this against you. But you must know the difference. Please explain it.

I don't know what you are talking about. Vas claimed Strelka to be his code. Someone compared this to fruit and said "aha, lots of copied/similar code". That started the discussion. Later others went directly to Rybka 1 binary and compared to fruit. End of that story.

Here, Vas has only claimed Robo* is a clone. No proof. No nothing. Makes the claim absolutely meaningless.

You misunderstand. THe point is if data from Vas would have any impact. Strelka showed a different result, he was even more accused. Same now. What all are asking you now is look at the code (without allowance from Vas) and join Don and PLEASE then help to get over the mess, Bob. Save computerchess now. Please.

Of course, if he has even more to "hide" then you could be correct. But then that would not exactly be flattering for him, would it? You can't have it all ways. This either falls on one side of the line or the other. We know what happened with Rybka 1 and fruit. And no amount of hand-waving and such is going to change that at all. We don't yet know what happened with Robo* and Rybka 3. And I suspect we are not going to know. What that implies I don't know, and don't care.

Well, to be precise a new program that was clearly made by decompiling Rybka 3 and putting it back together with changes to the evaluation function published its source code. So not the entire R3 code is public, but most of it other than the exact evaluation function is. The "author" does not publicly admit that his program was taken from R3, but Vas has said so and I can tell you that the similarities are so overwhelming that any claim to the contrary is a joke.

Does this help you or not?

Not one iota.

Sure you need details so that they could be utilized. But the text is from the Honorable GM Larry Kaufman, and you will perhaps treat it with respect please. Larry knows what he's talking about. But again unless the whole original source code of R isnt sent to you you will call Vas a liar. The story is well known. Tolerating anon dirt and scapegoating Brightness. -> Witchhunting.

If you have something useful to offer in this debate, feel free to do so. Otherwise, simply shut up.

I personally know Larry and have for _many_ years. But he only knows about some values he sent. He does not know _if_ they were used. He has not seen any source code. So that comment is completely unhelpful, particularly when he says "they are similar". Similar does _not_ mean "identical" which is what a clone would most likely be in that regard. This is a technical issue, not an issue for debate. There is one and only one truth. All I am seeing is multiple opinions with zero supporting data. So stay out of the debate unless you have something _new_ to add. Continually repeating the same things, over and over, is anything but helping Vas' case.

[bob]

So exactly what justification can be given for Vas _not_ providing some clear-cut examples of identical code? I can't think of a single valid explanation, unless Robo is not really a clone.

I can think of a few.

- He doesn't care if CCC thinks its a clone or not.
- He sees the debate as beneficial to him.
- He's focusing on other things.
- He sees clearly that providing proof won't help.

For that last point, recall the strelka/fruit/rybka issue. Proof was provided and did little to accomplish anything. Witness also the recent debates here, how some live in their little world with certain laws. The laws are unchanging, regardless of evidence to the contrary-- the evidence deconstructs their world-view, and they can't cope with that.

Actually the rybka/strelka debate _was_ put to rest, showing that strelka did come from Rybka 1. Unfortunately, it also showed that Rybka 1 came from Fruit, which I assume he did not want known.

[bob]

How about this if you're not prepared to believe people like Vas, Larry, Don, Gian-Carlo and others. If you respect Christopher Conkie as a good clone detector (as most seem to do), go and talk to him.

All of the mentioned guys just _think_ it's a Rybka clone, noone showed a single similary. Just because there is no one.

The people repeated what VR said.

Chris even said somewhere he thinks I am right, as long as there is no evidence, Robbo should be treated like it is no clone.

Fair enough. As I keep repeating, people will believe what they want to believe, no matter who says what.

"Opinions are like assholes. Everybody has one, and nobody wants to look at anybody elses."

[Rolf]

I can't imagine how adults can be so gullible and follow by blind devotion without pausing and thinking for a second. It is almost getting embarrassing when you read some of the comments. Bogus and ridiculous claims are made without backing them up. If you merely express reasonable doubt, you are labeled anti xxxxx. When someone hints that you have done wrong, and you come out and say you have not done wrong, who should describe what happened and provide proof?

Ok, I tell you what is embarassing. If Bob calls Vas a liar and the expects that today business as usual. Look, I'm a psychologist, I have a daily situation that people fight my innocent proposals or interpretations. And it really comes from the deep inside. But why should Vas take Bob for serious if it's simply true that there is no Fruit in his code? Should he now explain to this mountain of expertise what he had missed and then falsely accused and insulted? Would Bob really accept that? No, from all what we and Graham know, Bob will never admit of having made a mistake, which is very wrong because we all make mistakes. But not Bob. For me the explanation is simple. But if I stated it Bob would cry wolf again. But we all know the truth. I was shocked when I realised with how little coolness Bob looked upon questions and ideas from other nearby fields of science. Moral: Bob is a highly experienced specialist but he's not something above that, because that would require a certain distance in judging. Computer sciences is a technical discipline but nothing in classical sciences. Or social sciences for all. It's about engineering stuff. Without abstract questions of ethics.

Typical example from such a "specialist" today: dont want to harm business people - ok, Vas was harmed - but I couldnt care less what he thinks - ok, perhaps he thinks that stolen code is cool and therefore he stopped delivering new updates... but damages? who the hell cares - or less...

Sorry I doubt you are the right to understand the offense.

[bob]

I dont buy that. Since that code has been stolen anything added to that would give away too much. Wait for a later version if it is properly safe. Or the online thing. The games must go on! Sorry but you have to suffer because criminals had been involved. In such a case you just cant dream of living on a moon. It's here, it's real life. Period.

Where has Vas ever said the R3 code was stolen?

It was during his stay in Budapest, Wroclaw or Dresden.

Can you point us to it? He has said several times that he does not have it because he moved on with rybka development and did not keep a copy of the R3 source.

They are both lying Harvey and the ugly thing is that they have forgotten their old lies and hence the confusion of Rolf....

All that I would say here is that nobody in their right mind would release a commercial product and not maintain the source that matches that product somewhere. How do you _ever_ find a reported bug if you don't have the code for that specific version to look at.

"What a wicked web we weave, when first we practice to deceive..." - Shakespeare

If one is honest at every turn, one never has to eat his own words... -Hyatt