Komodo - Rybka in Danger?

[lkaufman]

We have some hints that you are correct about move-ordering. If so, Komodo may turn out to be stronger than we think at serious time limits.

[benstoker]

Is this info about R3 that you provide here and elsewhere at all public knowledge?

I taught Rybka 3 that two rooks on the seventh were worth more than two times one rook there, and it probably made R3 something like one Elo point stronger. We're not yet down to looking for single Elo point improvements in Komodo, so at this point Komodo only gets double credit for two rooks on the seventh, not more.

[lkaufman]

Well, this stuff was made public by the clones, and anyway Vas had asked me to keep such general eval knowledge confidential for a year, and it's now been a year and a half. Still I won't reveal exact values of terms in R3 even if they are more or less public due to the clones (or "derivatives" if you prefer). I doubt that they would be of much use to a non-clone program anyway, as different programs require different values for terms.

[maxchgr]

Larry, I did not realize that you are working with other authors on their engines now. Are you no longer helping Vas with Rybka? Are you also helping on engines other than Komodo now? Please do explain to me why you have made that choice as I am rather curious.

[lkaufman]

1. My actual employment with Rybka ended shortly after Rybka 3, basically because most of my easily-programmable chess knowledge was already in R3 and I found it difficult to improve the program just by tinkering with values, and Vas was just too busy to really work closely with me. I remain part of the team and remain willing to help Rybka in various ways, such as representation in tournaments, opening book related work, or creating alternate versions. I'm not working with anyone else other than Don; our friendship and collaboration in chess software dates back two decades and is the main reason I'm working with him on Komodo.

[benstoker]

That's great that you can make use of the intellectual property of Rybka and share some of the general ideas of the program with the engine community and incorporate some of the unique Rybka ideas into other projects.

Another thing ... you have have indicated that the robbolito code is originated from Rybka. Can we assume you have some personal knowledge of this, given your exposure to the Rybka code? Can you detail the evidence of the claim that it's Rybka-derived? Obviously, people want to know. Otherwise, your claims just create uncertainty and doubt. I would respectfully request that you set out in as much detail as you can the basis of the claim and get us closer to resolving the questions, either way it goes.

1. My actual employment with Rybka ended shortly after Rybka 3, basically because most of my easily-programmable chess knowledge was already in R3 and I found it difficult to improve the program just by tinkering with values, and Vas was just too busy to really work closely with me. I remain part of the team and remain willing to help Rybka in various ways, such as representation in tournaments, opening book related work, or creating alternate versions. I'm not working with anyone else other than Don; our friendship and collaboration in chess software dates back two decades and is the main reason I'm working with him on Komodo.

[slobo]

Well, this stuff was made public by the clones, and anyway Vas had asked me to keep such general eval knowledge confidential for a year, and it's now been a year and a half. Still I won't reveal exact values of terms in R3 even if they are more or less public due to the clones (or "derivatives" if you prefer). I doubt that they would be of much use to a non-clone program anyway, as different programs require different values for terms.

I would like to know something:
1. If only you and Vas had the Rybka 3 code, how did the "cloners" managed to get it ?
2. If Rybka's code is alredy "stealed", why you and Vas don't present evidences that the "cloners" code and the Rybka 3 one is the same?

[Michael Diosi]

Hi,


people from ChessOK will answer you if you ask them, at least they answered my questions (I know I am famous). Maybe one day I will publish what they told me now. The place to look for it is the Arena page.

I also published the info about their findings back then when Strelka/Belka was released and... It was in our forum. I decided not to publish it on the Arena page itself back then. Maybe I should... for those who didn't see it.



Michael
http://www.playwitharena.com
P.S. Many thanks for the hits after the ChessBase news item

[benstoker]

Hi,


people from ChessOK will answer you if you ask them, at least they answered my questions (I know I am famous). Maybe one day I will publish what they told me now. The place to look for it is the Arena page.

I also published the info about their findings back then when Strelka/Belka was released and... It was in our forum. I decided not to publish it on the Arena page itself back then. Maybe I should... for those who didn't see it.



Michael
http://www.playwitharena.com
P.S. Many thanks for the hits after the ChessBase news item

Just provide a link! What hide the ball?!

What's the point of making people go look for something on the "Arene page". Where's that?

[Don]

Well, this stuff was made public by the clones, and anyway Vas had asked me to keep such general eval knowledge confidential for a year, and it's now been a year and a half. Still I won't reveal exact values of terms in R3 even if they are more or less public due to the clones (or "derivatives" if you prefer). I doubt that they would be of much use to a non-clone program anyway, as different programs require different values for terms.

I would like to know something:
1. If only you and Vas had the Rybka 3 code, how did the "cloners" managed to get it ?
2. If Rybka's code is alredy "stealed", why you and Vas don't present evidences that the "cloners" code and the Rybka 3 one is the same?

For what reason should they do this? Every reasonable person already knows that these "clones" are based on Rybka, and whoever is left is not going to be convinced no matter what additional evidence is presented.

I know from my own dealings with people that if someone really want to believe something, no amount of evidence or logic is going to change them.