Moderator: Ras
CCRL is a group of free thinking individuals with diverse ideas and opinions. We make decisions through discussion and voting when compromise can't be found (did not happend yet). If you have any specific questions to me, I'll be glad to answer.Christopher Conkie wrote:CCRL is run by a Russian.
It is not hard to understand why he likes it either. He is wrong of course but at least he made his own mind up.
Thanks Graham. This is a good summary of my idea about Strelka. It was directed to another person originally, but I posted it at our forum as well to share with others. I am happy for Graham to quote it here.Graham Banks wrote:Chris - I can't read Russian. What I'd like to see is a reliable accurate translation that proves that Strelka is an illegally cloned engine.Christopher Conkie wrote:http://kasparovchess.crestbook.com/view ... =1869&p=17Graham Banks wrote:Show me a reliable translation of the Russian message where the Strelka author admits beyond all doubt that he has illegally cloned his engine.Christopher Conkie wrote: Do you think Strelka is a clone Graham? Don't you believe the guy who made it when he told you it was one?
Be serious.
Graham.
Message #844...Point One....
Show me where the GNU license is distributed with Strelka.
And the source code.....? Where's that? And that's just for starters. we have not even got to the Rybka stuff yet.
I'm sure Kirill won't mind me sharing his views on Strelka:
Regards, Graham.There is no doubt in my mind that Strelka is a clone. The ponder hit statistics in our 40/4 list shows that Strelka is more close to Rybka 1.0 than some Rybka versions to each other. However there is no enough evidence that this was done illegally. We have to be very clear about it, otherwise any person with doubts will be able to remove an engine from our list.
What we know from Osipov (assuming the guy who posted that is really Strelka author, which no one can guarantee either) is:
1. He took Fruit as a basis for Strelka, and re-wrote it to bitboards. Re-writing something is not re-using the code verbatim, so I am not sure GPL is violated. It looks like he reused algorithms. Also those who examined Strelka code did not confirm yet that GPL was violated.
2. He said that he did some statistical tuning of Strelka eval trying to approximate that of Rybka. This is totally acceptable way of making an engine, even if I myself don't see much point in doing that.
3. He said that he took some tables from Rybka. This may be problematic, but so far there is no enough information about it. Vasik keeps silence too.
I also remember the argument about same UCI output in Strelka and Rybka. But Rybka source is closed, so we have to assume that Osipov simply imitated Rybka's interface. This itself is not illegal.
Note that I myself dislike Strelka because even if cloning was legal, the author did not acknowledge his use of Fruit and Rybka.
Hot discussions that appear around Strelka every time show that situation is far from reaching any conclusion. I don't agree with the logic to remove Strelka when there are reasonable doubts. Because the data we provide, like correlation between Strelka and Rybka, may be very useful in resolving the issue. Our "ponder hit" stats show that Strelka plays nearly identical to Rybka 1.0. It is interesting to see at least. Without this data we will only have individual positions posted in the forums (positions evaluated the same by both engines), but someone will also say that there are also positions they evaluate differently. But our ponder hit tells us in more precise way how similar they play to each other.
Please note that we also discuss it a lot within the group. We have to think not only about Strelka, but about setting a precedent for future cases. So we really have to depend on some more sound reasoning than that presented so far.
The same stats also show that Loop plays nearly identical to the various Toga versions and settings. Comparison.Our "ponder hit" stats show that Strelka plays nearly identical to Rybka 1.0.
It's the percentage of ponder hits (i.e. predicted moves).Is there a direct interpretation of the ponder hit numbers (e.g., does 69 mean 69% of the time they pondered the same move)
Perhaps the poster on that forum of this topic, is not the author of Strelka and another one that simply uses his name.Christopher Conkie wrote: The word Fruit is readable but since it's you...here is message 844 point one in English.....
1. The basis of the program started with the sourcecode of Fruit a little less than two years ago.
It's that simple Graham. and we still have not got to the Rybka stuff.
Also remember than when first released he had been working on it for 20 years.
Do try to keep up.......
Perhaps it is simply materialism, when people think they own something new. Like advertising. Desperation!Daniel Mehrmann wrote:Wow, i thought here are a few nuts, but 45% would use Strelka in tournaments.....
Bye Bye, time to leave clone city
Daniel
I guess you like testing clones if they are strong and dont test weak original engines.....Chris Taylor wrote:As a member of the CCRL I must protest at your wild generalization. I also wonder why you made this attack. This is quite provocative! Methinks you have missed the target by quite a measure!
In CCRL we are encouraged to select those engines we want! Not what any one member says we must. Speaking for myself, I am happy to do some games, without harassment or any other pressure! It is guided by a sense of fun, a sense of enjoyment......
That Russian is a decent guy, and He has been helpful in my testing. Never once has he told me what to test. I can only remember good advice being passed to me to solve problems.
I admire and applaud his hard work and dedication towards the CCRL. His guidance in matters of computers and the playing of chess will have a place in my life...
Guess I enjoy CC and the CCRL...
Chris Taylor
Nottingham
England
It is him George.George Tsavdaris wrote:Perhaps the poster on that forum of this topic, is not the author of Strelka and another one that simply uses his name.Christopher Conkie wrote: The word Fruit is readable but since it's you...here is message 844 point one in English.....
1. The basis of the program started with the sourcecode of Fruit a little less than two years ago.
It's that simple Graham. and we still have not got to the Rybka stuff.
Also remember than when first released he had been working on it for 20 years.
Do try to keep up.......
I don't believe that, but we can't use the words someone said in one forum as a proof Strelka is a clone.
Of course i can't read Russian and perhaps there is other evidence in that forum that the author of this topic was the author of Strelka too but i can't know.