Rybka Coding Posts

[Graham Banks]

I can speak for myself only. I decided to stop testing Strelka because it had been proven by this very society to be a reverse engineered Rybka 1.0b (and even claimed by Vasik Rajlich to be his own engine) and I saw no reason to test 'the same' engine twice, only under a different name.

And, I have to admit, I felt some pressure which led to stop testing it when CCRL was severely criticized (in fact, rudely attacked) for 'supporting' clones.

Now it seems you are criticizing us for doing the opposite.
_________________
Gabor

I think people who donate all that time and effort to do testing can do what they want. You should be dictating your standards for what is necessary to qualify for testing, not the other way around. I would think many other engine writers would agree.

p.s. thank you for sending me the tolearn file from testing. I am back working on my program, and have put the files you sent me to good use .

-Sam

Just sent you my files also Sam. This post reminded me.

Regards, Graham.

[Graham Banks]

I am not criticizing you for stopping tests on Strelka.

On the other hand don't you think it would be good to have established rules in order to avoid any criticism next time it happens?

// Christophe

The only criterion I can think of is whether the engine is legal or not. And how can we tell it is legal when even the experts can't?

Exactly. There are still some "experts" who don't have a problem with Strelka.

[BubbaTough]

The only criterion I can think of is whether the engine is legal or not. And how can we tell it is legal when even the experts can't?
_________________
Gabor

Exactly. There are still some "experts" who don't have a problem with Strelka.

Graham Banks

Legal is a bit of a hard criteria to check...particularly for such an international group. I kind like (my interpretation of) how some tournaments do it. If there is an allegation, they quietly (non-publicly) take some steps, usually involving consulting predetermined expert(s), and then make a decision. The steps are fairly well established and known to all parties ahead of time, including accused and accuser. The accused may be required to hand over source, and if he/she chooses not to they can drop out with no other repercussions.

The tournament director makes the decision, including whether to pursue the issue or ask for code, not the expert(s); the expert(s) are there to help the tournament director reach a conclusion. If someone cries foul too many times, there are usually repercussions for the accuser (such as their program being removed from competition). Both the repercussions and how many accusations is too many is known ahead of time by all. Usually only another participant can make a claim about another program, not disinterested third parties (which is why the cry-wolf clause works). Repercussions for determining a program is invalid can be up to a ban for life, but only for that tournament director's tournaments (and how could they enforce it for other's really). Conclusions may be shared with other tournament directors which may also decide to follow suit.

Overall, tournament rules seem fairly well established, and leave an appropriate amount of decision authority where it belongs (with the person putting the work into organizing the tournament) while making sure they have technical consultants when needed. I don't hear a lot of complaints about whether tournament rules are fair or not. As the prestige of certain testing groups has been catching up with (and passing?) that of tournaments, it might make sense to start establishing some fairly concrete set of guidelines such as exist for tournaments. What those rules should be are really up to the testers, though I am sure the community here would be happy to help draft something if that was of interest. It would probably be nice though if each testing group acted as a cohesive unit, with a single decision maker (or voting system or something) deciding on things so that each individual tester of the group did not have to make their own decisions. That creates a bit of a confusing situation for all.

-Sam

[ozziejoe]

For using the word "clone" you must have skipped almost all of the discussion.



// Christophe[/quote]

Ok, I should say "stolen code"or "copied code."

. Christophe, this post of yours seems just a little, what, mean spirited? It implies i am either too lazy to read , or to ignorant too understand it. Of course what you say is subtly mean, and you will claim that you meant nothing by it and of course you are just providing clarification.

I am not that offended by this rather off hand post of yours, but I am noting a subtle tone that might occur in other posts and might seem inflammatory and unnecessary.

[tiger]

For using the word "clone" you must have skipped almost all of the discussion.



// Christophe

Ok, I should say "stolen code"or "copied code."

. Christophe, this post of yours seems just a little, what, mean spirited? It implies i am either too lazy to read , or to ignorant too understand it. Of course what you say is subtly mean, and you will claim that you meant nothing by it and of course you are just providing clarification.

I am not that offended by this rather off hand post of yours, but I am noting a subtle tone that might occur in other posts and might seem inflammatory and unnecessary.[/quote]



The word clone has been used improperly several times and inevitably starts unnecessary flames.

The discussion is already difficult enough that at least we should avoid misunderstandings in language.



// Christophe

[Dr.Ex]

Rybka has been extremely original right from the begining, in terms of its chess playing. I mean, it is nothing like any other engine.


The newest rybka is positionally brilliant. Play it at bishop or knight odds. If it does not find a tactic to equalize, it will slowly squeeze the life out of your position, till you have no moves left.

Indeed very impressive, I almost had no moves left...

[Event "5'"]
[Site "?"]
[Date "2008.09.02"]
[Round "?"]
[White "Rybka 3 32-bit 2CPU"]
[Black "Michael"]
[Result "0-1"]
[Annotator ""]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/R1BQKBNR w KQkq - 0 1"]
[PlyCount "77"]
[EventDate "2008.??.??"]

{512MB, RybkaII.ctg} 1. Nf3 {7} d5 {2} 2. g3 {4} Bg4 {2} 3. Bg2 {5} Nd7 {1} 4.
O-O {7} e6 {1} 5. d4 {5} Bd6 {1} 6. c4 {8} c6 {0} 7. c5 {4} Bc7 {1} 8. Bg5 {18}
Ne7 {1} 9. h3 {4} Bxf3 {2} 10. Bxf3 {4} O-O {1} 11. Re1 {7} f5 {1} 12. Qb3 {4}
Rb8 {2} 13. Qa3 {5} a6 {1} 14. e3 {7} Nf6 {1} 15. Bxf6 {3} Rxf6 {4} 16. b4 {27}
e5 {6} 17. Qb3 {3} e4 {2} 18. Bg2 {8} g5 {3} 19. Rf1 {8} Kh8 {2} 20. Qd1 {3}
Qe8 {7} 21. f4 {5} gxf4 {3} 22. gxf4 {3} Qg6 {4} 23. Kh1 {3} Rg8 {2} 24. Rg1 {4
} Qh6 {1} 25. Qe2 {4} Rfg6 {1} 26. Raf1 {3} R8g7 {14} 27. Qe1 {2} Ng8 {2} 28.
Qe2 {2} Nf6 {1} 29. Qf2 {20} Rg3 {19} 30. b5 {0} axb5 {2} 31. a4 {15} bxa4 {17}
32. Re1 {0} Qh4 {25} 33. Rgf1 {2} Ng4 {16} 34. Qxg3 {0} Qxg3 {4} 35. hxg4 {0}
Rxg4 {5} 36. Re2 {0} h5 {17} 37. Kg1 {5} h4 {3} 38. Kh1 {13} h3 {3} 39. Bxe4 {0
} 0-1