The rumors about Strelka beeing a Rybka clone is still alive

[MikeB]

... or has at least parts of Rybka code inside.

Try positions with Kings only, you will have the same behaviour in all of this positions:

4K3/8/4k3/8/8/8/8/8 w - - 0 1

Rybka:

213.490<--1:info depth 3 score cp -28 time 1 nodes 114 nps 116736 pv e8f8
213.490<--1:info depth 4 score cp -22 time 1 nodes 335 nps 343040 pv e8f8
213.490<--1:info depth 5 score cp -28 time 1 nodes 619 nps 633856 pv e8f8 e6f6
213.490<--1:info depth 6 score cp -22 time 1 nodes 1331 nps 1362944 pv e8f8 e6f6 f8e8
213.490<--1:info depth 7 score cp -16 time 1 nodes 2411 nps 2468864 pv e8f8 e6f6 f8e8 f6e5
213.490<--1:info depth 8 score cp -10 time 1 nodes 4296 nps 4399104 pv e8f8 e6f6 f8e8 f6e5 e8e7
213.490<--1:info depth 9 score cp -16 time 1 nodes 5804 nps 5943296 pv e8f8 e6f6 f8e8 f6e5 e8e7 e5d5 e7d7
213.490<--1:info depth 10 score cp -10 time 1 nodes 8634 nps 8841216 pv e8f8 e6f6 f8e8 f6e5 e8e7 e5d5 e7d7
213.507<--1:info depth 11 score cp 0 time 1 nodes 15453 nps 15823872 pv e8f8 e6f6 f8e8 f6e5 e8e7 e5d5 e7d7 d5c5 d7e6


Strelka:

226.897<--1:info depth 5 score cp -28 time 0 nodes 229 nps 0 pv e8f8 e6f6 f8e8 f6e5 e8e7
226.897<--1:info depth 6 score cp -22 time 0 nodes 448 nps 0 pv e8f8 e6f6 f8e8 f6e5 e8e7 e5d5
226.897<--1:info depth 7 score cp -28 time 0 nodes 811 nps 0 pv e8f8 e6f6 f8e8 f6e5 e8e7 e5d5 e7d7
226.897<--1:info depth 8 score cp -22 time 0 nodes 1569 nps 0 pv e8f8 e6f6 f8e8 f6e5 e8e7 e5d5 e7d7 d5e5
226.897<--1:info depth 9 score cp -16 time 0 nodes 2893 nps 0 pv e8f8 e6f6 f8e8 f6e5 e8e7 e5d5 e7d7 d5d4 d7e6
226.914<--1:info depth 10 score cp -10 time 0 nodes 4891 nps 0 pv e8f8 e6f6 f8e8 f6e5 e8e7 e5d5 e7d7 d5d4 d7e6 d4e4
226.914<--1:info depth 11 score cp -16 time 0 nodes 6678 nps 0 pv e8f8 e6f6 f8e8 f6e5 e8e7 e5d5 e7d7 d5d4 d7e6 d4e4 e6d6
226.914<--1:info depth 12 score cp -10 time 0 nodes 9715 nps 0 pv e8f8 e6f6 f8e8 f6e5 e8e7 e5d5 e7d7 d5d4 d7e6 d4e4 e6d6 e4d4
226.914<--1:info depth 13 score cp 0 time 0 nodes 18640 nps 0 pv e8f8 e6f6 f8e8 f6e5 e8e7 e5d5 e7f6 d5e4 f6e6 e4d4 e6d6 d4e4 d6e6

substract 2 of Strelka's depth and you will see there is always the same score in King vs. King positions. Note that Strelka gives more info strings as Rybka.

Though I defended Strelka in the past, this looks like a real evidence for me. STrelka is a clone.

OK Chris, you where right

Best,
Alex

good analysis - ypur comments , plus others' comments leads me to conclude that at least some of the codd in Strelka was taken from Rybka, the question is , how did it get there. It's just matter of time before many engines will start beenfitting from at least some Rybka engine code - although I beleive Vas will keep the true Rybka on top for a good number of more years.

[Alexander Schmidt]

Hi Werner,

please try once more, here is my analysis of your position:


30.641<--1:info depth 5 score cp 33 time 0 nodes 260 nps 0 pv f3e3 f1e1 e3d3 e1f2 d3d4
30.704<--1:info depth 6 score cp 27 time 0 nodes 485 nps 0 pv f3e3 f1e1 e3d3 e1f2 d3d4 f2e2
30.766<--1:info depth 7 score cp 21 time 0 nodes 992 nps 0 pv f3e3 f1e1 e3d3 e1f2 d3d4 f2e2 d4e4
30.829<--1:info depth 8 score cp 27 time 0 nodes 2095 nps 0 pv f3e3 f1e1 e3d3 e1f2 d3e4 f2e2 e4d4 e2d2
30.891<--1:info depth 9 score cp 21 time 0 nodes 2771 nps 0 pv f3e3 f1e1 e3d3 e1f2 d3e4 f2e2 e4d4 e2d2 d4e4
30.954<--1:info depth 10 score cp 15 time 0 nodes 4406 nps 0 pv f3e3 f1e1 e3d3 e1f2 d3e4 f2e2 e4d4 e2d2 d4e5 d2d3
31.172<--1:info depth 11 score cp 9 time 0 nodes 8342 nps 0 pv f3e3 f1e1 e3d3 e1f2 d3e4 f2e2 e4d4 e2d2 d4e5 d2d3 e5d5
31.314<--1:info depth 12 score cp 15 time 0 nodes 11765 nps 0 pv f3e3 f1e1 e3d3 e1f2 d3e4 f2e2 e4d4 e2d2 d4e5 d2e3 e5d5 e3f4

221.283<--1:info depth 3 score cp 33 time 1 nodes 165 nps 168960 pv f3e3
221.283<--1:info depth 4 score cp 27 time 1 nodes 356 nps 364544 pv f3e3
221.283<--1:info depth 5 score cp 21 time 1 nodes 826 nps 845824 pv f3e3 f1e1
221.283<--1:info depth 6 score cp 27 time 1 nodes 1729 nps 1770496 pv f3e3 f1e1 e3d3
221.283<--1:info depth 7 score cp 21 time 1 nodes 2376 nps 2433024 pv f3e3 f1e1 e3d3 e1f2
221.283<--1:info depth 8 score cp 15 time 1 nodes 3905 nps 3998720 pv f3e3 f1e1 e3d3 e1f2 d3e4
221.283<--1:info depth 9 score cp 9 time 1 nodes 7447 nps 7625728 pv f3e3 f1e1 e3d3 e1f2 d3e4 f2e2
221.298<--1:info depth 10 score cp 15 time 1 nodes 10910 nps 11171840 pv f3e3 f1e1 e3d3 e1f2 d3e4 f2e2 e4d4 e2d2

Maybe you forgot to restart your engine before analysing a new position (hashtables)

Best,
Alex

[Joachim Wolfram, PhD.]

Is Strelka a free program?

[gerold]

Is Strelka a free program?

Yes it is a free uci engine. Very strong. I have it at just about the
same as Rybka. It plays a lot like Rybka. I can't say its a clone
yet.

Gerold.

[Chris Taylor]

[quote="Joachim Wolfram, PhD."]Is Strelka a free program?[/quote]



http://sdchess.ru/engines/Strelka_18_UCI.rar

Chris

[Werner]

...sorry
it does in my second test

[MartinBryant]

FWIW, after looking at the analyses presented here, elsewhere and of my own, I have to agree that IMO Strelka is a clone. No doubt in my mind about it. In fact, I can't see how people earlier decided it wasn't?!

Of course, what that actually means to computer chess enthusiasts is another thing. Does it really matter? Perhaps only to a few testers who exclude clones from their tourneys. It is what it is.

[ozziejoe]

ok, this thread, and others in rybka forum, have convinced me to stop testing strelka.

An expert looked at the code and said it was not a clone. I accepted this at face value.. However, I am now realizing that the expert did not offer any evidence . He just said confidently it was no clone.


maybe if the expert can weigh in again, and offer evidence that strelka is no rybka, then we can become enthusiastic about this new engine.


Otherwise, """expertize" is not enough reason for for me to believe it.
I am a scientist and in my experience, the experts are wrong all the time
best
Joseph

[ozziejoe]

it does matter martin. We chess engine geeks get really excited about new engines, and praise the intellectual work that an engine entails. We love to see even weak engines improved. we cheer for engines the way some cheer for man. united or arsenal.

and, speaking for myself, I hate to be dupped. and I hate for someone elses hard intellectual work to be stolen.

I know it is all just a silly game and we "should" not take it so seriously...but if you look at most o the things we take as important in life, they are often quite arbitrary and silly.

so I choose to be excited about this silly hobby, and to be offended by fraud.

best
J

[mclane]

if he experts who saw the strelka sources judged about them without ever having seen the rybka sources, it was IMO no judgement that can really prove anything.