POLL: Whether Rybka Is An Original Work?

[Charles]

Ippolit is derived from Rybka 3.

[Peter C]

Evidence?

Read the pdf file attached to the first post in this thread:
http://www.open-chess.org/viewtopic.php?f=3&t=119

It's got a lot of good evidence that Ippolit is original (though the authors certainly looked at ideas in Rybka).

Peter

[Charles]

Evidence?

Read the pdf file attached to the first post in this thread:
http://www.open-chess.org/viewtopic.php?f=3&t=119

It's got a lot of good evidence that Ippolit is original (though the authors certainly looked at ideas in Rybka).

Peter

Yes, but did they not get most of their ideas from Rybka?
And how does Ippolit get its strength? Is there something unique within its design that does this or is it simply the ideas taken from Rybka 3?

I think that pdf pretty much makes the case that Ippolit is legal, but there are higher ethical standards needed in determining how original the ideas contained within a chess engine are.

[Matthias Gemuh]

Evidence?

Read the pdf file attached to the first post in this thread:
http://www.open-chess.org/viewtopic.php?f=3&t=119

It's got a lot of good evidence that Ippolit is original (though the authors certainly looked at ideas in Rybka).

Peter

Yes, but did they not get most of their ideas from Rybka?
And how does Ippolit get its strength? Is there something unique within its design that does this or is it simply the ideas taken from Rybka 3?

I think that pdf pretty much makes the case that Ippolit is legal, but there are higher ethical standards needed in determining how original the ideas contained within a chess engine are.

Are you indirectly saying that Ippolit got its ideas from Fruit ?

A nice quote: "I went through the Fruit 2.1 source code forwards and backwards and took many things".

[Charles]

Evidence?

Read the pdf file attached to the first post in this thread:
http://www.open-chess.org/viewtopic.php?f=3&t=119

It's got a lot of good evidence that Ippolit is original (though the authors certainly looked at ideas in Rybka).

Peter

Yes, but did they not get most of their ideas from Rybka?
And how does Ippolit get its strength? Is there something unique within its design that does this or is it simply the ideas taken from Rybka 3?

I think that pdf pretty much makes the case that Ippolit is legal, but there are higher ethical standards needed in determining how original the ideas contained within a chess engine are.

Are you indirectly saying that Ippolit got its ideas from Fruit ?

A nice quote: "I went through the Fruit 2.1 source code forwards and backwards and took many things".

No, I am saying it got its ideas from Rybka 3.
You can argue forever whether Rybka beta took ideas or copied directly from fruit. It is clear to most that if someone takes all ideas from fruit and re-arranges the code they cannot make it stronger.
I am saying that Rybka 3 & 4 have more original ideas than ippolit that's all.

Are you saying Ippolit is an entirely original engine that just appeared to have strength of Rybka 3?

[Peter C]

Evidence?

Read the pdf file attached to the first post in this thread:
http://www.open-chess.org/viewtopic.php?f=3&t=119

It's got a lot of good evidence that Ippolit is original (though the authors certainly looked at ideas in Rybka).

Peter

Yes, but did they not get most of their ideas from Rybka?
And how does Ippolit get its strength? Is there something unique within its design that does this or is it simply the ideas taken from Rybka 3?

I think that pdf pretty much makes the case that Ippolit is legal, but there are higher ethical standards needed in determining how original the ideas contained within a chess engine are.

The Ippolit authors took lots of ideas from Rybka and added some of their own. Like the search is more aggressive than Rybka's and added some new stuff.

Er, most engines take ideas from other engines. Vas even said he got tons of ideas from the Fruit code. The jury is still out on whether he just took ideas or copied code.

Peter

[Roger Brown]

Vas has to account for it not me. However, other programmers have disputed this. My definition of original is the amount of innovation contained in the program. I suspect that Vas studeied fruit and others as he said and built his engine from it. Maybe he copied something in the process. and so Rybka is not original. However, he innovated by making his engine considerably stronger. So I feel the engine has considerable VALUE for the enduser.

Hello Charles,

I see the above as the thin edge of the wedge. You accept innovation where this results in increased strength even where copying took place. I have a problem with that acceptance BUT let's go with it.

What then would be the case against Ippo*?

Isn't it an innovation when Houdini et al (and as yet I am not saying clone, I am saying ideas as the author himself stated on his website) turns out stronger than Rybka?

I prefer to argue the principle of copying first to establish what is right and wrong then treat with each engine based on that principle. Whether the thing is stronger or weaker becomes a slippery slope. What if Rybka was weaker than Fruit? Would the sale of it be any less worse or better (depending on your perspective)?

The amount of originality in rybka makes it a useful product.. Btw.. no one would buy rybka if it was same strength of fruit. ---Rybka's innovation is in its considerable strength.

Again, I am uncomfortable with the strength making it all right argument. Strength is not a justification in and of itself.

This is established with the "clone testing" in talkchess. It can show (not necessarily conclusively) which engines are close in evaluation. Also the latest report by BB shows clearly that Ippolit is not a copy and paste but does show that there are a lot of similarities.
And the innovation is ippolit is strenght in blitz that almost vanishes at long time controls --- I have yet to see enough LTC games showing dominance over Rybka 3!

Should you be referring to the evaluation comparisons I would want to be careful with that analysis. When it was expanded, similarities among other engines - commercial and free - were also noted. That would suggest that either the method lacks the requisite rigour or there is an incestuous arrangement in terms of chess engine coding.

Rybka 4 - adjusting time control factors is much stronger. and is equal or better than ivanhoe. However, houdini might be stronger ...My tests show that.

Are the games or results of the games available somewhere? If not, could you state the time-control you used?

Anyway, if Rybka is NOT original, then neither is ippolit by the same standards.

Ippo* was not sold.

Later.

[kingliveson]

Houdini smashes Rybka tactically in a recent tournament and what happens afterwards is laughable comments by Rybka's (team) henchmen in attempt to discredit the wicked win.

[Event "TCEC - Tournament 5"]
[Site "www.tcec-chess.org"]
[Date "2010.11.26"]
[Round "8.3"]
[White "Houdini 1.03a 64-bit 6CPU"]
[Black "Rybka 4 Exp-61 64-bit 6CPU"]
[Result "1-0"]
[Time "23:32:53"]
[ECO "D15"]
[Opening "Three Knights Variation, Slav; D15"]
[TimeControl "40/7200:20/3600:900+30"]
[PlyCount "163"]
[Number "24"]
[Termination "GUI adjudication"]
[WhiteType "program"]
[BlackType "program"]
[Variant "normal"]

{ Intel Core i7 980x @ 4303 MHz }
1. d4 d5 2. c4 c6 3. Nf3 Nf6 4. Nc3 e6 5. e3 Nbd7 6. Qc2 Bd6 7. Bd3 O-O 8. O-O
dxc4 9. Bxc4 b5 10. Be2 a6 11. Ng5 Qc7 12. e4 Bxh2+ 13. Kh1 h6 14. Nf3 Bf4 15.
e5 Bxc1 16. Raxc1 Ne8 17. Ne4 Bb7 18. Rfd1 Rd8 19. Qc3 Qb8 20. Kg1 Nc7 21. Qe1
Nd5 22. b4 Nf4 23. Bf1 Nb6 24. a3 Nbd5 25. Nfd2 Ne7 26. Nb3 Nf5 27. g3 Nd5 28.
Na5 Ba8 29. Nc5 Qc8 30. Qe4 Rfe8 31. Rd3 Nb6 32. Qf4 Rf8 33. Bh3 Nd5 34. Qe4
Nde7 35. Kh2 Kh8 36. Rh1 Nd5 37. Kg1 Kg8 38. Bf1 Nfe7 39. Be2 f5 40. Qh4 Ng6
41. Qh5 Nh8 42. f4 Kh7 43. Bf3 Ng6 44. Rd2 Rh8 45. Rdh2 Nde7 46. Nab3 Kg8 47.
Rc2 Nf8 48. g4 Nfg6 49. Qh2 Rf8 50. Rf2 Rf7 51. gxf5 Nxf5 52. Bh5 Nge7 53.
Bxf7+ Kxf7 54. Qh5+ g6 55. Qg4 Nd5 56. Nd2 h5 57. Qh3 Qe8 58. Nde4 Qe7 59. Ng5+
Ke8 60. Ngxe6 Bb7 61. Qg2 Rh6 62. Ng5 Bc8 63. Rh3 Kd8 64. e6 Kc7 65. Qe4 Qf6
66. Nf7 Rh7 67. Ne5 Nfe7 68. Ned7 Qf5 69. Qe5+ Qxe5 70. fxe5 Nf5 71. Nf8 Rh6
72. Kh2 Nde7 73. Rhf3 Kd8 74. Ne4 Kc7 75. Rg2 Nd5 76. Rxf5 gxf5 77. Rg7+ Kd8
78. Ng5 Ke8 79. Ng6 Bxe6 80. Nxe6 h4 81. Nc5 f4 82. e6 1-0

The claim is that Houdini is a modified derivative of an older version of Rybka. Funny because there is no publicly available Rybka source code. For the record, Rybka is not an original program. Rybka is A Fruit Derivative.

Rybka on the ranking table before very delicious open-source Peach Fruit code was released:

Cross Table

No  Name                Feder Rtg   1    2    3    4    5    6    7    8    9

1.  Hiarcs,                   2812 28:W 16:W 10:L 20:D 34:W 13:W 15:W  9:W  3:D
2.  Junior,                   2701 29:D  7:W 14:W  5:L 38:W 19:W 16:W  3:D  9:D
3.  Crafty,                   2607 30:W 20:W 12:W 10:D  8:W  9:W 21:D  2:D  1:D
4.  Rebel 12,                 2600 31:W 19:W  9:L 21:L 33:W 25:W  5:L  7:W 40:L
5.  Ruffian,                  2595 32:W 22:D 34:W  2:W  9:L 20:W  4:W 21:W 10:L
6.  Quark,                    2586 33:W 21:D 15:D 19:L 36:W 31:L 26:W 24:W 14:L
7.  Arasan,                   2557 34:L  2:L 48:W 36:L 45:W 39:W 32:W  4:L 31:W
8.  Searcher,                 2532 35:W 23:W 13:W  9:L  3:L 40:D 38:W 10:L 16:W
9.  Yace,                     2531 36:W 26:W  4:W  8:W  5:W  3:L 10:D  1:L  2:D
10. Zappa,                    2530 37:W 27:W  1:W  3:D 21:L 11:W  9:D  8:W  5:W
11. Falcon,                   2500 38:W 34:D 22:D 40:W 19:D 10:L 23:L 28:W 51:L
12. Pepito,                   2500 39:W 45:W  3:L 23:L 37:D 51:L 33:D 35:W 36:W
13. Green Light Chess,        2495 40:W 44:W  8:L 22:W 23:W  1:L 51:D 20:L 26:W
14. Comet B68,                2489 41:W 51:W  2:L 29:D 15:L 37:W 31:D 25:W  6:W
15. King of Kings,            2479 42:W 52:D  6:D 38:D 14:W 22:W  1:L 51:W 23:W
16. Post Modernist,           2471 43:W  1:L 33:W 51:W 31:W 21:D  2:L 40:D  8:L
17. Chezzz,                   2468 44:L 40:L 54:W 42:L 46:W 38:L 45:W 34:D 28:W
18. Ikarus,                   2466 45:L 29:D 43:W 39:D 40:L 36:L 41:L   :    :
19. The Baron,                2465 46:W  4:L 32:W  6:W 11:D  2:L 40:L 27:W 29:L
20. Pharaon,                  2450 47:W  3:L 37:W  1:D 29:W  5:L 42:W 13:W 21:D
21. Thinker 4.5a,             2450 48:W  6:D 52:W  4:W 10:W 16:D  3:D  5:L 20:D
22. WildCat,                  2425 49:W  5:D 11:D 13:L 39:W 15:L 27:D 29:L 37:W
23. XiniX,                    2420 50:W  8:L 36:W 12:W 13:L 42:D 11:W 31:W 15:L
24. OliThink,                 2400 51:L 39:L 47:W 35:W 42:L 41:W 36:W  6:L 33:L
25. BlackBishop,              2400 52:L 42:W 38:L 44:W 50:W  4:L 34:W 14:L 32:W
26. SpiderChess,              2396 53:W  9:L 39:D 31:L 52:W 29:W  6:L 38:W 13:L
27. Frenzee,                  2380 54:W 10:L 40:L 45:D 28:W 34:D 22:D 19:L 38:L
28. Djinn,                    2378  1:L 41:D 51:L 47:W 27:L 44:W 37:W 11:L 17:L
29. Movei,                    2367  2:D 18:D 44:W 14:D 20:L 26:L 50:W 22:W 19:W
30. messchess,                2367  3:L 43:D 35:D 37:L 41:L 52:W 47:W 36:L 46:L
31. Amateur,                  2361  4:L 46:W 45:W 26:W 16:L  6:W 14:D 23:L  7:L
32. Averno,                   2354  5:L 47:W 19:L 41:W 51:L 49:W  7:L 42:W 25:L
33. Bodo,                     2351  6:L 48:W 16:L 46:W  4:L 50:D 12:D 39:W 24:W
34. Butcher,                  2337  7:W 11:D  5:L 52:W  1:L 27:D 25:L 17:D 42:L
35. Dorky 4.0,                2324  8:L 50:D 30:D 24:L 44:D 45:D 46:W 12:L 43:W
36. Chepla,                   2323  9:L 49:W 23:L  7:W  6:L 18:W 24:L 30:W 12:L
37. Hossa,                    2317 10:L 54:W 20:L 30:W 12:D 14:L 28:L 50:W 22:L
38. Amyan,                    2292 11:L 53:W 25:W 15:D  2:L 17:W  8:L 26:L 27:W
39. Alarm,                    2290 12:L 24:W 26:D 18:D 22:L  7:L 49:W 33:L 52:L
40. Bringer 1.9,              2283 13:L 17:W 27:W 11:L 18:W  8:D 19:W 16:D  4:W
41. Chompster,                2274 14:L 28:D 50:D 32:L 30:W 24:L 18:L   :    :
42. Tao 5.6,                  2274 15:L 25:L 49:W 17:W 24:W 23:D 20:L 32:L 34:W
43. Tinker,                   2232 16:L 30:D 18:L 50:L 53:D 46:L 54:W 47:W 35:L
44. Rascal,                   2225 17:W 13:L 29:L 25:L 35:D 28:L 53:W 52:D 49:W
45. Nullmover,                2213 18:W 12:L 31:L 27:D  7:L 35:D 17:L 49:L 53:D
46. Chiron,                   2200 19:L 31:L 53:W 33:L 17:L 43:W 35:L 48:W 30:W
47. SEE,                      2196 20:L 32:L 24:L 28:L 54:W 53:W 30:L 43:L 48:D
48. Noonian Chess,            2128 21:L 33:L  7:L 53:D 49:L 54:W 52:L 46:L 47:D
49. 31337/Celes,              2127 22:L 36:L 42:L 54:W 48:W 32:L 39:L 45:W 44:L
50. Cheetah,                  2104 23:L 35:D 41:D 43:W 25:L 33:D 29:L 37:L 54:W
51. Jonny 2.54,               2075 24:W 14:L 28:W 16:L 32:W 12:W 13:D 15:L 11:W
52. MatadorX,                 2000 25:W 15:D 21:L 34:L 26:L 30:L 48:W 44:D 39:W
53. Rybka,                    2000 26:L 38:L 46:L 48:D 43:D 47:L 44:L 54:W 45:D
54. Tohno,                    1800 27:L 37:L 17:L 49:L 47:L 48:L 43:L 53:L 50:L

And the Original Cloner Award goes to...

The exact numerology is much beyond a similarity (as I mentioned, merely having the same Rank/File/Line centralisation strategy for PST is a different bailiwick than having the same numbers from them). I can't find any engines other than Fruit and Rybka whose PST values are derived from (minor exceptions with Rybka in central pawns):

Code: Select all

    static const int PawnFile[8] = {-3, -1, +0, +1, +1, +0, -1, -3,};
    static const int KnightLine[8] = {-4, -2, +0, +1, +1, +0, -2, -4,};
    static const int KnightRank[8] = {-2, -1, +0, +1, +2, +3, +2, +1,};
    static const int BishopLine[8] = {-3, -1, +0, +1, +1, +0, -1, -3,};
    static const int RookFile[8] = {-2, -1, +0, +1, +1, +0, -1, -2,};
    static const int QueenLine[8] = { -3, -1, +0, +1, +1, +0, -1, -3,};
    static const int KingLine[8] = {-3, -1, +0, +1, +1, +0, -1, -3,};
    static const int KingFile[8] = { +3, +4, +2, +0, +0, +2, +4, +3,};
    static const int KingRank[8] = { +1, +0, -2, -3, -4, -5, -6, -7,};

I can imagine someone with the same idea producing 4 or 5 of these arrays that are the same, but not all 9.

Rybka's evaluation has been the subject of much speculation ever since its appearance. Various theories have been put forth about the inner workings of the evaluation, but with the publication of Strelka, it was shown just how wrong everyone was. It is perhaps ironic that Rybka's evaluation is its most similar part to Fruit; it contains, in my opinion, the most damning evidence of all. Simply put, Rybka's evaluation is virtually identical to Fruit's.

You can see the original report showing Rybka is modified Fruit here or get the attached pdf file.

Taking an open-source GPL program, modifying it, and releasing binaries for sale is fine -- requires however written offer to provide source code that must accompany binary-only release; not releasing the source-code violates GPL terms of agreement. Even worse of a violation is a violator accusing another without offering a single hair strand of proof.

Yes, the publication of Fruit 2.1 was huge. Look at how many engines took a massive jump in its wake: Rybka, Hiarcs, Fritz, Zappa, Spike, List, and so on. I went through the Fruit 2.1 source code forwards and backwards and took many things.

[thorstenczub]

do not download this pdf-file.
my avast claims a virus.

[BB+]

As appears in the TalkChess thread, the lousiness of the a8 bishop plays a notable rôle in the above game. I might point out (again) that bad bishops are one of the major differences in method (beyond just parameter tweaking) of evaluation between Rybka 3 and the IPPOLIT series. [I am hesitant to publish the R3 details (and it would take a bit of work to do so completely), but perhaps I will do so if this continues to be an issue].

However I do not think IPPOLIT/IvanHoe has any particular decrement for an a8 bishop blocked by its own pawns [there is the "bishop on back rank penalty" but nothing else], and so Houdini seems to have added an additional evaluation element here [I'm not quite sure it is as important as 40cp -- usually a bishop on a8 can move to b7 where presumably it is would "just" be a bad bishop, but here that is not possible at various junctures due to the White knights].

PS. I didn't Letouzey know was involved in Othello, and was also surprised the 8x8 game is "drawish" (though independent confirmation agrees --- 32-32 is the expected result, though my sources indicate the first player has the chances). And the PDF download was fine for me (not that it did anything besides collate ZW's web pages).