Not logged inRybka Chess Community Forum
Up Topic The Rybka Lounge / Computer Chess / Problems with the general 50-move rule & UCI 'Syzygy 50mr'
- By MrKris (***) Date 2020-03-12 11:19
The Lc0 and Cute Chess authors seem to underestimate the important differnce between the general 50 move rule and the top engine (except Lc0) setting 'Syzygy 50 move rule (on/off)'.

1) The ICCF rules: https://iccfwebfiles.blob.core.windows.net/rules/2018/ICCF%20Rules%20-%20Final%20version.pdf
" 2. As concerns a 6-piece tablebase claim, if the tablebase shows a win that supersedes the 50 moves rule, the win will be awarded.
In a position that is not solvable by the certified ICCF tablebase, the 50 move rule as described by FIDE Laws of Chess is valid even in case such a solvable position will arise immediately after the 50th move.
In case the solvable won position arises, and is claimed, before draw according to 50 move rule, the win will be awarded. [Reference:  Playing Rule Guidelines]
There is one other exception to the 50-move rule.  If there are a total of 7 pieces remaining on the board, the 50-move rule no longer applies. [Congress 2017 decision] "

2a) Arena's 50-move rule can be turned off.
2b) Arena's drawscore is unchangeable at 0.00 - top engines almost never reach that, leading to many 200+ move games, unless a repition is inimate (sometime not even then) or the last possible set of 50 moves is inimate: the side with a false-advantage will sacrifice pawn(s) just to re-set the 50-move rule.

3) Cute Chess's 50-move rule cannot be turned off. (However its drawscore can be set, which helps a lot in normal circumstances.)

4) Cute Chess will use Syzygy-6 to adjudicate but it is stuck at 'Syzygy 50 move rule ON'.
---This is in violation of ICCF's rule above, and gives meaningless evaluations below.---

5) Lc0's (v0.23.2) has no 'Syzygy 50-move rule' on/off setting like the Stockfish-family, as some other top engines, do.
Lc0 is internally set to 'Syzygy 50 move rule ON' and can not be turned off.

After 42.g8=Q Bxg8 43.Rxg8 , 6-man ending (no pawns)
6R1/8/8/2n4K/8/k1b5/8/3B4 b - - 0 43
Namilov 6-man : https://www.shredderchess.com/online-chess/online-databases/endgame-database.html
43...Kb4 Lose in 78    43...Be5 Lose in 71    43...Ne6 Lose in 64
43...Nd7 Lose in 73    43...Ne4 Lose in 71    43...Bd2 Lose in 63
43...Kb2 Lose in 72    43...Bb4 Lose in 69    ... etc.


Top engines with Syzygy-6 but with 'Syzygy 50 move rule ON' (always ON for Lc0) their evaluations are meaningless to fit with the expectation that the GUI will call the general 50 move rule even if it is won/lost per ICCF.
And, the evaluations are deceptive: black has not played well enough for a += (the -0.19), only that there is only a small chance of being checkmated before the 50 move rule.
They are definitely not the same thing, and there is to indication of which it is. [Event "G/4'+1" 4GBhash Sy-6"]
[Site "2700X | 2060"]
[Date "2020.03.10"]
[Round "4"]
[White "Lc0-.23.2K_Ls13"]
[Black "Bluefish-FD-030720"]
[Result "1/2-1/2"]
[FEN "6R1/8/8/2n4K/8/k1b5/8/3B4 b - - 0 1"]
[PlyCount "100"]
[SetUp "1"]
[TimeControl "240+1"]

1... Kb4 {-0.19/26 5.7s} 2. Rg4+ {+0.41/6 2.8s} Kb5 {-0.19/28 4.4s}
3. Kg6 {+0.55/6 4.3s} Kc6 {-0.19/30 3.8s} 4. Bf3+ {+0.63/7 7.5s}
Kd6 {-0.19/28 3.7s} 5. Kf5 {+0.52/8 5.6s} Nb3 {-0.19/29 4.6s}
6. Bg2 {+0.58/8 3.2s} Nd2 {-0.19/32 3.7s} 7. Rg3 {+0.58/9 5.0s}
Bb4 {-0.19/33 3.3s} 8. Rd3+ {+0.67/10 7.8s} Kc5 {-0.19/30 4.5s}
9. Kf4 {+0.73/10 1.6s} Nb1 {-0.19/29 3.0s} 10. Ke3 {+0.76/9 5.0s}
Nc3 {-0.19/32 4.2s} 11. Rd7 {+0.94/8 6.9s} Ba5 {-0.19/29 3.9s}
12. Kd3 {+1.01/9 5.8s} Nb5 {-0.19/29 6.0s} 13. Rd5+ {+0.97/8 8.8s}
Kb4 {-0.19/29 3.3s} 14. Rh5 {+0.86/8 5.8s} Bd8 {-0.19/30 7.0s}
15. Rh6 {+0.88/7 5.1s} Be7 {-0.19/29 2.8s} 16. Rc6 {+0.83/7 5.1s}
Bc5 {-0.19/32 4.0s} 17. Bh3 {+0.80/8 5.5s} Nd6 {-0.19/35 12s}
18. Be6 {+0.71/9 4.9s} Nb7 {-0.19/30 2.9s} 19. Bd7 {+0.92/11 5.2s}
Bf2 {-0.19/31 5.4s} 20. Rc4+ {+1.03/12 4.1s} Ka5 {-0.19/33 3.2s}
21. Kc3 {+1.13/11 3.5s} Nc5 {-0.19/33 3.9s} 22. Be8 {+1.16/11 5.3s}
Be1+ {-0.19/31 4.9s} 23. Kd4 {+1.26/11 7.3s} Kb6 {-0.19/35 7.6s}
24. Bf7 {+0.69/7 1.2s} Nb7 {-0.19/32 45s} 25. Kd5 {+1.29/8 9.0s}
Nd8 {-0.19/27 2.5s} 26. Bh5 {+1.42/8 7.2s} Bd2 {-0.19/31 19s}
27. Rg4 {+1.35/8 13s} Bc3 {-0.19/31 33s} 28. Rg2 {+1.48/7 4.1s}
Be1 {-0.19/21 25s} 29. Rg6+ {+1.20/7 6.6s} Kc7 {-0.19/25 4.0s}
30. Rg7+ {+1.09/8 4.8s} Kb6 {-0.19/22 5.0s} 31. Rd7 {+0.99/8 5.3s}
Bh4 {-0.19/29 7.8s} 32. Bf3 {+1.10/8 4.5s} Bg5 {-0.19/21 1.5s}
33. Ke5 {+0.86/8 5.9s} Bh4 {-0.19/20 1.1s} 34. Kf5 {+0.81/8 4.3s}
Kb5 {-0.19/22 2.8s} 35. Ba8 {+0.72/7 6.1s} Kc5 {-0.19/22 1.3s}
36. Rd3 {+0.50/7 5.5s} Kc4 {-0.19/26 1.4s} 37. Rd5 {+0.47/8 5.1s}
Nf7 {-0.19/18 1.3s} 38. Kg6 {+0.36/8 6.4s} Nd8 {-0.19/18 1.2s}
39. Re5 {+0.22/7 5.4s} Bf2 {-0.19/22 1.1s} 40. Re2 {+0.16/6 5.2s}
Bg3 {-0.19/20 3.5s} 41. Re8 {+0.15/7 4.9s} Bc7 {-0.19/19 1.6s}
42. Kf6 {+0.14/6 5.3s} Kc5 {-0.19/19 3.4s} 43. Ke7 {+0.17/6 4.8s}
Ba5 {-0.19/21 1.7s} 44. Kd7 {+0.15/7 4.7s} Nf7 {-0.19/19 1.3s}
45. Re7 {+0.19/7 4.5s} Nh8 {-0.19/18 1.1s} 46. Re5+ {+0.19/7 4.7s}
Kb6 {-0.19/17 1.0s} 47. Bc6 {+0.10/6 4.6s} Ng6 {-0.19/18 0.90s}
48. Rb5+ {+0.06/6 4.1s} Ka6 {-0.19/19 1.8s} 49. Kd6 {+0.02/5 7.0s}
Bd2 {-0.19/19 1.1s} 50. Rb3 {0.00/3 2.8s} Ka5 {-0.19/245 0.014s}
51. Kc5 {0.00/2 3.3s, Draw by fifty moves rule} 1/2-1/2

Cute Chess: Syzygy through 6 in 'Tools | Settings | General | Syzgy tablebase path'; and 'Tablebase adjudiction' set to ON in Tournament settings:
- All games from the top position result in a draw: [Event "G/4'+1" 4GBhash Sy-6"]
[Site "2700X | 2060"]
[Date "2020.03.11"]
[Round "1"]
[White "Bluefish-FD-030720"]
[Black "Komodo-13.3_Sy50mrOFF"]
[Result "1/2-1/2"]
[FEN "6R1/8/8/2n4K/8/k1b5/8/3B4 b - - 0 1"]
[PlyCount "1"]
[SetUp "1"]
[Termination "adjudication"]
[TimeControl "240+1"]

1... Kb4 {-280.00/5 0.047s, Draw by adjudication: SyzygyTB ---- -280.01 is Komodo's code for a >50 move Sy. loss, -M93 is Stockfish's; Syzygy does not report the actual # of moves} 1/2-1/2 -- In fact black has made a valid ICCF claim on itself that it is lost! -see above (ICCF all tablebase claims are optional).
-- Similarly: white Komodo Sy50mrOFF - black Bluefish-FD (default 'Sy. 50mr ON') is:
1... Kb4 {-0.19/28 42s, Draw by adjudication: SyzygyTB} 1/2-1/2
Komodo Sy50mrOFF would say +280.01 for itself (see below) making a valid ICCF win claim had it not been stopped by Cute Chess's adjudication that has 'Sy. 50mr ON' only.

Here is what happens WITHOUT Cute Chess's TB adjudication
all games start from the position at the top:

Game/5 mins. + 1 sec.
R7 2700X 16smt | RTX 2060 | Cute Chess 1.0.0
Syzygy-6 (SSD) Sy.50mr on * , 4 GB hash
Drawscore 0.16 move 35 for 2; resign 10.00
*engines defaults except where noted

Cute Chess 50mr is always ON     Score     Bl Lc Ho St Ko
----------------------------------------------------------
1: Bluefish-FD-030720          5.5 /  8   XX == =1 R= =R  (+3 -0 =5)
2: Lc0-.23.2K_Ls13             5.5 /  8   == XX 1= R= R=  (+3 -0 =5)
3: Honey--030720_NO-Sy         3.0 /  8   =0 0= XX 0R R0  (+2 -4 =2)
4: Stockfish_20030922_Sy50off  3.0 /  8   0= 0= 10 XX R0  (+2 -4 =2)
5: Komodo-13.3_Sy50mrOFF       3.0 /  8   =0 0= 01 0R XX  (+2 -4 =2)
----------------------------------------------------------
R = 1 (win) as white vs. black Sy.50-off resign score. see the next game
20 games: white +12 -0 =8

Stockfish and Komodo had Syzygy but with the 'Sy. 50mr OFF' they always reported a very high losing disadvantage with black -
in Cute Chess above any normal to a 'resign score' so Cute Chess immediately ends the game as black lost (the winner's evaluation does not mater, my setting was for 2 moves -10.00.). [Event "G/4'+1" 4GBhash Sy-6"]
[Site "2700X | 2060"]
[Date "2020.03.10"]
[Round "3"]
[White "Lc0-.23.2K_Ls13"]
[Black "Stockfish_20030922_S50off"]
[Result "1-0"]
[FEN "6R1/8/8/2n4K/8/k1b5/8/3B4 b - - 0 1"]
[PlyCount "3"]
[SetUp "1"]
[Termination "adjudication"]
[TimeControl "240+1"]

1... Kb4 {-M93/39 4.6s} 2. Rg4+ {+0.42/6 2.8s}
Kb5 {-M93/36 2.2s, White wins by adjudication} 1-0 But Lc0 will NEVER return the favor - its Sy.50mr can not be turned off! (And as above Cute Chess' for adjudication can not be either.) [Event "G/4'+1" 4GBhash Sy-6"]
[Site "2700X | 2060"]
[Date "2020.03.10"]
[Round "3"]
[White "Stockfish_20030922_S50off"]
[Black "Lc0-.23.2K_Ls13"]
[Result "1/2-1/2"]
[FEN "6R1/8/8/2n4K/8/k1b5/8/3B4 b - - 0 1"]
[PlyCount "100"]
[SetUp "1"]
[TimeControl "240+1"]

1... Kb4 {-0.46/8 2.5s} 2. Rg4+ {+M93/34 4.9s} Kb5 {-0.58/10 3.7s}
3. Kg5 {+M93/34 3.5s} Kc6 {-0.54/9 4.7s} 4. Kf5 {+M93/36 2.5s}
Kd6 {-0.57/10 4.7s} 5. Rg3 {+M93/35 4.2s} Be5 {-0.53/10 8.3s}
6. Rg8 {+M93/35 4.5s} Ba1 {-0.52/9 5.2s} 7. Bf3 {+M93/39 2.7s}
Nb3 {-0.53/11 3.5s} 8. Bg2 {+M93/32 2.4s} Kc5 {-0.72/12 7.2s}
9. Rd8 {+M93/34 2.4s} Nd4+ {-0.78/11 11s} 10. Ke4 {+M93/39 5.7s}
Nb5 {-0.85/11 4.6s} 11. Rd5+ {+M93/40 2.5s} Kb4 {-0.85/10 1.6s}
12. Rh5 {+M93/37 2.4s} Bf6 {-0.84/10 4.8s} 13. Rh6 {+M93/32 3.0s}
Be7 {-0.80/9 5.5s} 14. Bf1 {+M93/41 3.0s} Nc3+ {-0.78/10 3.5s}
15. Ke5 {+M93/34 3.0s} Na4 {-0.70/8 6.5s} 16. Rg6 {+M93/34 3.2s}
Kc3 {-0.69/9 8.0s} 17. Kd5 {+M93/36 2.6s} Kb3 {-0.81/10 5.6s}
18. Bd3 {+M93/42 4.9s} Kb4 {-0.78/10 5.2s} 19. Rg4+ {+M93/44 2.6s}
Kb3 {-0.82/10 5.8s} 20. Bg6 {+M93/37 3.2s} Bf6 {-0.73/11 4.4s}
21. Ke4 {+M93/45 8.2s} Nc5+ {-0.78/11 5.8s} 22. Ke3 {+M93/43 2.9s}
Nd7 {-1.01/11 7.3s} 23. Bf7+ {+M93/44 3.2s} Ka3 {-0.98/11 3.5s}
24. Bg8 {+M93/41 9.2s} Ne5 {-0.98/10 6.4s} 25. Re4 {+M93/46 3.0s}
Bg7 {-0.88/10 5.1s} 26. Bd5 {+M93/47 3.6s} Nd7 {-0.76/11 5.8s}
27. Re7 {+M93/36 3.3s} Nb6 {-0.81/11 3.9s} 28. Bf3 {+M93/40 10s}
Bh8 {-0.80/11 6.4s} 29. Re8 {+M93/38 3.3s} Ba1 {-0.90/11 6.7s}
30. Re4 {+M93/40 46s} Na4 {-0.87/12 3.0s} 31. Kd2 {+M93/35 9.3s}
Nc5 {-0.87/12 4.0s} 32. Rg4 {+M93/33 2.6s} Bh8 {-0.82/10 8.4s}
33. Rg8 {+M93/32 7.7s} Be5 {-0.58/10 6.4s} 34. Re8 {+M93/30 7.6s}
Bd4 {-0.55/11 4.6s} 35. Rd8 {+M93/25 2.5s} Ne6 {-0.66/12 8.3s}
36. Rb8 {+M93/29 6.6s} Be5 {-0.69/15 1.6s} 37. Rb1 {+M93/29 7.8s}
Bb2 {-0.34/8 7.0s} 38. Kd3 {+M93/29 9.4s} Nc5+ {-0.36/9 6.6s}
39. Kc2 {+M93/25 1.9s} Bd4 {-0.33/9 5.6s} 40. Bc6 {+M93/36 4.5s}
Be5 {-0.33/9 10s} 41. Rb5 {+M93/38 1.9s} Bd4 {-0.35/9 2.4s}
42. Kd2 {+M93/34 1.9s} Ne6 {-0.36/9 5.3s} 43. Kd3 {+M93/34 1.9s}
Bg7 {-0.34/8 1.5s} 44. Rb1 {+M93/35 2.0s} Nd4 {-0.23/7 5.6s}
45. Bd5 {+M93/32 1.9s} Bf6 {-0.20/8 10s} 46. Kc4 {+M93/34 1.9s}
Bg7 {-0.17/7 2.6s} 47. Rb7 {+M93/39 1.9s} Bh8 {-0.11/7 3.1s}
48. Bg8 {+M93/46 1.8s} Be5 {-0.08/6 3.5s} 49. Rb6 {+M93/59 1.7s}
Bg7 {0.00/3 6.3s} 50. Rg6 {+M93/72 2.1s} Be5 {0.00/2 2.3s}
51. Rg2 {+M93/245 0.006s, Draw by fifty moves rule} 1/2-1/2

**Honest evaluations with Komodo Sy50mrOFF vs. Honey-NO-Sy, and an honest result.
Honey is looking far enough ahead, even at blitz, to know its lost and reports that: [Event "G/4'+1" 4GBhash Sy-6"]
[Site "2700X | 2060"]
[Date "2020.03.10"]
[Round "1"]
[White "Komodo-13.3_50mrOFF"]
[Black "Honey--030720_NO-Sy"]
[Result "1-0"]
[FEN "6R1/8/8/2n4K/8/k1b5/8/3B4 b - - 0 1"]
[PlyCount "83"]
[SetUp "1"]
[Termination "adjudication"]
[TimeControl "240+1"]

1... Nd7 {-0.47/32 5.6s} 2. Rc8 {+280.01/5 0.016s} Kb4 {-0.48/34 4.1s}
3. Be2 {+280.01/5 0.007s} Ne5 {-0.54/40 4.4s} 4. Kg5 {+280.01/5 0.011s}
Nf7+ {-0.54/42 3.4s} 5. Kf4 {+280.01/31 12s} Be5+ {-0.54/46 2.8s}
6. Kf3 {+280.01/5 0.005s} Nd6 {-0.54/47 2.7s} 7. Rc6 {+280.01/5 0.003s}
Bh2 {-0.54/49 3.3s} 8. Bf1 {+280.01/5 0.010s} Be5 {-0.54/39 4.9s}
9. Ke3 {+280.01/5 0.013s} Kb3 {-0.54/43 3.8s} 10. Rc5 {+280.01/5 0.006s}
Bg3 {-0.54/45 6.9s} 11. Kf3 {+280.01/5 0.003s} Bh2 {-0.54/41 2.7s}
12. Kg2 {+280.01/32 5.0s} Bf4 {-0.54/47 3.1s} 13. Rd5 {+280.01/5 0.009s}
Kc2 {-0.61/47 5.1s} 14. Kf3 {+280.01/5 0.003s} Bh2 {-0.61/47 3.8s}
15. Rh5 {+280.01/5 0.003s} Bg1 {-0.60/44 3.7s} 16. Rg5 {+280.01/5 0.011s}
Bb6 {-0.61/49 3.4s} 17. Rd5 {+280.01/5 0.004s} Ne8 {-0.61/46 3.7s}
18. Rd3 {+280.01/5 0.011s} Bc7 {-0.61/42 3.3s} 19. Ra3 {+280.01/5 0.005s}
Nd6 {-0.61/43 3.2s} 20. Bd3+ {+280.01/34 13s} Kd2 {-0.61/38 3.5s}
21. Bg6 {+280.01/5 0.008s} Nc4 {-0.61/44 3.1s} 22. Rd3+ {+280.01/5 0.002s}
Ke1 {-0.61/42 3.2s} 23. Rc3 {+280.01/5 0.002s} Nd2+ {-0.61/37 4.7s}
24. Kg2 {+280.01/5 0.004s} Ba5 {-0.61/37 3.4s} 25. Rc5 {+280.01/5 0.002s}
Bb4 {-0.61/39 8.3s} 26. Re5+ {+280.01/5 0.002s} Kd1 {-0.59/43 18s}
27. Kf2 {+280.01/5 0.008s} Nc4 {-0.53/39 12s} 28. Rd5+ {+280.01/5 0.016s}
Nd2 {-0.48/35 3.5s} 29. Ke3 {+280.01/5 0.007s} Ke1 {-0.46/31 4.6s}
30. Bh5 {+280.01/5 0.012s} Ba3 {-0.40/33 7.3s} 31. Rd8 {+280.01/33 6.9s}
Bc5+ {-0.43/30 6.6s} 32. Kf4 {+280.01/5 0.002s} Bb4 {-0.42/28 5.1s}
33. Re8+ {+280.01/5 0.009s} Kf2 {-0.38/27 4.6s} 34. Re2+ {+280.01/5 0.005s}
Kf1 {-0.33/27 4.2s} 35. Ke3 {+280.01/5 0.001s} Nc4+ {-0.39/24 5.0s}
36. Kd3 {+280.01/5 0.003s} Nd6 {-0.28/24 5.2s} 37. Bg4 {+280.01/5 0.009s}
Kg1 {-0.28/25 5.5s} 38. Be6 {+280.01/28 14s} Ba3 {-0.21/23 15s}
39. Rc2 {+280.01/24 3.7s} Nb5 {-3.03/36 5.7s} 40. Bh3 {+280.01/28 8.3s}
Be7 {-3.44/32 9.0s} 41. Rg2+ {+280.01/5 0.014s} Kh1 {-35.12/41 19s}
42. Ke2 {+280.01/5 0.004s} Nd6 {-34.14/40 19s, White wins by adjudication} 1-0

*** Stockfish is probably using Syzygy correctly but Cute Chess stops the game as a draw: [Event "G/4'+1" 4GBhash Sy-6"]
[Site "2700X | 2060"]
[Date "2020.03.10"]
[Round "3"]
[White "Stockfish_20030922_S50off"]
[Black "Lc0-.23.2K_Ls13"]
[Result "1/2-1/2"]
[FEN "6R1/8/8/2n4K/8/k1b5/8/3B4 b - - 0 1"]
[PlyCount "100"]
[SetUp "1"]
[TimeControl "240+1"]

1... Kb4 {-0.46/8 2.5s} 2. Rg4+ {+M93/34 4.9s} Kb5 {-0.58/10 3.7s}
3. Kg5 {+M93/34 3.5s} Kc6 {-0.54/9 4.7s} 4. Kf5 {+M93/36 2.5s}
Kd6 {-0.57/10 4.7s} 5. Rg3 {+M93/35 4.2s} Be5 {-0.53/10 8.3s}
6. Rg8 {+M93/35 4.5s} Ba1 {-0.52/9 5.2s} 7. Bf3 {+M93/39 2.7s}
Nb3 {-0.53/11 3.5s} 8. Bg2 {+M93/32 2.4s} Kc5 {-0.72/12 7.2s}
9. Rd8 {+M93/34 2.4s} Nd4+ {-0.78/11 11s} 10. Ke4 {+M93/39 5.7s}
Nb5 {-0.85/11 4.6s} 11. Rd5+ {+M93/40 2.5s} Kb4 {-0.85/10 1.6s}
12. Rh5 {+M93/37 2.4s} Bf6 {-0.84/10 4.8s} 13. Rh6 {+M93/32 3.0s}
Be7 {-0.80/9 5.5s} 14. Bf1 {+M93/41 3.0s} Nc3+ {-0.78/10 3.5s}
15. Ke5 {+M93/34 3.0s} Na4 {-0.70/8 6.5s} 16. Rg6 {+M93/34 3.2s}
Kc3 {-0.69/9 8.0s} 17. Kd5 {+M93/36 2.6s} Kb3 {-0.81/10 5.6s}
18. Bd3 {+M93/42 4.9s} Kb4 {-0.78/10 5.2s} 19. Rg4+ {+M93/44 2.6s}
Kb3 {-0.82/10 5.8s} 20. Bg6 {+M93/37 3.2s} Bf6 {-0.73/11 4.4s}
21. Ke4 {+M93/45 8.2s} Nc5+ {-0.78/11 5.8s} 22. Ke3 {+M93/43 2.9s}
Nd7 {-1.01/11 7.3s} 23. Bf7+ {+M93/44 3.2s} Ka3 {-0.98/11 3.5s}
24. Bg8 {+M93/41 9.2s} Ne5 {-0.98/10 6.4s} 25. Re4 {+M93/46 3.0s}
Bg7 {-0.88/10 5.1s} 26. Bd5 {+M93/47 3.6s} Nd7 {-0.76/11 5.8s}
27. Re7 {+M93/36 3.3s} Nb6 {-0.81/11 3.9s} 28. Bf3 {+M93/40 10s}
Bh8 {-0.80/11 6.4s} 29. Re8 {+M93/38 3.3s} Ba1 {-0.90/11 6.7s}
30. Re4 {+M93/40 46s} Na4 {-0.87/12 3.0s} 31. Kd2 {+M93/35 9.3s}
Nc5 {-0.87/12 4.0s} 32. Rg4 {+M93/33 2.6s} Bh8 {-0.82/10 8.4s}
33. Rg8 {+M93/32 7.7s} Be5 {-0.58/10 6.4s} 34. Re8 {+M93/30 7.6s}
Bd4 {-0.55/11 4.6s} 35. Rd8 {+M93/25 2.5s} Ne6 {-0.66/12 8.3s}
36. Rb8 {+M93/29 6.6s} Be5 {-0.69/15 1.6s} 37. Rb1 {+M93/29 7.8s}
Bb2 {-0.34/8 7.0s} 38. Kd3 {+M93/29 9.4s} Nc5+ {-0.36/9 6.6s}
39. Kc2 {+M93/25 1.9s} Bd4 {-0.33/9 5.6s} 40. Bc6 {+M93/36 4.5s}
Be5 {-0.33/9 10s} 41. Rb5 {+M93/38 1.9s} Bd4 {-0.35/9 2.4s}
42. Kd2 {+M93/34 1.9s} Ne6 {-0.36/9 5.3s} 43. Kd3 {+M93/34 1.9s}
Bg7 {-0.34/8 1.5s} 44. Rb1 {+M93/35 2.0s} Nd4 {-0.23/7 5.6s}
45. Bd5 {+M93/32 1.9s} Bf6 {-0.20/8 10s} 46. Kc4 {+M93/34 1.9s}
Bg7 {-0.17/7 2.6s} 47. Rb7 {+M93/39 1.9s} Bh8 {-0.11/7 3.1s}
48. Bg8 {+M93/46 1.8s} Be5 {-0.08/6 3.5s} 49. Rb6 {+M93/59 1.7s}
Bg7 {0.00/3 6.3s} 50. Rg6 {+M93/72 2.1s} Be5 {0.00/2 2.3s}
51. Rg2 {+M93/245 0.006s, Draw by fifty moves rule} 1/2-1/2

Here is a complete explaination of the scores in the above crosstable:
------
1: Bluefish-FD (+3 -0 =5)
2: Lc0 (+3 -0 =5)
All +3 with white vs.: Honey-NO-Sy, Sf_Sy50off and Komodo-Sy50off;
=5: all 4 games with black draws because Syzygy helps them survive 50 moves, CC always uses the general 50 move draw rule;
and they drew with each other as white.
------
3: Honey_NO-Sy (+2 -4 =2)
+2: It won only as white vs. Sf_Sy50mrOFF & Ko_Sy50mrOFF as explained above;
-4: It always lost as black, all the other engines having Syzygy - it could not defend for the general 50 move rule without Syzygy;
=2: It drew as white, CC general 50 move rule, vs. Bf & Lc0 with black & Syzygy (not losing within CC's general 50 move rule).
------
4: Sf_Sy50mrOFF (+2 -4 =2)
5: Komodo_Sy50mrOFF (+2 -4 =2)
+2: as white they only won vs. the other because their HONEST evalutions with 'Syzygy 50 move rule OFF' caused the other to give a resign score loss.
and Honey-NoSy could not defend for 50 moves vs. either with black **;
-4: they lost all with black because of their HONEST Syzygy evaluation (even if >50 general move rule) interpereted as a resign loss;
=2: although giving a correct winning evaluations vs. the top 2 *** as white Cute Chess's unstopable general 50 move was invoked for draws.
Attachment: Syzygy-50MR_Problems.pgn - Games (31k)
Up Topic The Rybka Lounge / Computer Chess / Problems with the general 50-move rule & UCI 'Syzygy 50mr'

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill