Latest TCEC has 379 moves draw

[Frank Quisinsky]

Connor,

ah... very rude of me.

First of all, thanks for your new Seer.
The move average of the Seer is absolutely OK.

Nice-to-have and for me always a very nice engine development.
Also in the group of hard mid-game fighters and not easy to beat with a fast king attack.

And that's the point!
As long as all engines don't produce the same style of chess I like to play my tournaments.

Best
Frank

[Frank Quisinsky]

Hi Gabor,

I do that in the past with one of my older rating systems.
I stoped the games after 160 moves with a bad conscience.

Not sure, I believe it was Dieter Buerssner, where I had a discussion about it. For programmers much more interesting if the game run to the end. So engine mistakes can be found in endgames and can be fix. At this time I produced the ratinglist with a fast Dual Core Pentium I with 1Ghz, very expensive and not many people have such a system. On this system Deep Shredder vs. the World are still running. Games are with longer time controls and some programmers are happy for looking in the games.

Often I am thinking on it.
But maybe today all this isn't longer important?

I believe around 30.000 games played on this system at this time.
43 games goes over 160 moves ...

And today ...
over 500 from 10.000 games over 180 moves!

That is a different!

Best
Frank

Move-average without resign mode = 69 moves (from TOP-20) around the year 2000.
Today 95 moves.

[connor_mcmonigle]

...
A GUI should never affect a game between two engines. Not with endgame tables and not with other things. That is the job of the engine programmers. I have never changed my opinion!
...

Apologies if my message game across as a tad aggressive. I think this is a reasonable opinion, but it does imply that you'll suffer through some long endgames, especially given your choice of a cyclical time control for your tournament. Have you considered switching to incremental time controls for your tournaments? Doing so should significantly reduce the compute time wasted on long drawn games. Additionally, most engine authors, especially younger engine authors, focus almost exclusively on incremental time controls for testing/tuning their engines' respective time managers. Consequently, you'll not uncommonly see engines fail to perform at their best when using cyclical time controls.

[RubiChess]

My guess is that in general long games are not caused by what you call contempt (most of the engine probably don't have implemented anything like that) but by wrong/bad eval of one of the engines seeing some advantage (material or advanced pawn that leads to nothing or whatever) that is none.

And if SF games/draws are longer in average than probably because its evaluation is superior to most/all other engines and it saves many games into drawn positions that the opponent doesn't accept to be drawn.

Sorry, this is nonsense in most cases.

Took your pgn, ordered the games by plies and took the longest SF game:
[pgn][Event "40 Moves in 20 min"] [Site "fcp-tourney-2024, WASP-3"] [Date "2023.11.03"] [Round "1.22"] [White "Stockfish 16 NN"] [Black "Booot 7.2 NN"] [Result "1/2-1/2"] [EventDate "????.??.??"] [ECO "B33"] [PlyCount "517"] 1.e4 {book 0s} 1...c5 {book 0s} 2.Nf3 {book 0s} 2...Nc6 {book 0s} 3.d4 {book 0s} 3...cxd4 {book 0s} 4.Nxd4 {book 0s} 4...Nf6 {book 0s} 5.Nc3 {book 0s} 5...e5 {book 0s} 6.Ndb5 {book 0s} 6...d6 {book 0s} 7.Bg5 {book 0s} 7...a6 {book 0s} 8.Na3 {book 0s} 8...b5 {book 0s} 9.Bxf6 {book 0s} 9...gxf6 {book 0s} 10.Nd5 {book 0s} 10...Bg7 {book 0s} 11.c3 {book 0s} 11...f5 {book 0s} 12.exf5 {+0.02/44 2:55m} 12...Bxf5 {+0.32/32 1:05m} 13.Nc2 {+0.03/40 17s} 13...O-O {+0.27/33 51s} 14.a4 {0.00/41 20s} 14...Be6 {+0.01/33 53s} 15.g3 {0.00/42 19s} 15...bxa4 {+0.50/35 2:18m} 16.Rxa4 {0.00/47 51s} 16...Rb8 {+0.09/36 40s} 17.Bxa6 {0.00/44 14s} 17...Rxb2 {+0.01/38 53s} 18.Nce3 {0.00/45 25s} 18...f5 {+0.11/38 56s} 19.O-O {0.00/45 22s} 19...f4 {0.00/38 47s} 20.gxf4 {0.00/50 22s} 20...Nd4 {+0.05/38 46s} 21.Qh5 {0.00/49 26s} 21...exf4 {+0.01/38 44s} 22.Bd3 {0.00/51 25s} 22...Qg5+ {0.00/39 39s} 23.Qxg5 {0.00/51 27s} 23...Nf3+ {+0.01/39 38s} 24.Kh1 {0.00/55 28s} 24...Nxg5 {+0.07/40 46s} 25.Rxf4 {0.00/57 27s} 25...Bxd5+ {+0.07/39 42s} 26.Nxd5 {0.00/56 33s} 26...Rd2 {+0.07/40 33s} 27.Rxf8+ {0.00/59 30s} 27...Kxf8 {+0.01/38 26s} 28.f4 {0.00/63 33s} 28...Nh3 {+0.01/40 22s (Rxd3)} 29.Bc4 {0.00/60 33s} 29...Nf2+ {+0.01/39 29s} 30.Kg2 {0.00/66 32s} 30...Ne4+ {+0.01/38 28s} 31.Kf3 {0.00/67 40s} 31...Nxc3 {0.00/39 29s} 32.Nxc3 {0.00/65 36s} 32...Bxc3 {0.00/41 29s} 33.Be2 {0.00/70 38s} 33...Bb4 {0.00/40 29s (d5)} 34.Rc1 {0.00/73 39s} 34...Rd4 {0.00/41 24s} 35.Rc7 {0.00/76 41s} 35...Bd2 {0.00/43 25s (d5)} 36.Rc4 {0.00/88 49s} 36...Rxc4 {0.00/50 24s} 37.Bxc4 {0.00/96 50s} 37...Be1 {0.00/55 32s (d5)} 38.Bd5 {0.00/77 55s} 38...Bd2 {0.00/56 33s (Kg7)} 39.f5 {0.00/82 1:04m} 39...Bc3 {-0.01/51 32s (Ke7)} 40.Ke3 {0.00/79 46s} 40...Kg7 {-0.01/51 16s} 41.h3 {0.00/74 23s} 41...Ba1 {-0.01/56 35s (Bb4)} 42.f6+ {0.00/76 24s} 42...Kxf6 {-0.07/51 32s} 43.h4 {0.00/75 24s} 43...Ke5 {-0.07/54 34s (Bc3)} 44.Be4 {0.00/84 24s} 44...h6 {-0.07/57 39s (d5)} 45.Kd3 {0.00/77 25s} 45...Bc3 {-0.07/56 34s} 46.h5 {0.00/76 25s} 46...Ba1 {-0.07/60 34s (d5)} 47.Bg2 {0.00/74 25s} 47...Bb2 {-0.07/63 39s (d5)} 48.Be4 {0.00/74 27s} 48...Ba1 {-0.07/64 41s (d5)} 49.Bh1 {0.00/74 26s} 49...Bb2 {-0.07/64 31s (d5)} 50.Bg2 {0.00/76 27s} 50...Ba1 {-0.07/64 32s (d5)} 51.Bh3 {0.00/73 28s} 51...Bb2 {-0.07/65 35s (Kf6)} 52.Kc2 {0.00/74 37s} 52...Ba3 {-0.07/63 32s (Ba1)} 53.Kb1 {0.00/70 27s} 53...Bb4 {-0.07/52 30s (d5)} 54.Ka2 {0.00/77 29s} 54...Ba5 {-0.07/52 31s (Bc3)} 55.Kb3 {0.00/74 29s} 55...Be1 {-0.07/61 30s (d5)} 56.Kc4 {0.00/73 29s} 56...Bd2 {-0.07/63 33s (Bf2)} 57.Bf1 {0.00/71 31s} 57...d5+ {-0.07/64 33s (Be1)} 58.Kc5 {0.00/74 31s} 58...Bg5 {-0.19/57 30s (Be1)} 59.Kc6 {0.00/75 37s} 59...d4 {-0.19/61 29s (Be7)} 60.Kc5 {0.00/72 31s} 60...Bf6 {-0.19/64 30s (d3)} 61.Kc4 {0.00/71 32s} 61...Ke4 {-0.19/64 21s (Kd6)} 62.Be2 {0.00/71 32s} 62...Bh4 {-0.09/60 28s (d3)} 63.Bd3+ {0.00/71 33s} 63...Ke3 {-0.09/59 27s (Ke5)} 64.Bc2 {0.00/70 32s} 64...Be1 {-0.07/61 27s (Kf4)} 65.Bb3 {0.00/73 33s} 65...d3 {-0.62/59 22s} 66.Bd1 {0.00/70 33s} 66...Bd2 {-0.62/60 22s (d2)} 67.Ba4 {0.00/76 33s} 67...Bc1 {-0.62/61 20s (Be1)} 68.Bd7 {0.00/70 38s} 68...d2 {-0.62/63 23s} 69.Bg4 {0.00/74 34s} 69...Kf2 {-0.62/65 21s (Kf4)} 70.Bd1 {0.00/81 34s} 70...Ke1 {-9.95/54 22s} 71.Bg4 {0.00/84 1:50m} 71...d1=Q {-9.95/61 22s} 72.Bxd1 {0.00/92 46s} 72...Kxd1 {-9.95/65 27s} 73.Kd3 {0.00/58 2:10m} 73...Bd2 {-9.95/62 23s (Ke1)} 74.Ke4 {0.00/74 11s} 74...Ke2 {-9.95/64 25s} 75.Kd4 {0.00/73 11s} 75...Kf3 {-9.95/62 32s (Kf2)} 76.Kd3 {0.00/74 17s} 76...Bg5 {-9.95/61 34s} 77.Kd4 {0.00/70 30s} 77...Ke2 {-9.95/65 34s (Kg4)} 78.Kc4 {0.00/63 4s} 78...Bf6 {-9.95/60 34s (Kf2)} 79.Kb3 {0.00/44 1s} 79...Ke3 {-9.95/61 34s (Kf3)} 80.Kc2 {0.00/40 1s} 80...Ke2 {-9.95/54 17s (Kf3)} 81.Kc1 {0.00/43 32s} 81...Bg5+ {-9.95/60 45s (Kf3)} 82.Kc2 {0.00/50 2:31m} 82...Be3 {-9.95/61 33s (Bf6)} 83.Kb3 {0.00/56 29s} 83...Bg5 {-9.95/61 45s (Kf3)} 84.Kc2 {0.00/53 5:41m} 84...Bf4 {-9.95/60 41s (Bf6)} 85.Kb1 {0.00/53 28s} 85...Kd3 {-9.95/60 33s (Kf3)} 86.Kb2 {0.00/54 31s} 86...Be5+ {-9.95/59 43s (Ke2)} 87.Kb3 {0.00/58 1:29m} 87...Bc3 {-9.95/64 43s (Ke4)} 88.Ka2 {0.00/51 3:01m} 88...Bf6 {-9.95/63 31s (Be5)} 89.Kb1 {0.00/42 17s} 89...Kd2 {-9.95/61 42s} 90.Ka2 {0.00/48 1:51m} 90...Kc3 {-9.95/62 41s (Bd4)} 91.Ka1 {0.00/45 8s} 91...Kc2+ {-9.95/53 40s (Kd2+)} 92.Ka2 {0.00/35 0s} 92...Be7 {-9.95/59 40s (Kd2)} 93.Ka1 {0.00/41 0s} 93...Ba3 {-9.95/53 39s (Bf6+)} 94.Ka2 {0.00/36 0s} 94...Bc5 {-9.95/55 38s (Be7)} 95.Ka1 {0.00/44 3s} 95...Bd4+ {-9.95/52 38s (Bb4)} 96.Ka2 {0.00/43 1s} 96...Be5 {-9.95/52 26s (Kd2)} 97.Ka3 {0.00/40 0s} 97...Kc3 {-9.95/51 37s (Bd4)} 98.Ka2 {0.00/47 1:14m} 98...Bg3 {-9.95/47 36s (Kd2)} 99.Ka3 {0.00/47 42s} 99...Bh4 {-9.95/47 35s (Kc2)} 100.Ka2 {0.00/45 23s} 100...Bf6 {-9.94/43 2:01m} 101.Kb1 {0.00/52 12s} 101...Kd2 {-9.94/36 18s} 102.Ka2 {0.00/45 6s} 102...Kd3 {-9.94/32 18s (Ke3)} 103.Kb1 {0.00/46 0s} 103...Be5 {-9.94/34 55s (Ke3)} 104.Kc1 {0.00/45 3s} 104...Bd6 {-9.94/37 24s (Ke3)} 105.Kd1 {0.00/44 0s} 105...Bb4 {-9.95/33 13s (Be5)} 106.Kc1 {0.00/36 1s} 106...Bd6 {-9.61/33 32s} 107.Kb1 {0.00/45 0s} 107...Be5 {-9.51/30 12s (Ke3)} 108.Kc1 {0.00/46 0s} 108...Ke2 {-9.86/29 46s (Ke3)} 109.Kc2 {0.00/37 0s} 109...Bc7 {-9.94/28 35s (Ke3)} 110.Kc3 {0.00/38 0s} 110...Bb6 {-8.35/27 8s (Be5+)} 111.Kc2 {0.00/30 0s} 111...Ba5 {-8.20/24 12s (Bc7)} 112.Kb1 {0.00/31 0s} 112...Kd1 {-9.94/25 20s (Bb6)} 113.Kb2 {0.00/31 0s} 113...Bb4 {-9.54/23 15s (Bb6)} 114.Kb3 {0.00/24 0s} 114...Be7 {-6.70/23 11s (Be1)} 115.Kc4 {0.00/21 0s} 115...Ke2 {-6.57/21 8s} 116.Kd4 {0.00/25 0s} 116...Kf3 {-7.00/21 6s} 117.Kc3 {0.00/25 0s} 117...Kf4 {-6.23/18 4s (Kg4)} 118.Kd2 {0.00/26 0s} 118...Kf3 {-0.01/24 4s (Kg5)} 119.Ke1 {0.00/31 0s} 119...Bd6 {-2.40/44 4s (Bc5)} 120.Kf1 {0.00/25 0s} 120...Kg4 {-2.40/44 2s (Bc5)} 121.Kg2 {0.00/37 3:20m} 121...Ba3 {-2.40/50 45s (Kxh5)} 122.Kf1 {0.00/35 59s} 122...Kxh5 {-2.40/24 44s} 123.Ke2 {0.00/45 6:56m} 123...Kg4 {-2.40/35 44s (Kh4)} 124.Kf1 {0.00/33 31s} 124...Kg3 {-2.40/38 43s} 125.Kg1 {0.00/32 3:43m} 125...Bc1 {-2.40/34 34s (Bd6)} 126.Kh1 {0.00/30 18s} 126...Kf2 {-2.40/32 42s (Bf4)} 127.Kh2 {0.00/26 0s} 127...Bb2 {-2.40/28 31s (Bf4+)} 128.Kh1 {0.00/38 11s} 128...Kg3 {-2.40/41 42s (Bd4)} 129.Kg1 {0.00/25 0s} 129...Ba1 {-2.40/31 41s (Be5)} 130.Kh1 {0.00/27 11s} 130...Bg7 {book 40s (Be5)} 131.Kg1 {0.00/25 0s} 131...Bd4+ {-2.11/31 37s (Bc3)} 132.Kh1 {0.00/30 23s} 132...Kf4 {-2.11/37 39s (Be5)} 133.Kg2 {0.00/42 1:43m} 133...Ke4 {-2.36/56 27s (Ke3)} 134.Kg3 {0.00/40 50s} 134...Ke3 {-2.36/46 38s (Be5+)} 135.Kg2 {0.00/36 3s} 135...Bb2 {-2.36/47 37s (Be5)} 136.Kg1 {0.00/38 23s} 136...Bc1 {-2.36/46 36s (Kf3)} 137.Kh1 {0.00/36 11s} 137...Kf2 {-2.36/33 35s (Kf3)} 138.Kh2 {0.00/25 1s} 138...Bb2 {-2.36/32 35s (Bf4+)} 139.Kh1 {0.00/31 4s} 139...Ba1 {-2.36/33 34s (Kg3)} 140.Kh2 {0.00/35 1s} 140...Bc3 {-2.36/33 33s (Bf6)} 141.Kh1 {0.00/35 0s} 141...Bd2 {-2.36/28 19s (Ba5)} 142.Kh2 {0.00/37 1s} 142...Bb4 {-2.36/33 22s (Bf4+)} 143.Kh1 {0.00/26 0s} 143...Ba5 {-2.36/48 22s} 144.Kh2 {0.00/26 0s} 144...Be1 {-2.36/47 22s (Bc7+)} 145.Kh1 {0.00/21 0s} 145...Kg3 {-2.36/27 16s} 146.Kg1 {0.00/20 0s} 146...Bc3 {-2.36/35 22s (Bb4)} 147.Kh1 {0.00/20 0s} 147...Bb4 {-2.36/28 22s (Ba5)} 148.Kg1 {0.00/19 0s} 148...Bd2 {-2.36/41 22s (Be1)} 149.Kh1 {0.00/21 0s} 149...Kh3 {-2.36/30 17s (Ba5)} 150.Kg1 {0.00/21 0s} 150...Be1 {-2.36/34 22s (Kg3)} 151.Kh1 {0.00/23 0s} 151...Kg3 {-2.36/33 18s (h5)} 152.Kg1 {0.00/22 0s} 152...Bd2 {-2.28/39 50s (Bb4)} 153.Kh1 {0.00/19 0s} 153...Bf4 {-2.28/29 18s (h5)} 154.Kg1 {0.00/19 0s} 154...Be3+ {-2.28/27 19s} 155.Kh1 {0.00/19 0s} 155...Kf4 {-2.11/31 27s (Bc1)} 156.Kg2 {0.00/26 0s} 156...Ba7 {-2.28/32 22s (Kg4)} 157.Kh1 {0.00/27 0s} 157...Kg5 {-2.28/26 17s (Kf3)} 158.Kh2 {0.00/32 0s} 158...Bc5 {-2.28/27 17s (Kf4)} 159.Kg3 {0.00/33 0s} 159...Kf5 {-1.90/25 17s (Bd6+)} 160.Kh4 {0.00/30 0s} 160...Ke6 {-1.90/29 8s (Kf4)} 161.Kh5 {0.00/47 7:17m} 161...Bf8 {-0.01/51 34s} 162.Kg4 {0.00/47 6:50m} 162...Bd6 {-0.01/56 33s (Ke5)} 163.Kh5 {0.00/44 30s} 163...Bf4 {-0.01/60 35s} 164.Kg4 {0.00/45 30s} 164...Ke5 {-0.01/60 31s} 165.Kh3 {0.00/34 18s} 165...Ke4 {-0.01/59 32s} 166.Kg2 {0.00/36 2:32m} 166...Be5 {-0.01/59 32s (Ke3)} 167.Kf2 {0.00/37 6s} 167...Bf4 {-0.01/58 34s (Kf4)} 168.Kg2 {0.00/33 1:04m} 168...h5 {-0.01/59 33s (Kf5)} 169.Kf2 {0.00/30 6s} 169...Bb8 {-0.01/61 37s (h4)} 170.Kg1 {0.00/30 23s} 170...Kf3 {-0.01/57 42s (h4)} 171.Kh1 {0.00/35 9s} 171...Bg3 {-0.01/62 32s (h4)} 172.Kg1 {0.00/32 3s} 172...Bf4 {-0.01/62 35s (Bd6)} 173.Kh1 {0.00/30 1s} 173...Bd6 {-0.01/50 34s (Kg3)} 174.Kg1 {0.00/23 0s} 174...h4 {-0.01/60 42s (Bc7)} 175.Kh1 {0.00/18 0s} 175...Bf4 {-0.01/68 30s} 176.Kg1 {0.00/18 0s} 176...h3 {-0.01/53 33s (Be5)} 177.Kh1 {0.00/17 0s} 177...Ke4 {-0.01/61 40s} 178.Kg1 {0.00/18 0s} 178...Be5 {-0.01/56 29s (Bg3)} 179.Kh1 {0.00/15 0s} 179...Kf4 {-0.01/52 39s (Bf4)} 180.Kh2 {0.00/18 0s} 180...Kg4+ {-0.01/45 27s} 181.Kg1 {0.00/16 0s} 181...Kg5 {-0.01/48 25s (Bb8)} 182.Kh1 {0.00/18 0s} 182...Bg3 {-0.01/56 19s (h2)} 183.Kg1 {0.00/19 0s} 183...Kf5 {-0.01/39 25s (Be5)} 184.Kh1 {0.00/26 0s} 184...Ke4 {-0.01/49 20s (h2)} 185.Kg1 {0.00/25 0s} 185...Ke5 {-0.01/52 20s (Bb8)} 186.Kh1 {0.00/23 0s} 186...Kf6 {-0.01/48 18s (h2)} 187.Kg1 {0.00/17 0s} 187...Be5 {-0.01/43 27s (h2+)} 188.Kh1 {0.00/18 0s} 188...Kf5 {-0.01/59 27s (Bg3)} 189.Kg1 {0.00/16 0s} 189...Kg6 {-0.01/43 25s (Bc7)} 190.Kh1 {0.00/19 0s} 190...Bb8 {-0.01/64 27s (Bc7)} 191.Kg1 {0.00/18 0s} 191...Kf5 {-0.01/56 22s (Bg3)} 192.Kh1 {0.00/17 0s} 192...Be5 {-0.01/57 19s (Kg5)} 193.Kg1 {0.00/19 0s} 193...Bf4 {-0.01/57 26s (Bb8)} 194.Kh1 {0.00/19 0s} 194...Kg5 {-0.01/53 29s (Bg3)} 195.Kg1 {0.00/19 0s} 195...Be5 {-0.01/58 29s (Kf5)} 196.Kh1 {0.00/19 0s} 196...Bg3 {-0.01/50 20s (Bd6)} 197.Kg1 {0.00/21 0s} 197...Kf5 {-0.01/59 31s (Kh6)} 198.Kh1 {0.00/20 0s} 198...Ke4 {-0.01/58 31s (Kg4)} 199.Kg1 {0.00/19 0s} 199...Kd5 {-0.01/53 31s (Kf3)} 200.Kh1 {0.00/20 0s} 200...Kd6 {-0.01/54 15s (h2)} 201.Kg1 {0.00/24 0s} 201...Ke7 {-0.01/55 33s (Kd5)} 202.Kh1 {0.00/50 5:09m} 202...Kd6 {-0.01/57 45s (h2)} 203.Kg1 {0.00/41 2s} 203...Ke5 {-0.01/51 34s (h2+)} 204.Kh1 {0.00/41 9:40m} 204...Kf4 {-0.01/61 44s (h2)} 205.Kg1 {0.00/31 0s} 205...Kg4 {-0.01/61 41s (h2+)} 206.Kh1 {0.00/43 3:23m} 206...Kf4 {-0.01/61 43s (h2)} 207.Kg1 {0.00/21 1s} 207...Ke3 {-0.01/59 35s (Ke4)} 208.Kh1 {0.00/30 10s} 208...Kd2 {-0.01/59 35s (h2)} 209.Kg1 {0.00/22 0s} 209...Kc3 {-0.01/58 29s (Ke3)} 210.Kh1 {0.00/34 1:01m} 210...Kb4 {-0.01/46 42s (Kd3)} 211.Kg1 {0.00/24 0s} 211...Kc5 {-0.01/53 41s (Bf4)} 212.Kh1 {0.00/23 1s} 212...Kc4 {-0.01/46 32s (h2)} 213.Kg1 {0.00/27 0s} 213...Bd6 {-0.01/60 31s (Kd4)} 214.Kf2 {0.00/28 2s} 214...Kb4 {-0.01/57 29s (h2)} 215.Kg1 {0.00/31 1s} 215...Bf4 {-0.01/62 33s (h2+)} 216.Kf2 {0.00/28 2s} 216...Bh2 {-0.01/52 39s (h2)} 217.Kf1 {0.00/31 1s} 217...Kb3 {-0.01/55 34s (Bg3)} 218.Kf2 {0.00/33 2s} 218...Ka2 {-0.01/53 37s (Bc7)} 219.Kf3 {0.00/33 3s} 219...Bg1 {-0.01/54 37s (Bc7)} 220.Kg3 {0.00/25 1s} 220...h2 {-0.01/48 36s} 221.Kg2 {0.00/37 0s} 221...Kb2 {-0.01/42 19s (Ka1)} 222.Kh1 {0.00/35 0s} 222...Kb1 {-0.01/53 16s (Ka2)} 223.Kg2 {0.00/46 0s} 223...Ka2 {-0.01/51 24s (Kc1)} 224.Kh1 {0.00/47 0s} 224...Ka3 {-0.01/59 20s (Kb1)} 225.Kg2 {0.00/44 0s} 225...Kb2 {-0.01/60 24s (Ka4)} 226.Kh1 {0.00/53 0s} 226...Ka1 {-0.01/45 24s (Ka2)} 227.Kg2 {0.00/49 0s} 227...Ka2 {-0.01/59 18s} 228.Kh1 {0.00/54 0s} 228...Kb1 {-0.01/42 19s (Kb3)} 229.Kg2 {0.00/58 0s} 229...Kc1 {-0.01/62 25s} 230.Kh1 {0.00/60 0s} 230...Kd2 {-0.01/64 24s (Kb2)} 231.Kg2 {0.00/57 0s} 231...Ke3 {-0.01/59 21s (Kc1)} 232.Kh1 {0.00/37 0s} 232...Ke4 {-0.01/47 26s (Ke2)} 233.Kg2 {0.00/44 0s} 233...Kf5 {-0.01/61 26s (Kd5)} 234.Kh1 {0.00/46 0s} 234...Kf6 {-0.01/45 26s (Ke4)} 235.Kg2 {0.00/44 0s} 235...Ke6 {-0.01/60 26s (Ke5)} 236.Kh1 {0.00/45 0s} 236...Kd7 {-0.01/54 20s (Kf6)} 237.Kg2 {0.00/45 0s} 237...Ke6 {-0.01/61 27s (Ke8)} 238.Kh1 {0.00/45 0s} 238...Kf7 {-0.01/50 27s (Kf6)} 239.Kg2 {0.00/43 0s} 239...Kf6 {-0.01/47 27s (Kg6)} 240.Kh1 {0.00/40 0s} 240...Ke5 {-0.01/51 13s (Kg5)} 241.Kg2 {0.00/51 0s} 241...Kd4 {-0.01/49 45s} 242.Kh1 {0.00/79 1:10m} 242...Kc5 {-0.01/63 35s (Ke4)} 243.Kg2 {0.00/67 0s} 243...Kb4 {-0.01/49 45s (Kc4)} 244.Kh1 {0.00/85 2:13m} 244...Kb3 {-0.01/53 44s (Ka3)} 245.Kg2 {0.00/71 0s} 245...Ka4 {-0.01/58 43s (Kc2)} 246.Kh1 {0.00/83 2:58m} 246...Ka5 {-0.01/62 42s (Kb4)} 247.Kg2 {0.00/73 20s} 247...Kb6 {-0.01/66 42s (Ka4)} 248.Kh1 {0.00/82 10:30m} 248...Kc5 {-0.01/53 41s (Kc7)} 249.Kg2 {0.00/86 12s} 249...Kd5 {-0.01/57 41s (Kb6)} 250.Kh1 {0.00/82 13s} 250...Kd4 {-0.01/65 40s (Ke4)} 251.Kg2 {0.00/69 0s} 251...Kc3 {-0.01/61 29s (Kc4)} 252.Kh1 {0.00/64 12s} 252...Kd2 {-0.01/53 39s (Kb4)} 253.Kg2 {0.00/63 4s} 253...Ke1 {-0.01/50 32s (h1Q+)} 254.Kh1 {0.00/68 21s} 254...Kd1 {-0.01/59 30s} 255.Kg2 {0.00/65 0s} 255...Kc2 {-0.01/52 38s (Ke1)} 256.Kh1 {0.00/67 14s} 256...Kd3 {-0.01/48 37s} 257.Kg2 {0.00/61 1s} 257...Ke3 {-0.01/54 36s (Kd2)} 258.Kh1 {0.00/62 37s} 258...Bf2 {-0.01/51 35s (Ke4)} 259.Kxh2 {0.00/80 4s} 1/2-1/2 [/pgn]

1. Both engines more or less agree on a draw out of the book (this opening is probably useless in modern computer chess giving almost always a draw for engines > 3000 at this time control).
2. Starting with move 65. Booot believes in an advantage (60cp), from move 70 to move 160 Booot is a bishop and a pawn up and believes in an advantage of 995cp, later ~200cp and doesn't allow to draw. Stockfish always knows that this is a draw.
3. From move 161 to last move 259 you'll find four pawn pushes of Booot preventing the draw. Stockfish offers a draw by constantly evaluate 0.00.

As I said, I didn't search for this game, it was the first one I found in the list of games ordered by number of plies and it is a perfect example of what I said before.

Edit: Small correction/addition: The game above is the longest with SF playing white. There are three even longer games with SF playing black, all perfect examples of what I said before:

Round 1.39 Superultra vs. SF, 308 moves, SF eval constant at 0.00 from move 24 to end, Superultra two pawns up, evaluating +176cp and preventing the draw
Round 3.39 Superultra vs. SF, 300 moves, SF eval constant at 0.00 from move 36 to end, Superultra one pawn up, evaluating +160cp and preventing the draw. Moves 251-300 are KRvKR and run into 50-moves-draw.
Round 2.24 Black Marlin vs. SF, 278 moves, SF eval at 0.00 from move 27 to end, Black Marlin a pawn up and preventing draw until move 266.

[Frank Quisinsky]

Apologies if my message game across as a tad aggressive.

No, no ... all was OK with your message and your opinion was fine for me.

In the past, I could not understand my father with his old-fashioned opinions. It took him 5 years to buy his first flat screen TV. The old tube TV was expensive, why should he change it? And today I try to understand him because I do the same ... unfortunately he is no longer alive.

I have been using the same GUI for many years, the same time controls, and I always do the same in computer chess.

I will try a different GUI, different settings and the time control most people use for my next tournament. There must be a reason for what the younger generation is doing. You have written about it.

For the ongoing tournament the most important thing is to animate John for his Wasp 7 development. John likes my tournaments. I also need more positions for a test set with balanced opening positions (less draws). When the tournament is over, the test set is ready, I will change something in testing engines.

[Frank Quisinsky]

Good evening Andreas,

my last message for today, have tomorrow a long day.
Thanks for this example, a good one!

I am not told that everything you wrote is nonsense, but maybe a bit too hard to wrote "nonsense", sorry!
You often try to hold the programmer flag, for sure OK for me.

Please have a look at the game again, around move 70 ...
You understand why I use resign = off and not GUI features to stop the game.
The database is full of such bishop endgames and databases with resign=on are also full of wrong results.

Right, what you wrote here ...
If one engine does not understand the positions, the game goes a very long time.
It's not always contempt. So I am talking about "most of the time".

An example:
In round 8 a game is draw after less than 20 moves. I have replayed such games. I want to quickly update the download file and my stats in the evening. I am now waiting for the replayed game. And the game goes over 250 moves. An engine with 0.06 eval made a pawn move after move 49 and I have to wait again and again and again for the last game. Of course the game was clearly draw. The database is also full of such games from different engines.

If I have more time, I can manually create a statistic for the reasons for the long games, sorted by engines. OK, engines have produced a very long move average after many games, 95% have a problem with this and that. That is a fact when I look at it in detail.

Honestly, I prefer to look in the fast games than to search all the mistakes from the longer games.

Best regards
Frank

PS: You can make a small experiment if you like to do that.
In stats in my files you can find the move-average of the games for every engine.
Put the 20 engines with the best results in a database and looking how long the games are running, compare the move-average and search in the database for bad endgames (to 95% all is OK). Now put the other 20 engines with a long move-average in a second database, partly terrifying.

[RubiChess]

Please have a look at the game again, around move 70 ...
You understand why I use resign = off and not GUI features to stop the game.
The database is full of such bishop endgames and databases with resign=on are also full of wrong results.

As long as only one engine claims win like here, nothing should happen. There is no resign.
If Booot would have played black in this game it probably wouldn't have played into a bishop down endgame and avoided the -900 evaluation.
If an engine really thinks, it is -900 for multiple consecutive moves when it is down a wrong bishop, well, no mercy, let the GUI accept the "resign".

In round 8 a game is draw after less than 20 moves. I have replayed such games.

Why do you replay? First you complain about drawn games too long, here you complain about drawn game too short.
Why is a short draw worse than a long draw?
With replaying you are destroying a good move average of these two engines which is one of the most important qualities of an engine if I understand correctly what you tell us all the time.

And the game goes over 250 moves. An engine with 0.06 eval made a pawn move after move 49 and I have to wait again and again and again for the last game. Of course the game was clearly draw. The database is also full of such games from different engines.

Draw adjudication helps in this case. Moves > 100, eval of both engines inside [-0.2, 0.2] for 10 moves => draw!
And a better time control also helps.
Your 40 moves/20 minutes:
Short game: 40 moves = 20 minutes per engine
Long game: 240 moves = 120 minutes per engine
Better time control like 20 minutes + 5 sec:
Short game: 40 moves = ~23 minutes per engine
Long game: 240 moves = 40 minutes per engine

[Frank Quisinsky]

Hi Andreas,

Replay draws less than 20 moves ...

The FEOBOS (opening book project) ends in 2017. Most opening books produced with Eng-Eng games ~1.5 - 2.5% draws less than 20 moves. At the moment the FEOBOS book has produced 23 draws after 10,440 games = 0.2%. These are more or less detail optimisations after FEOBOS. In the last 6 years I was able to reduce the fast draws from 0.45% (results after FEOBOS) to 0.20% (today). This was one of the goals of the FEOBOS project. The main goal was to have 10 engines working together to optimise the FEOBOS database.

All the bigger tournaments (FCP Tourney-2020, 2021 and so on) I need for the statistics to the FEOBOS positions. With the last FEOBOS work I produced today about 500.000 games with longer time controls. Material I need for the last step around the long-time project FEOBOS. To find out the most interesting 500 balanced A00-E99 positions, produced the lowest draw quotes. On two other systems I tested the best lines again and again with engines prodcued different playing-styles.

This is more or less the main work I do when the day is long ....

An example: With the current 500 positions I have found out ... OK, only with 320 I am very sure, the others are unclear ...

Stockfish 16 - Dragon 3.3 (40 moves in 4 minutes) produced a draw rate of only 72%. But this is not the best engine-engine combination for a test. Much harder are combinations like Arasan - Seer, Minic - Seer, Arasan - Minic. The draw rate is around 78-82%. And with Velvet or Wasp against 200 Elo weaker I tested how many short games can be produced. I also try to keep the move average low.

---

Time control 40 moves in x minutes.
This is based on statistics from the year 2000 with 1-0, 0-1 games from TOP-20. To which time, engines produced the winning advantage. Somewhere between moves 53-57. The time controls should be adjusted in the middle. Around the year 2000 many people like to play 60 moves in x minutes. Not a good idea, because in the most important phase of the game different engines have not enough time on the clock. Today the time management of most engines is clearly better as for more than 20 years.

It is an old time-control I use, your example is more modern ... I will change that for future tourneys.
The time control I use helps the "weaker" engines a bit in endgames for a draw.
I think in neural network times the time control I use is indeed a bit old-fashioned.

When my test set is ready, I will try other GUIs and use a more modern time-control for engine-engine testing.

---

A GUI should never influence an eng-eng game!!
To your example [-0,2, 0,2] ... for the first moment I think it's not a bad idea!
I immediately thought of a game I saw today from the still running tournament.

A lost game of Berserk, not to understand what Berserk do here!
Could be a bug in Berserk ...

Let me search the game ... I believe vs. Devre!

[pgn][Event "40 Moves in 20 min"] [Site "fcp-tourney-2024, WASP-3"] [Date "2023.11.06"] [Round "6.8"] [White "Devre 4.0 NN"] [Black "Berserk 12 NN"] [Result "1-0"] 1. d4 {book 0s} Nf6 {book 0s} 2. c4 {book 0s} c5 {book 0s} 3. d5 {book 0s} e6 {book 0s} 4. Nf3 {book 0s} exd5 {book 0s} 5. cxd5 {book 0s} d6 {book 0s} 6. Nc3 {book 0s} g6 {book 0s} 7. Bf4 {book 0s} a6 {book 0s} 8. a4 {book 0s} Bg7 {book 0s} 9. e4 {book 0s} Bg4 {book 0s} 10. Be2 {book 0s} O-O {book 0s} 11. h3 {book 0s} Bxf3 {book 0s} 12. Bxf3 {book 0s} Qc7 {book 0s} 13. O-O {book 0s} Nbd7 {book 0s} 14. Qc2 {+0.68/25 48s} c4 {+0.26/35 47s} 15. a5 {+0.40/27 1:22m} Nc5 {+0.33/37 42s} 16. Na4 {+0.60/27 1:22m} Nb3 {+0.33/37 26s} 17. Nb6 {+0.38/29 1:02m} Nxa1 {+0.42/38 27s} 18. Rxa1 {+0.55/27 30s} Nd7 {+0.33/39 49s} 19. Qxc4 {+0.53/26 42s (Nxa8)} Qxc4 {+0.17/36 25s} 20. Nxc4 {+0.24/27 1:06m} Rac8 {+0.13/38 44s} 21. Be2 {+0.63/26 47s} Rfd8 {+0.13/38 32s} 22. Bxd6 {+0.73/27 32s} Nf6 {+0.12/37 28s} 23. Be7 {+0.72/27 5s} Nxe4 {+0.12/37 26s} 24. Bxd8 {+0.61/25 7s (Nb6)} Rxd8 {+0.10/38 28s} 25. Rd1 {+0.57/26 14s} Nd6 {+0.09/41 29s} 26. b3 {+0.46/27 1:24m} Bc3 {+0.07/43 1:23m} 27. Rc1 {+0.46/30 1:06m} Bb4 {+0.07/45 1:27m} 28. Nxd6 {+0.48/29 52s} Bxd6 {+0.07/45 41s} 29. Rc4 {+0.42/28 1:07m} Rd7 {+0.07/41 26s} 30. Kf1 {+0.35/27 1:07m (Bf3)} Rc7 {+0.07/43 1:51m} 31. Rxc7 {+0.32/30 56s} Bxc7 {+0.05/43 2:32m} 32. Bf3 {+0.34/29 40s (d6)} h5 {+0.05/41 33s} 33. Ke2 {+0.35/30 40s (d6)} Kf8 {+0.05/41 1:17m} 34. d6 {+0.39/28 50s} Bxd6 {+0.05/41 23s} 35. Bxb7 {+0.40/31 31s} Bb4 {+0.05/40 36s} 36. Bxa6 {+0.40/28 17s} Bxa5 {+0.05/39 1:33m} 37. Bc4 {+0.40/28 33s (Kf3)} Ke7 {+0.05/36 18s} 38. Kd3 {+0.40/26 33s} h4 {+0.05/35 1s} 39. Ke4 {+0.40/25 7s (f3)} Be1 {+0.05/33 1s} 40. f3 {+0.40/28 25s} g5 {+0.05/33 1s} 41. Kd5 {+0.44/30 34s (Kd4)} Kd7 {+0.05/39 55s} 42. Be2 {+0.44/31 22s (Bb5+)} f6 {+0.05/40 1:50m} 43. f4 {+0.44/33 57s (Bb5+)} gxf4 {+0.05/40 1:02m} 44. Bf3 {+0.36/32 30s (Kc5)} Kc7 {+0.05/41 1:10m} 45. Ke6 {+0.36/34 45s (Kc5)} Kb6 {+0.01/36 18s} 46. Kxf6 {+0.53/40 44s (Kd5)} Kc5 {+0.01/41 36s} 47. Kf5 {+0.53/40 29s} Kb4 {+0.01/42 1:05m} 48. Bd5 {+0.53/35 5s} Bg3 {+0.01/43 53s} 49. Bf7 {+0.53/40 38s (Kg4)} Kc5 {+0.01/42 33s} 50. Be6 {+0.53/42 42s (Kg5)} Kb4 {+0.01/41 1:34m} 51. Bc4 {+0.53/41 46s (Kg5)} Kc5 {+0.01/44 2:01m} 52. Be6 {+0.53/40 45s (Kg5)} Kc6 {+0.01/42 30s} 53. Bc4 {+0.53/38 35s (Ke4)} Kc5 {+0.01/42 47s} 54. Ke6 {+0.53/37 7s (Kg4)} Kb4 {+0.01/42 42s} 55. Kd5 {+0.53/41 32s} Be1 {+0.01/44 23s} 56. Ke4 {+0.53/40 35s} Bg3 {+0.01/44 8s} 57. Bd5 {+0.53/37 8s (Kf3)} Kc5 {+0.01/42 8s} 58. Bc4 {+0.53/42 53s (Bf7)} Bh2 {+0.01/45 51s} 59. Be2 {+0.53/39 48s (Kf3)} Kb4 {+0.01/44 2:28m} 60. Bc4 {+0.53/41 34s} Bg3 {+0.01/44 10s} 61. Bf7 {+0.53/40 24s (Bd5)} Kc5 {+0.01/42 18s} 62. Kf3 {+0.53/39 7s} Kb4 {+0.01/42 13s} 63. Bd5 {+0.53/40 43s (Be6)} Kc5 {+0.01/41 19s} 64. Be6 {+0.53/37 13s (Ke4)} Kb4 {+0.01/41 37s} 65. Bc4 {+0.53/34 4s (Ke4)} Kc5 {+0.01/42 1s} 66. Bf7 {+0.53/42 42s} Kb4 {+0.01/41 2s} 67. Bd5 {+0.53/39 45s (Ke2)} Kc5 {+0.01/42 3s} 68. Bc4 {+0.53/38 32s (Ke4)} Kb4 {+0.01/39 6s} 69. Ke2 {+0.53/39 27s (Ke4)} Bh2 {+0.01/38 1s} 70. Kf3 {+0.53/38 7s (Bd5)} Bg3 {+0.01/37 0s} 71. Be6 {+0.53/40 41s} Kc5 {+0.01/37 0s} 72. Kg4 {+0.53/38 13s (Ke2)} Kb4 {+0.01/39 0s} 73. Kg5 {+0.53/39 39s (Kf3)} Kc5 {+0.01/23 0s} 74. Kg6 {+0.53/39 39s (Kg4)} Kb4 {+0.01/20 0s} 75. Kg7 {+0.53/37 6s (Kh5)} Ka3 {+0.01/21 0s} 76. Bd5 {+0.53/41 34s (Kg6)} Kb4 {+0.01/10 0s} 77. Kg6 {+0.53/38 17s (Be6)} f3 {book 0s} 78. gxf3 {+2.80/33 24s} Bb8 {book 0s} 79. Kg5 {+3.63/29 17s (Kh5)} Ba7 {book 0s} 80. Kxh4 {+4.92/27 11s} Bb8 {book 0s} 81. Kg5 {+6.33/25 57s (Kg4)} Ba7 {+13.27/33 1:23m} 82. f4 {+7.84/23 49s} Be3 {+M-992/37 5:26m} 83. Kf5 {+8.58/22 54s} Kc3 {+M-994/36 53s} 84. Ke5 {+9.96/22 36s} Bd4+ {+M26/37 50s} 85. Ke6 {+10.42/23 25s} Bg7 {+M25/35 12s} 86. h4 {+13.13/22 16s (f5)} Kd4 {+M22/36 31s} 87. f5 {+15.33/23 30s} Bh8 {+M18/36 33s} 88. f6 {+17.27/25 54s (Bc4)} Bxf6 {+M12/42 27s} 89. Kxf6 {+M13/27 51s} Kxd5 {+M11/52 11s} 90. h5 {+M11/35 51s} Kd4 {+M10/54 10s} 91. h6 {+M10/36 41s} Kc3 {+M9/59 10s} 92. h7 {+M9/34 13s} Kb2 {+M8/60 10s} 93. h8=Q {+M8/34 48s} Kxb3 {+M7/64 10s} 94. Ke5 {+M7/34 48s (Qc8)} Kc3 {+M6/68 13s} 95. Kd5+ {+M6/33 44s (Ke4+)} Kd3 {+M5/72 17s} 96. Qd4+ {+M5/36 31s} Kc2 {+M4/75 14s} 97. Kc4 {+M4/74 7s} Kb1 {+M3/88 18s} 98. Kc3 {+M3/100 0s} Ka1 {+M2/200 0s} 99. Kc2+ {+M2/100 0s} Ka2 {+M1/200 0s} 100. Qb2# {+M1/100 0s (Qa4+)} 1-0[/pgn]

Move 77 ...
I think it make sense to check the other 4 lost games from Berserk in my tourney.
And "book" comes from the Shredder GUI, one of the mistakes in Shredder I dislike a lot.

Often both engines have been producing evals like 0.2 for a long time and suddenly the game is over. So many thinkable constellations ... never, a GUI should influence an engine-engine match. The arguments for a different time control is OK for me, but every game should run to the end. It is the job of the programmer to try to reduce the move average. Here my opinion is very clear.

Best
Frank

Andreas:
No pull-ups to avoid engine faults.
It's a feature that with resign-off games much more is to see as with resign=on games.
For fun-tourneys I am using on my Notebook often the time-control "3+2".

and ...
Du hast ferner gut reden ... Dein Programm macht das was mir wichtig ist richtig. Ich finde RubiChess wirklich großartig weil es alles sehr ausgeglichen spielt. Kein Book mehr auf für mich kompliziertes english.

[chessica]

[d]k7/P5P1/8/8/3B4/5K2/b7/8 w - - 49 250 500 ply`s

[Uri Blass]

Hi Andreas,

Replay draws less than 20 moves ...

The FEOBOS (opening book project) ends in 2017. Most opening books produced with Eng-Eng games ~1.5 - 2.5% draws less than 20 moves. At the moment the FEOBOS book has produced 23 draws after 10,440 games = 0.2%. These are more or less detail optimisations after FEOBOS. In the last 6 years I was able to reduce the fast draws from 0.45% (results after FEOBOS) to 0.20% (today). This was one of the goals of the FEOBOS project. The main goal was to have 10 engines working together to optimise the FEOBOS database.

All the bigger tournaments (FCP Tourney-2020, 2021 and so on) I need for the statistics to the FEOBOS positions. With the last FEOBOS work I produced today about 500.000 games with longer time controls. Material I need for the last step around the long-time project FEOBOS. To find out the most interesting 500 balanced A00-E99 positions, produced the lowest draw quotes. On two other systems I tested the best lines again and again with engines prodcued different playing-styles.

This is more or less the main work I do when the day is long ....

An example: With the current 500 positions I have found out ... OK, only with 320 I am very sure, the others are unclear ...

Stockfish 16 - Dragon 3.3 (40 moves in 4 minutes) produced a draw rate of only 72%. But this is not the best engine-engine combination for a test. Much harder are combinations like Arasan - Seer, Minic - Seer, Arasan - Minic. The draw rate is around 78-82%. And with Velvet or Wasp against 200 Elo weaker I tested how many short games can be produced. I also try to keep the move average low.

---

Time control 40 moves in x minutes.
This is based on statistics from the year 2000 with 1-0, 0-1 games from TOP-20. To which time, engines produced the winning advantage. Somewhere between moves 53-57. The time controls should be adjusted in the middle. Around the year 2000 many people like to play 60 moves in x minutes. Not a good idea, because in the most important phase of the game different engines have not enough time on the clock. Today the time management of most engines is clearly better as for more than 20 years.

It is an old time-control I use, your example is more modern ... I will change that for future tourneys.
The time control I use helps the "weaker" engines a bit in endgames for a draw.
I think in neural network times the time control I use is indeed a bit old-fashioned.

When my test set is ready, I will try other GUIs and use a more modern time-control for engine-engine testing.

---

A GUI should never influence an eng-eng game!!
To your example [-0,2, 0,2] ... for the first moment I think it's not a bad idea!
I immediately thought of a game I saw today from the still running tournament.

A lost game of Berserk, not to understand what Berserk do here!
Could be a bug in Berserk ...

Let me search the game ... I believe vs. Devre!

[pgn][Event "40 Moves in 20 min"] [Site "fcp-tourney-2024, WASP-3"] [Date "2023.11.06"] [Round "6.8"] [White "Devre 4.0 NN"] [Black "Berserk 12 NN"] [Result "1-0"] 1. d4 {book 0s} Nf6 {book 0s} 2. c4 {book 0s} c5 {book 0s} 3. d5 {book 0s} e6 {book 0s} 4. Nf3 {book 0s} exd5 {book 0s} 5. cxd5 {book 0s} d6 {book 0s} 6. Nc3 {book 0s} g6 {book 0s} 7. Bf4 {book 0s} a6 {book 0s} 8. a4 {book 0s} Bg7 {book 0s} 9. e4 {book 0s} Bg4 {book 0s} 10. Be2 {book 0s} O-O {book 0s} 11. h3 {book 0s} Bxf3 {book 0s} 12. Bxf3 {book 0s} Qc7 {book 0s} 13. O-O {book 0s} Nbd7 {book 0s} 14. Qc2 {+0.68/25 48s} c4 {+0.26/35 47s} 15. a5 {+0.40/27 1:22m} Nc5 {+0.33/37 42s} 16. Na4 {+0.60/27 1:22m} Nb3 {+0.33/37 26s} 17. Nb6 {+0.38/29 1:02m} Nxa1 {+0.42/38 27s} 18. Rxa1 {+0.55/27 30s} Nd7 {+0.33/39 49s} 19. Qxc4 {+0.53/26 42s (Nxa8)} Qxc4 {+0.17/36 25s} 20. Nxc4 {+0.24/27 1:06m} Rac8 {+0.13/38 44s} 21. Be2 {+0.63/26 47s} Rfd8 {+0.13/38 32s} 22. Bxd6 {+0.73/27 32s} Nf6 {+0.12/37 28s} 23. Be7 {+0.72/27 5s} Nxe4 {+0.12/37 26s} 24. Bxd8 {+0.61/25 7s (Nb6)} Rxd8 {+0.10/38 28s} 25. Rd1 {+0.57/26 14s} Nd6 {+0.09/41 29s} 26. b3 {+0.46/27 1:24m} Bc3 {+0.07/43 1:23m} 27. Rc1 {+0.46/30 1:06m} Bb4 {+0.07/45 1:27m} 28. Nxd6 {+0.48/29 52s} Bxd6 {+0.07/45 41s} 29. Rc4 {+0.42/28 1:07m} Rd7 {+0.07/41 26s} 30. Kf1 {+0.35/27 1:07m (Bf3)} Rc7 {+0.07/43 1:51m} 31. Rxc7 {+0.32/30 56s} Bxc7 {+0.05/43 2:32m} 32. Bf3 {+0.34/29 40s (d6)} h5 {+0.05/41 33s} 33. Ke2 {+0.35/30 40s (d6)} Kf8 {+0.05/41 1:17m} 34. d6 {+0.39/28 50s} Bxd6 {+0.05/41 23s} 35. Bxb7 {+0.40/31 31s} Bb4 {+0.05/40 36s} 36. Bxa6 {+0.40/28 17s} Bxa5 {+0.05/39 1:33m} 37. Bc4 {+0.40/28 33s (Kf3)} Ke7 {+0.05/36 18s} 38. Kd3 {+0.40/26 33s} h4 {+0.05/35 1s} 39. Ke4 {+0.40/25 7s (f3)} Be1 {+0.05/33 1s} 40. f3 {+0.40/28 25s} g5 {+0.05/33 1s} 41. Kd5 {+0.44/30 34s (Kd4)} Kd7 {+0.05/39 55s} 42. Be2 {+0.44/31 22s (Bb5+)} f6 {+0.05/40 1:50m} 43. f4 {+0.44/33 57s (Bb5+)} gxf4 {+0.05/40 1:02m} 44. Bf3 {+0.36/32 30s (Kc5)} Kc7 {+0.05/41 1:10m} 45. Ke6 {+0.36/34 45s (Kc5)} Kb6 {+0.01/36 18s} 46. Kxf6 {+0.53/40 44s (Kd5)} Kc5 {+0.01/41 36s} 47. Kf5 {+0.53/40 29s} Kb4 {+0.01/42 1:05m} 48. Bd5 {+0.53/35 5s} Bg3 {+0.01/43 53s} 49. Bf7 {+0.53/40 38s (Kg4)} Kc5 {+0.01/42 33s} 50. Be6 {+0.53/42 42s (Kg5)} Kb4 {+0.01/41 1:34m} 51. Bc4 {+0.53/41 46s (Kg5)} Kc5 {+0.01/44 2:01m} 52. Be6 {+0.53/40 45s (Kg5)} Kc6 {+0.01/42 30s} 53. Bc4 {+0.53/38 35s (Ke4)} Kc5 {+0.01/42 47s} 54. Ke6 {+0.53/37 7s (Kg4)} Kb4 {+0.01/42 42s} 55. Kd5 {+0.53/41 32s} Be1 {+0.01/44 23s} 56. Ke4 {+0.53/40 35s} Bg3 {+0.01/44 8s} 57. Bd5 {+0.53/37 8s (Kf3)} Kc5 {+0.01/42 8s} 58. Bc4 {+0.53/42 53s (Bf7)} Bh2 {+0.01/45 51s} 59. Be2 {+0.53/39 48s (Kf3)} Kb4 {+0.01/44 2:28m} 60. Bc4 {+0.53/41 34s} Bg3 {+0.01/44 10s} 61. Bf7 {+0.53/40 24s (Bd5)} Kc5 {+0.01/42 18s} 62. Kf3 {+0.53/39 7s} Kb4 {+0.01/42 13s} 63. Bd5 {+0.53/40 43s (Be6)} Kc5 {+0.01/41 19s} 64. Be6 {+0.53/37 13s (Ke4)} Kb4 {+0.01/41 37s} 65. Bc4 {+0.53/34 4s (Ke4)} Kc5 {+0.01/42 1s} 66. Bf7 {+0.53/42 42s} Kb4 {+0.01/41 2s} 67. Bd5 {+0.53/39 45s (Ke2)} Kc5 {+0.01/42 3s} 68. Bc4 {+0.53/38 32s (Ke4)} Kb4 {+0.01/39 6s} 69. Ke2 {+0.53/39 27s (Ke4)} Bh2 {+0.01/38 1s} 70. Kf3 {+0.53/38 7s (Bd5)} Bg3 {+0.01/37 0s} 71. Be6 {+0.53/40 41s} Kc5 {+0.01/37 0s} 72. Kg4 {+0.53/38 13s (Ke2)} Kb4 {+0.01/39 0s} 73. Kg5 {+0.53/39 39s (Kf3)} Kc5 {+0.01/23 0s} 74. Kg6 {+0.53/39 39s (Kg4)} Kb4 {+0.01/20 0s} 75. Kg7 {+0.53/37 6s (Kh5)} Ka3 {+0.01/21 0s} 76. Bd5 {+0.53/41 34s (Kg6)} Kb4 {+0.01/10 0s} 77. Kg6 {+0.53/38 17s (Be6)} f3 {book 0s} 78. gxf3 {+2.80/33 24s} Bb8 {book 0s} 79. Kg5 {+3.63/29 17s (Kh5)} Ba7 {book 0s} 80. Kxh4 {+4.92/27 11s} Bb8 {book 0s} 81. Kg5 {+6.33/25 57s (Kg4)} Ba7 {+13.27/33 1:23m} 82. f4 {+7.84/23 49s} Be3 {+M-992/37 5:26m} 83. Kf5 {+8.58/22 54s} Kc3 {+M-994/36 53s} 84. Ke5 {+9.96/22 36s} Bd4+ {+M26/37 50s} 85. Ke6 {+10.42/23 25s} Bg7 {+M25/35 12s} 86. h4 {+13.13/22 16s (f5)} Kd4 {+M22/36 31s} 87. f5 {+15.33/23 30s} Bh8 {+M18/36 33s} 88. f6 {+17.27/25 54s (Bc4)} Bxf6 {+M12/42 27s} 89. Kxf6 {+M13/27 51s} Kxd5 {+M11/52 11s} 90. h5 {+M11/35 51s} Kd4 {+M10/54 10s} 91. h6 {+M10/36 41s} Kc3 {+M9/59 10s} 92. h7 {+M9/34 13s} Kb2 {+M8/60 10s} 93. h8=Q {+M8/34 48s} Kxb3 {+M7/64 10s} 94. Ke5 {+M7/34 48s (Qc8)} Kc3 {+M6/68 13s} 95. Kd5+ {+M6/33 44s (Ke4+)} Kd3 {+M5/72 17s} 96. Qd4+ {+M5/36 31s} Kc2 {+M4/75 14s} 97. Kc4 {+M4/74 7s} Kb1 {+M3/88 18s} 98. Kc3 {+M3/100 0s} Ka1 {+M2/200 0s} 99. Kc2+ {+M2/100 0s} Ka2 {+M1/200 0s} 100. Qb2# {+M1/100 0s (Qa4+)} 1-0[/pgn]

Move 77 ...
I think it make sense to check the other 4 lost games from Berserk in my tourney.
And "book" comes from the Shredder GUI, one of the mistakes in Shredder I dislike a lot.

Often both engines have been producing evals like 0.2 for a long time and suddenly the game is over. So many thinkable constellations ... never, a GUI should influence an engine-engine match. The arguments for a different time control is OK for me, but every game should run to the end. It is the job of the programmer to try to reduce the move average. Here my opinion is very clear.

Best
Frank

Andreas:
No pull-ups to avoid engine faults.
It's a feature that with resign-off games much more is to see as with resign=on games.
For fun-tourneys I am using on my Notebook often the time-control "3+2".

and ...
Du hast ferner gut reden ... Dein Programm macht das was mir wichtig ist richtig. Ich finde RubiChess wirklich großartig weil es alles sehr ausgeglichen spielt. Kein Book mehr auf für mich kompliziertes english.

I guess the problem of berserk is bad time management in x minutes/40 moves time control.
It already got only depth 10 at move 76 and used 0 seconds because of bad time management and it probably played move 77 immediately.