Battle of the Goths 2012 (live broadcast)

[Evert]

Sjaak 470 is now available (source and Win32/Win64 binaries for now, I'll add Linux and OS X binaries later today).
I have not tested other variants, but it should not be worse than the odl 399 release for Gothic/Capablanca.

[Evert]

After the Battle of the Goths has finished I will do a gauntlet with Heretic 0.2 versus all playoff partcipants, to get ratings. There is a slight chance I will also be able to add the commercial program Gothic Vortex to that. (If Ed Trice sends me the sources, and I manage to convert it to WB protocol.)

If you do that, would it be possible to replay some of the early games with the new version of Sjaak as well? I would be interested in seeing the results of that.

[hgm]

Cross table of the second division after three cycles:

Code: Select all

                              Fair Sjaa Here BigL ArcB Ches
 1. Fairy-Max                 #### 1110 111= 11=1 111= 0100
                              #### 0111 1100 ==   10   11     69%  23.5 (525.0, 373.5)

 2. Sjaak                     0001 #### 1101 1101 =011 1111
                              1000 #### 1100 11   11   11     66%  22.5 (533.0, 298.0)

 3. Heretic                   000= 0010 #### 1111 1111 11=1
                              0011 0011 #### 1=   11   11     66%  22.5 (533.0, 282.5)

 4. BigLion                   00=0 0010 0000 #### 1101 0111
                              ==   00   0=   #### 10   =1     38%  11.5 (507.0, 136.0)

 5. ArcBishop                 000= =100 0000 0010 #### 1110
                              01   00   00   01   #### 11     33%  10.0 (516.0, 122.0)

 6. ChessV_Winboard.exe       1011 0000 00=0 1000 0001 ####
                              00   00   00   =0   00   ####   20%   6.0 (540.0, 109.0)

It turns out the version of ChessV I am playing is buggy. It is actually a private version, the binary of which Gregory Strong e-mailed me about a year ago. It differs from the last official release by supporting Xiangqi and Spartan Chess and fixed a bug in Knightmate support). But it seems there is something very wrong with it...

At some point (which can be very late in the game) there suddenly appear ridiculous scores, like totally imagined checkmates for one side of the other. And these persist then for a series of blunder moves, until the sddenly fall back to a realistic value. (Which, by that time, is usually between -16 and -32...). The bug does not strike always, which is why id did still score some points against quite strong opponents.

Too bad, because normally ChessV would only be about 50 Elo below Fairy-Max.

Anyway, I started the last cycle of this division now (from the Carrera position). After that we will see the top engines!

[hgm]

Final rankings of the Second Division:

Code: Select all

                              Fair Sjaa Here BigL ArcB Ches
 1. Fairy-Max                 #### 1110 111= 11=1 111= 0100
                              #### 0111 1100 ==1= 1011 1111   72%  29.0 (728.0, 529.8)

 2. Sjaak                     0001 #### 1101 1101 =011 1111
                              1000 #### 1100 1111 1111 1111   71%  28.5 (732.0, 440.5)

 3. Heretic                   000= 0010 #### 1111 1111 11=1
                              0011 0011 #### 1=11 1111 1111   71%  28.5 (732.0, 419.0)

 4. BigLion                   00=0 0010 0000 #### 1101 0111
                              ==0= 0000 0=00 #### 1001 =101   35%  14.0 (848.0, 204.8)

 5. ArcBishop                 000= =100 0000 0010 #### 1110
                              0100 0000 0000 0110 #### 1110   30%  12.0 (864.0, 176.3)

 6. ChessV_Winboard.exe       1011 0000 00=0 1000 0001 ####
                              0000 0000 0000 =010 0001 ####   20%   8.0 (896.0, 160.3)

Fairy-Max managed to keep the suspense to the very last minute: in the last game it was badly losing against BigLion (at some point it was below -6), which would have meant a 3-way tie for the top position. But both sides still had Chancellors, and at one point BigLion let its guard down, and allowed Fairy-Max to actually get ahead eval wise, by picking off some unprotected pieces while chasing BigLion's King. But BigLion could counter with an uncurable mate threat, so there finally was no choice than to go for a perpetual. The half point was enough for Fairy-Max to finish first place.

Unfortunately the ranking has very little reality value; Sjaak and Heretic played their important games with weak versions, which were later replaced. I used a version for ChessV that later turned out quite buggy, and threw more than half its games. In the games that were spared by the bug, it could play quite strong, so for its opponent's playing it was more or less like flipping a coin whether they would win.

Oh well, better luck next time...

Now I will prepare the round-robin for the playoffs, with the 7 top engines, I hope to start that within an hour or so.

[Daniel Shawul]

You are lucky that you have older versions in git. Before using that for Nebiyu, I introduced a bug that made it hundreds of elo weaker and had an only exe to work with. I had to decompile and compare codes to find out it was a problem with history counters!

[Evert]

Couldn't live without version control!
It's really one of those things that I find I can't do without now that I've grown used to it. In fact, I tend to not think of changes as "saved" until they've savely been committed to the repository.

[hgm]

OK, the playoffs have begun. Participants:

1.Bihasa
2.Joker80
3.Smirf
4.TJchess10x8
5.Spartacus
6.NebiyuChess
7.TSCP Gothic

The games from the qualifier between Spartacus, Nebiyu and TSCP Gothic will keep counting. To remind you:

Code: Select all

                              Spar Nebi tscp
 1. Spartacus                 #### 10=1 1111
                              #### 1110 1111   84%  13.5 ( 84.0,  62.8)

 2. NebiyuChess               01=0 #### 0011
                              0001 #### 1111   53%   8.5 (124.0,  45.8)

 3. tscpgothic                0000 1100 ####
                              0000 0000 ####   12%   2.0 (176.0,  17.0)

[hgm]

In the first cycle of the playoffs Bihasa already dealt a devastating blow to all opponents, and leads with 11 out of 12.

Code: Select all

                              Biha Spar TJch Nebi Joke SMIR tscp
 1. Bihasa v                  #### 11   11   11   1=   =1   11  
                              ####                                 92%  11.0

 2. Spartacus                 00   #### 10   10=1 01   11   1111
                                   ####      1110           1111   73%  17.5

 3. TJchess                   00   01   #### 10   11   11   11  
                                        ####                       67%   8.0

 4. NebiyuChess               00   01=0 01   #### 11   01   0011
                                   0001      ####           1111   52%  12.5

 5. Joker                     0=   10   00   00   #### 11   01  
                                                  ####             37%   4.5

 6. SMIRF MS                  =0   00   00   10   00   #### 01  
                                                       ####        21%   2.5

 7. tscpgothic                00   0000 00   1100 10   10   ####
                                   0000      0000           ####   17%   4.0 

[George Tsavdaris]

In the first cycle of the playoffs Bihasa already dealt a devastating blow to all opponents, and leads with 11 out of 12.

12 out of 13 now and it is going to win again as it seems. It has TJChess on the ropes after the 18th move only!
I predicted it will have only 1 draw out of 48 games but i was wrong. I guess a loss will come sooner or later although it's not that easy as it seems. In the game against Smirf which ended a draw, Smirf should have won by all means but Bihasa managed a draw.

Code: Select all

    name                       perc.  pnt games  Buchh    S-B

 1. Bihasa v                    92%  12.0 / 13  (110.0, 106.0)
 2. Spartacus                   71%  18.5 / 26  (199.0, 124.8)
 3. TJchess                     62%   8.0 / 13  (122.0,  55.0)
 4. NebiyuChess                 52%  12.5 / 24  (236.0,  91.8)
 5. Joker                       39%   5.5 / 14  (152.0,  52.0)
 6. SMIRF MS                    21%   2.5 / 12  (121.0,  22.5)
 7. tscpgothic                  17%   4.0 / 24  (304.0,  33.0)

[hgm]

Playoffs after two cycles (Gothic + Capablanca).

Code: Select all

                    Biha Spar TJch Nebi Joke SMIR tscp
 1. Bihasa v        #### 1111 1111 1111 1=01 =111 1111
                    ####                                 92%  22.0 (296.0, 276.0)

 2. Spartacus       0000 #### 1011 10=1 0101 1101 1111
                         ####      1110           1111   67%  21.5 (376.0, 204.8)

 3. TJchess         0000 0100 #### 1011 1111 1111 1111
                              ####                       67%  16.0 (320.0, 152.0)

 4. NebiyuChess     0000 01=0 0100 #### 1101 0111 0011
                         0001      ####           1111   48%  15.5 (424.0, 144.8)

 5. Joker           0=10 1010 0000 0010 #### 1111 0111
                                        ####             48%  11.5 (338.0, 125.5)

 6. SMIRF MS        =000 0010 0000 1000 0000 #### 0111
                                             ####        23%   5.5 (362.0,  60.0)

 7. tscpgothic      0000 0000 0000 1100 1000 1000 ####
                         0000      0000           ####   12%   4.0 (516.0,  48.0)

In the second cycle, Bihasa again crushed the opposition with 11 out of 12. But this time there was a single loss, showing at least it is not completely invincible.

A bit overshadowed by the Bihasa violence is how much TJchess10x8 has actually been improved compared to the last tourney. If it were not for the newcomers, TJchess10x8 would be very firmly in the lead by now! I must point out that the version 1.1 that is playing here is not yet released, and should not be confused with the latest released version 1.01.

Now a cycle of Bird's Chess has been started.