Fairy-Max 4.8 released

[hgm]

Note I just released a derivative of uMax that can handle (almost) arbitrary fairy pieces, and board sizes upto 12x8 (and as a result can play Capablanca Chess):

http://wbforum.vpittlik.org/viewtopic.php?t=6880

A sample game, played under Winboard_F:

[Event "Computer Chess Game"]
[Site "FOM-RHKA8J2A5WY"]
[Date "2007.10.18"]
[Round "-"]
[White "fmax4_8g"]
[Black "fmax4_8g"]
[Result "1/2-1/2"]
[TimeControl "40/2400"]
[Variant "capablanca"]
[Annotator "1. +0.24 1... +0.39"]

1. c3 d5 2. Nh3 Nd7 3. d4 h6 4. f4 e6 5. Nd2 i5 6. j3 a5 7. i4 j5 8. e4 f5
9. exf5 exf5 10. Qxe8+ Kxe8 11. g4 Nf6 12. g5 hxg5 13. Nxg5 Nh6 14. h3 Ne4
15. ixj5 Nxg5 16. fxg5 Bxg5 17. Cg3 Ae6 18. j6 Rxj6 19. j4 f4 20. Ch5 Rj5
21. Nf3 Cf7 22. Nxg5 Axg5 23. Bi3 Ae6 24. O-O g6 25. Re1 gxh5 26. Rxe6+ Kd7
27. Re7+ Cxe7 28. Bxe7 Kxe7 29. Axf4 Ra6 30. jxi5 Rxi5+ 31. Kh2 h4 32. Axc7
a4 33. Bf3 Rf5 34. Kg2 Rg6+ 35. Kf2 Be6 36. Re1 Nf7 37. Ke2 Nd6 38. Bg4
Rfg5 39. Rg1 b5 40. a3 Rh6 41. Ke3 Rf6 42. Bxe6 Nc4+ 43. Ke2 Rxg1 44. Axd5+
Kd6 45. Axf6 Rg2+ 46. Kf3 Kxe6 47. Ad8+ Kd7 48. Axh4 Rxb2 49. Af5+ Kc7 50.
d5 Rb3 51. Ae6+ Kd6 52. Af7+ Kc5 53. Ae6+ Kd6
{Draw by repetition} 1/2-1/2

[George Tsavdaris]

Note I just released a derivative of uMax that can handle (almost) arbitrary fairy pieces, and board sizes upto 12x8 (and as a result can play Capablanca Chess):

Many thanks but without Winboard_F i've found it impossible to play from the DOS executable of FMax4_8w, since i didn't know the commands i had to put for the starting board, time control etc....

Any idea of when you will release Winboard_F? I'm very interested and have some experience in 10x8 variants so i'm looking forward!

[hgm]

Any idea of when you will release Winboard_F?

Very soon, I hope. There still are some loose ends that I have to tie up. Like allowing a wider variety of piece-indicating letters in FENs in such a way that they are passed to the engine as they occur in the FEN, rather than being determined by the Winboard display symbol. (The engine might know many more piece types than the 11 that Winboard_F now has, which is OK, because they never occur on the board simultaneously, and the user can already decide how they look.)

For adjudications I still want it to verify engine claims (repetition and 50-move draws, illegal-move claims). Curent Winboard gets into a deadlock if one of the engines does not accept the move of the other, as it waits forever for the obstinate engine to come with a move, while it thinks it is not its turn. This means the entire tourney hangs...

Another problem I still want to tackle is the pause between games of a match. I already made that user adjustable, but I really would like to include an 'as-fast-as-possible' option, dependent on a signal from the engine that it is ready to start the new game.

Finally, there is already someone working on an xboard version, and we want to make sure that the sources do not needlessly diverge. So we will have to do some source comparison before releasing the source, and under GPL it is not allowed to release without source.

Btw, ChessGUI does support 10x8 boards and Winboard protocol, FMax should work under that as well.

[George Tsavdaris]

OK i have finally been able to setup FMax to play in Winboard.
My main problem was the pieces Archbishop and Chancellor that didn't appear in Winboard. The only way i could find is the font solution of the new winboard. So i had to manually create a font with these 2 new pieces!
Here is a screenshot:


How did you play making the 2 new pieces appear? Was any other available font or workaround possible?


After this i played 2 games against the latest freeware version of Smirf. 40/10 was the time control.
In the first game FMax and Smirf played strangely at first and FMax reached a winning position in 10th move only having an Archbishop for a Knight. Yet Smirf managed to win.... In the second game Smirf outplayed Fmax and won easily.


[Event "Computer Chess Game"]
[Site "????????"]
[Date "2007.10.20"]
[Round "1"]
[White "Smirf"]
[Black "fmax4_8w"]
[Result "1-0"]
[TimeControl "40/600"]
[Variant "capablanca"]
[Annotator "1... +0.12"]

1. Nh3 e5 2. Cg3 d5 3. Ab3 f5 4. e3 d4 5. Ac5+ Be7 6. Ad3 e4 7. Af4 Bd6 8.
Axi7 Bxg3 9. Axh8 Bj6 10. Axj6 Nxj6 11. Na3 d3 12. cxd3 exd3 13. Nf4 Qd8
14. h3 a5 15. Bi3 Qd6 16. Nb5 Qb6 17. a4 c6 18. Nc3 Qxb2 19. Rb1 Qa3 20. e4
fxe4 21. Qxe4 Ad6 22. Ne6+ Kf7 23. Rxb7+ Ke8 24. Nxg7+ Kf8 25. Ne6+ Ke8 26.
Nc7+ Kf8 27. Rxb8+ Ac8 28. Rxc8+ Rxc8 29. Bj4+ h6 30. Bxh6+ Kf7 31. Qe6#
{White mates} 1-0


[Event "Computer Chess Game"]
[Site "?????"]
[Date "2007.10.20"]
[Round "2"]
[White "fmax4_8w"]
[Black "Smirf"]
[Result "0-1"]
[TimeControl "40/600"]
[Variant "capablanca"]
[Annotator "1. +0.15"]

1. d3 Nh6 2. e3 Ad6 3. f4 Nc6 4. Nd2 e5 5. Ne4 Ae7 6. c4 exf4 7. exf4 Ah4
8. Ng3 Qxe1+ 9. Kxe1 Ni4 10. Bg4 d6 11. j3 Nxg3 12. hxg3 Af6 13. Bh3 h6 14.
i3 Bi6 15. i4 Cg6 16. Bf5 Ce7+ 17. Be4 Bh7 18. Bi3 Axi3 19. Cxi3 Cxe4+ 20.
dxe4 Bxi3 21. e5 Nb4 22. Kd2 Rd8 23. Kc3 Nc2 24. Rb1 dxe5 25. Nh3 Bf6 26.
f5 Bxf5 27. g4 Bg6 28. g5 Be7 29. gxh6 gxh6 30. b4 Ne3 31. Ae2 Bxb1 32.
Rxb1 Rg8 33. Axj7 Rxg2 34. a3 Rc2+ 35. Kb3 Rd3+ 36. Ka4 Nxc4 37. Ra1 Rb2
38. Axh6+ ixh6 39. Kb5 Nxa3+ 40. Ka5 Bxb4+ 41. Ka4 b5#
{Black mates} 0-1

[George Tsavdaris]

I played a game against FMax and won as i expected. I can mostly win against every other program in the 10x8 variants, especially Gothic Chess that i have enough experience. And this is due to hidden attacks that programs don't see.

Yet i have to note that Ed Trice is absolutely right. I played first time a Capablanca Chess game and i have to say that this game is awful! Gothic Chess is 5 million times better.
In Capablanca you have amazing difficulties to develop your pieces in the opening and you literally can't castle! If you decide to castle then you have to spend 6-7 moves maneuvering and improvising to be able to castle, but then your pieces are so bad developed that you are lost anyway! Not to mention that you have to protect right away the i-Pawn.

I should note also that strangely enough FMax when it has to play a forced move, it actually thinks! Why is that? Why it doesn't play immediately....?

[Event "Computer Chess Game"]
[Site "USER"]
[Date "2007.10.20"]
[Round "1"]
[White "Me"]
[Black "fmax4_8w"]
[Result "1-0"]
[TimeControl "40/600"]
[Variant "capablanca"]
[Annotator "1... +0.14"]

1. e4 h6 2. h3 c5 3. Bh2 e6 4. f4 d6 5. Nj3 i5 6. Cf2 j5 7. d3 i4 8. Ni1
Bf6 9. j3 i3 10. Bg3 Nc6 11. c3 h5 12. Cc2 h4 13. Bf2 Ae7 14. Bf3 Qd7 15.
Nd2 O-O-O 16. a4 g5 17. Nc4 gxf4 18. a5 Bg5 19. b4 Nh6 20. Ab3 d5 21. Bxc5
dxc4 22. Axc4 Ag6 23. b5 Ne5 24. a6 Nxc4 25. axb7+ Qxb7 26. Rxa7 Qxb5 27.
Ra8+ Kc7 28. Ca2 Ne3+ 29. Kg1 Rxa8 30. Cxa8+ Kc6 31. Ca7+ Kxc5 32. Cc8+ Qc6
33. d4+ Kb5 34. Qb1+ Ka5 35. Cxc6+ Ka4 36. Qa2+ Kb5 37. Cc5+ Kb6 38. Qa5#
{White mates} 1-0

After sacrificing a great amount of material(An Archbishop and Knight for some Pawns) i've built an attack and here i played the pretty Ca2! that wins....

[Uri Blass]

<snipped>

I should note also that strangely enough FMax when it has to play a forced move, it actually thinks! Why is that? Why it doesn't play immediately....?

I think that the reason is clear.
FMax is based on micromax and the idea is to have a small program
that is best relative to the size(adding characters to the program is justified only if the elo rating increase is bigger than the number of bits that you add).

Having code to play immediately when there is a forced move is going to increase the size too much.

Uri

[hgm]

OK i have finally been able to setup FMax to play in Winboard.
My main problem was the pieces Archbishop and Chancellor that didn't appear in Winboard. The only way i could find is the font solution of the new winboard. So i had to manually create a font with these 2 new pieces!

Indeed, Winboard_F is supposed to use the font solution for representing new pieces. It would be hard to do it any other way, as there are zillions of fairy pieces. I added 5 new piece types to the internal board representation, to be able to distinguish more than the 6 piece types occurring in FIDE Chess. But how the user uses those piece types is entirely upto him. In one game they might be Archbishop and Chancellor, in another game they might be Nightrider, Elephant or Commoner. And I usually find using the same display symbol for an Archbishop and an Elephant very confusing, even if they cannot appear in the same game.

So I see no other solution than to have the user define the 5 non-FIDE pieces (or as many of this as he needs) for every game separately. There are Chess fonts that support fairy pieces as rotated versions of the FIDE symbols, e.g. 'Good Companion'. I built in some support for this: if the user sets the font to GC2004D (and has this font installed!), the 5 new pieces are assigned the symbols of upside-down normal FIDE pieces. (I don't like that very much, as it is still confusing, as my eye/brain automatically seems to rotate the symbols back. But at least it makes them show up differently on careful examination.

I did not want to make font-based rendering the default, as I cannot be sure which fonts the user has installed (if any). But the WB native bitmap rendering only supports FIDE pieces, so I could do little else than mapping the new pieces onto the FIDE pieces. The idea is to make Winboard_F 100% upward compatible with the old Winboard for people that play FIDE 8x8 Chess (i.e. they must be able to do so without having fonts installed). This shifts somewhat more of a burdon on people that play variants, as they really have to think on how to represent the new pieces.

Great that you are able to design fonts. I have no idea how to do that. I assumed there would exist fonts for Capablanca / Gothic Chess. I hope you will make your font publically available as well.

[hgm]

I should note also that strangely enough FMax when it has to play a forced move, it actually thinks! Why is that? Why it doesn't play immediately....?

Yes, I never took the trouble to add special code to suppress that, even in Joker. The time gains are so minimal, that it might not even buy you 1 Elo point. Of course in micro-Max that means that I would only add it if I could implement it with less than 1 character. Which does not seem possible ever.

The purpose of the Fairy-Max project, however, is not to write the smallest or best/size Capablanca program. I developed it for being able to precisely determine piece values of arbitrary fairy piece, on various board sizes. I do this by playing unequal-material games, where, say, one of the sides has Nightriders in stead of Rooks (possibly giving Pawn odds if the difference turns out to be too large). Micro-Max seemed an ideal choice for this, as its algorithm is so simple that it basically has no Chess-knowledge at all. So the FIDE pieces do not have an unfair advantage over the fairy pieces I am testing because, say, they would have exquisitely tuned Piece-Square Tables which the fairies lack. The only thing uMax knows is that light pieces should draw toward the center, which seems just as applicable to fairies. So all pieces are compared on an equal basis.

Don't expect GM-level Chess from an engine that measures only about 100 lines of code. Even in FIDE Chess micro-Max is only rated slightly over 2000 Elo, which is already close to a miracle. And my first impression is that Capablance Chess is relatively more difficult to computers, because the new pieces are heavyweights that really crank up the branching factor, as does the bigger board.

[George Tsavdaris]

First: Congratulations on you for creating this flawless(at least in the 10 games i played so far) GUI for 10x8 Chess variants!
Also thanks for FMax too that seems to play with no apparent bugs. And despite its very very small size(code) it playes reasonably well....
And thanks to Matthias Gemuh also for the 2 new engines that support 10x8 Chess too(Capablanca variant).

I have played some games in 40/10. No problems found except the repetition claims i say below.

-ArcBishop 1.00b -Chancellor 1.00b 0-1

-Chancellor 1.00b - ArcBishop 1.00b 1-0
Game ended with a nice mate by the ArchBishop piece mating alone the King! Pretty to see such end

-ArcBishop 1.00b - FMax4_8w 1-1
In game one ArcBishop offered a pretty Queen sacrifice 11.d4! that showed the sacrificial nature of 10x8 Chess and the power of Archbishop and Chancellor pieces....
In game 2 FMax showed the power of Archbishop into a nice ending and forced the Arcbishop(the program) to give its Queen for the A, so from then on FMax won easily.

-FMax4_8w - ArcBishop 1.00b 1-1
The strange thing is that in the first game with 26.Rb1 the engines claimed and declared a draw by repetition. But there was no 3-fold repetition occured. Whose engine's bug is this? I guess Arcbishop's right?
Game 2 was more exciting and FMax was winning but then Arcbishop played the surprise and amazing 38...Cg6!! giving its Chancellor and forced white into a very strange draw trap that couldn't escape. And although white couldn't ascape from repeating moves and it would repeat anyway, the game again ended with a premature draw claim by repetition! Yet there was not 3-fold repetition again.
Here Arcbishop played 38...Cg6!! saving the game.


-FMax4_8w - Me 0-2
In the second game i played the terrific 27...Ne5!! and won. FMax replied correctly but i've seen deeper.
------------------------------------------------------------------------------

I assumed there would exist fonts for Capablanca / Gothic Chess. I hope you will make your font publically available as well.

I assumed the same but have been disappointed. I didn't find anything. I found some strange looking pieces(turned upside down) but i didn't like them at all.
I was interested only in Archbishop and Chancellor pieces. But didn't find anything as i've said even for these 2 pieces.
I also didn't know anything of how creating fonts but Google helped me.
So i created 2 sets of A+C.
One that i designed myself and one that i designed looking at the pieces from the program Gothic Vortex of Ed Trice. Perhaps Ed may confirm here that i did not violate any copyrights here if i allow this font for downloading.

So here it is:
Set-1(Chess10x8): The font-pieces of Chancellor and Archbishop as i have designed them, by looking at the pieces of Gothic Vortex.
This is by far the best set of the 2 i think, but i have to wait Ed to confirm i can offer them for downloading....


Set-2(Chess10x8n): The font-pieces i designed for Chancellor and Archbishop:


You can download set-2 (for set-1 i have to wait Ed to confirm) here:
SET-2 LINK


You can use them by adding this font into your computer.
I did this by just paste and copy the file Chess10x8n.ttf into C:\Windows\Fonts folder.

Then you have to add the following line to Winboard.INI:
/renderPiecesWithFont="Chess10x8"
or
/renderPiecesWithFont="Chess10x8n"
For using Set-1 or 2 respectively.

And then adding the following line to winboard.ini:
/fontPieceToCharTable="bcdefguhijklmnoupq"

And then it works.

++++++++++++++++++
The aforementioned games:

[Event "Computer Chess Game"]
[Site "????"]
[Date "2007.10.20"]
[Round "1"]
[White "George"]
[Black "fmax4_8w"]
[Result "1-0"]
[TimeControl "40/600"]
[Variant "capablanca"]
[Annotator "1... +0.14"]

1. e4 h6 2. h3 c5 3. Bh2 e6 4. f4 d6 5. Nj3 i5 6. Cf2 j5 7. d3 i4 8. Ni1
Bf6 9. j3 i3 10. Bg3 Nc6 11. c3 h5 12. Cc2 h4 13. Bf2 Ae7 14. Bf3 Qd7 15.
Nd2 O-O-O 16. a4 g5 17. Nc4 gxf4 18. a5 Bg5 19. b4 Nh6 20. Ab3 d5 21. Bxc5
dxc4 22. Axc4 Ag6 23. b5 Ne5 24. a6 Nxc4 25. axb7+ Qxb7 26. Rxa7 Qxb5 27.
Ra8+ Kc7 28. Ca2 Ne3+ 29. Kg1 Rxa8 30. Cxa8+ Kc6 31. Ca7+ Kxc5 32. Cc8+ Qc6
33. d4+ Kb5 34. Qb1+ Ka5 35. Cxc6+ Ka4 36. Qa2+ Kb5 37. Cc5+ Kb6 38. Qa5#
{White mates} 1-0

[Event "Computer Chess Game"]
[Site "????"]
[Date "2007.10.21"]
[Round "2"]
[White "fmax4_8w"]
[Black "George"]
[Result "0-1"]
[TimeControl "40/600"]
[Variant "capablanca"]
[Annotator "1. +0.04"]

1. e4 h6 2. Bg4 e6 3. Nc3 Bh7 4. f4 d6 5. d4 g6 6. a4 Ng7 7. j4 f5 8. Bf3
Cf7 9. exf5 Nxf5 10. d5 O-O 11. dxe6 Cf8 12. Bd5 c6 13. Bb3 Be7 14. g4 Ng7
15. Bd4 Rg8 16. Bxg7 Rxg7 17. h4 d5 18. Nxd5 cxd5 19. Bxd5 Nc6 20. Qc3 Bf6
21. Qi3 Bg8 22. g5 hxg5 23. hxg5 Rh7 24. Nh3 Bh8 25. Ch2 Bxe6 26. Bxe6 Axe6
27. O-O Ne5 28. Re1 Ng4 29. Cxg4 Axg4 30. Rxe8 Cxe8 31. Qg1 Rxh3 32. ixh3
Axh3+ 33. Kh2 Axj4+ 34. Kh1 Axg1 35. Kxg1 Ce1+ 36. Kf2 Cxc2+ 37. Kf3 Ce1+
38. Kg4 Cg1+ 39. Kh4 Ch1+ 40. Kg4 Ci3+ 41. Kh4 Cj3+ 42. Ki4 Ci1+ 43. Kh4
i5+ 44. Kg3 Bj6 45. Rb1 i4+ 46. Kh2 Ch3+ 47. Ki2 Cf2+ 48. Ki3 Bh4+ 49. Kxh4
Rh8+ 50. Kg3 Ch3+ 51. Kg2 Ci3+ 52. Kf1 Rh1+ 53. Ke2 Rh2+ 54. Ke1 Ci1#
{Black mates} 0-1

[Event "Computer Chess Game"]
[Site "????"]
[Date "2007.10.21"]
[Round "1"]
[White "ArcBishop 1.00b WB"]
[Black "fmax4_8w"]
[Result "1-0"]
[TimeControl "40/600"]
[Variant "capablanca"]
[Annotator "1. -0.42 1... +0.16"]

1. Ci3 Nh6 2. f3 i6 3. e4 e5 4. Ci5 j6 5. Ch3 d5 6. g4 dxe4 7. Qxe4 f5 8.
Bc5+ Kf7 9. Qd5+ Kg6 10. gxf5+ Kf6 11. d4 Bxd5 12. Ag5+ Kxf5 13. Cg3+ Kg6
14. Ae7+ Kf7 15. Cg5#
{ArcBishop says White Mates} 1-0

[Event "Computer Chess Game"]
[Site "????"]
[Date "2007.10.21"]
[Round "2"]
[White "ArcBishop 1.00b WB"]
[Black "fmax4_8w"]
[Result "0-1"]
[TimeControl "40/600"]
[Variant "capablanca"]
[Annotator "1. -0.31 1... +0.10"]

1. Nh3 e5 2. e4 f6 3. Ci3 i6 4. Nc3 d6 5. Ci5 h6 6. Cj3 i5 7. Ci1 i4 8. Nj4
f5 9. Bh5 Bf7 10. Bxf7 Kxf7 11. j3 ixj3 12. Cxj3 Nc6 13. Ae2 Nd4 14. Ni6
jxi6 15. Cxj8 Nxe2 16. Qxe2 fxe4 17. Qxe4 h5 18. O-O-O Bg5 19. Cj7 Cxj7 20.
Rxj7 c6 21. h4 Bf6 22. Qh7 Axi2 23. Qxi6 Axh4 24. Qxh5+ Ag6 25. Qi6 Qh8 26.
Qg4 Nh6 27. Qd7+ Ae7 28. Rj4 b6 29. Ne4 d5 30. Nxf6 gxf6 31. Rh4 Qj6 32. g3
a5 33. Qb7 Re8 34. Qxb6 Ad8 35. Qc5 d4 36. Re1 Qi6 37. c4 d3 38. Qa7+ Ke6
39. Rd4 Re7 40. Qb8 Ac7 41. Qb3 Aa6 42. Qxd3 Kf7 43. Rh4 Ab4 44. Qb3 c5 45.
Bj4 Nf5 46. Rhh1 a4 47. Qxb4 cxb4 48. b3 a3 49. Rh8 Qj6 50. Rf8+ Kg6 51.
Bi3 Re6 52. Rg8+ Kf7 53. Rb8 Qi6 54. Bg1 Nd4 55. Rb7+ Re7 56. Rxe7+ Kxe7
57. f3 Kf7 58. Bxd4 exd4 59. g4 d3 60. Rf1 Qi2 61. f4 Qxg4 62. c5 f5 63.
Ri1 Qg7 64. Ri7 Qxi7 65. c6 Qi1#
{Black mates} 0-1

[Event "Computer Chess Game"]
[Site "?????"]
[Date "2007.10.21"]
[Round "1"]
[White "fmax4_8w"]
[Black "ArcBishop 1.00b WB"]
[Result "1/2-1/2"]
[TimeControl "40/600"]
[Variant "capablanca"]
[Annotator "1. +0.12 1... +0.39"]

1. d4 Nh6 2. j4 Ci6 3. i3 d6 4. e4 Axj1 5. Bxi6 hxi6 6. Nh3 Ai2 7. Ci1 Axg1
8. Cxg1 e5 9. dxe5 Qxe5 10. f4 Qj5 11. c4 Nc6 12. Nc3 Bf6 13. Nd5 O-O-O 14.
Nxf6 gxf6 15. Qc3 Bh7 16. Ad2 Rje8 17. Nf2 f5 18. Qg7 Bg6 19. exf5 Nxf5 20.
Qc3 d5 21. Qc1 d4 22. Ch3 Ne3+ 23. Kg1 Nc2 24. Rb1 N2b4 25. Ra1 Nc2 26. Rb1
{Draw by repetition} 1/2-1/2

[Event "Computer Chess Game"]
[Site "????"]
[Date "2007.10.21"]
[Round "2"]
[White "fmax4_8w"]
[Black "ArcBishop 1.00b WB"]
[Result "1/2-1/2"]
[TimeControl "40/600"]
[Variant "capablanca"]
[Annotator "1. +0.13 1... +0.79"]

1. h3 Nc6 2. d4 Nh6 3. d5 Ne5 4. f4 Ng6 5. e4 Ad6 6. f5 Ne5 7. i4 j6 8. Nc3
e6 9. g4 exd5 10. g5 Ni8 11. Nxd5 Ac8 12. Bj4 i5 13. Bi3 c6 14. g6 Nh6 15.
gxf7 Bxf7 16. Nf4 Bxi3 17. jxi3 Bc4+ 18. Kg2 Qe7 19. b3 Bf7 20. Nh5 Bxh5
21. Bxh5 Nef7 22. c4 Qf6 23. Ad2 Ne5 24. a4 d6 25. a5 a6 26. h4 ixh4 27.
ixh4 Nd3 28. Qe2 Nc5 29. Nh3 Qb2 30. Cb1 Qe5 31. Af3 Qe7 32. Ag5 Qe5 33. b4
Nd7 34. Af4 Ae7 35. Ae6+ Qxe6 36. fxe6 Axh4+ 37. Kh2 Nf6 38. Qf1 Cg6 39.
Bxg6 Ng4+ 40. Kh1 Ag3+ 41. Kg1 Ai2+ 42. Kg2 Ah4+ 43. Kh1
{Draw by repetition} 1/2-1/2

[Event "Computer Chess Game"]
[Site "????"]
[Date "2007.10.22"]
[Round "1"]
[White "ArcBishop 1.00b WB"]
[Black "Chancellor 1.00b WB"]
[Result "0-1"]
[TimeControl "40/600"]
[Variant "capablanca"]
[Annotator "1. -0.42 1... +0.38"]

1. Ci3 i6 2. f3 e5 3. Ci5 j6 4. Cj3 d6 5. Nh3 i5 6. Ci1 i4 7. Nj4 j5 8. g4
Nh6 9. Nc3 Nc6 10. Nh5 f5 11. h3 Bf7 12. Ni3 j4 13. Ng2 ixh3 14. ixh3 Ri8
15. Qj6 Cg8 16. Ch1 Bj3 17. Cg3 f4 18. Ce4 Bi2 19. Ri1 Bxj2 20. Rxi2 Rxi2
21. Qxj4 Bd8 22. Nh4 Bxh4 23. Qxh4 Ae7 24. Qj6 Aj2 25. Qxj2 Rxj2 26. Cg5
Ci7 27. Ab3 Nd8 28. Rb1 Rj5 29. Ce4 Ci2 30. Ad4 exd4 31. Cxf4+ Qf7 32.
Cxf7+ Nhxf7 33. d3 Rj1 34. Ke1 Cxg1+ 35. Kd2 Cf1+ 36. Kc1 dxc3 37. bxc3
Cxd1+
{Chancellor says Black Mates} 0-1

[Event "Computer Chess Game"]
[Site "????"]
[Date "2007.10.22"]
[Round "1"]
[White "Chancellor 1.00b WB"]
[Black "ArcBishop 1.00b WB"]
[Result "1-0"]
[TimeControl "40/600"]
[Variant "capablanca"]
[Annotator "1. -0.48 1... +0.21"]

1. e4 Nh6 2. e5 j5 3. j4 d6 4. Nh3 dxe5 5. Qxe5 Nc6 6. Qe1 e5 7. g3 Bf6 8.
Nc3 Ad6 9. g4 Bh4 10. i3 Bj6 11. Ni5 Ci6 12. f3 Nd4 13. f4 f6 14. Be3 Qi4+
15. h3 Bxi5 16. jxi5 Qxi5 17. g5 Qxe1+ 18. Kxe1 exf4 19. gxh6 fxe3 20. Bxi6
Nxc2+ 21. Kd1 exd2 22. Axd2 gxh6 23. Kxc2 hxi6 24. Ne4 Ae7 25. Kb1 Bh7 26.
Ac3 Ac5 27. Ce1 Re8 28. Rj4 Af2 29. Cf3 Axe4+ 30. Rxe4 O-O 31. Cxf6 Ref8
32. Ce6 Kj7 33. Ce7 Bi8 34. Kc1 Bxe4 35. Cxe4 Rhg8 36. Ce1 Rg3 37. Ch1 Rff3
38. Ah8+ Kj6 39. Ad4 Rxh3 40. Cxh3 Rxh3 41. Af2+ Kj7 42. Axh3 b6 43. Ae6 i5
44. Axc7 j4 45. i4 Kj8 46. Kc2 Kj7 47. Rd1 Kj6 48. Ae8 Kj7 49. Ah5+ i6 50.
Rd7+ Ki8 51. Ag6+ Kj8 52. Rd8+ Kj7 53. Ah8+ Ki8 54. Axi6+ Ki7 55. Aj5+ Kj7
56. Rj8+ Kxj8 57. Axh6#
{Chancellor says White Mates} 1-0

[hgm]

The strange thing is that in the first game with 26.Rb1 the engines claimed and declared a draw by repetition. But there was no 3-fold repetition occured. Whose engine's bug is this?

It might actually be Fairy-Max', as I sometimes set that to claim a draw on the first repetition for self play. (To save time, knowing that it plays reproducibly anywaay.) I might accidentally have released a version where this is the case. I will look into it.