The game of chess needs 2 major changes.......

[lkaufman]

But if a rule change is contemplated, my suggestion is to permit only opposite side castling. So if your opponent castles first, you have only one side available for castling. This does not make the game unrecognisably different to be considered a variant.

This is actually a great rule if combined with Armageddon rule. The point is that it greatly increases white's advantage, he can castle short first and force Black to aim for long castling, which takes longer and even if achieved a draw isn't so likely. White's winning chances should be close to fifty percent. You can roughly simulate it by setting castling rights as short for White and long for Black.

I will try to simulate this on the weekend. But now that I think about it, it seems somewhat like simply using unbalanced openings.

Well, unbalanced openings are used to provide variety as well as avoid draws. This rule doesn't help with variety, but it has the merit of having clear, simple rules that are fully within the spirit of normal chess. This is more likely to catch on with human players, since it doesn't involve some arbitrary set of starting positions. But we would still have the problem of computer analysis making it a memory contest. My guess is that Black will score a bit more than White with the Armageddon rule, but it should be close and I could be wrong.

[chesskobra]

I am now running a match between Stockfish 16.1 and Berserk 13, and so far no draws. The starting position is

Code: Select all

rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w Kq - 0 1

i.e., white can only castle short, black can only castle long. So far no draws. Mostly white wins, but as I reported in another thread, Berserk lost many games with white in a good position with the error

Code: Select all

{polyglot: resign (illegal engine move by white: c4d5)}

which is kind of spoiling the fun. Stockfish has won all its white games. Tomorrow I will post the results, and I will also do a match from the position

Code: Select all

rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w Qk - 0 1

Incidentally, cutechess gui ignores the castling restriction. Luckily I am able to run the match in xboard, which always works as claimed. One of the best programs along with pgn-extract. Top demais!

[lkaufman]

I am now running a match between Stockfish 16.1 and Berserk 13, and so far no draws. The starting position is

Code: Select all

rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w Kq - 0 1

i.e., white can only castle short, black can only castle long. So far no draws. Mostly white wins, but as I reported in another thread, Berserk lost many games with white in a good position with the error

Code: Select all

{polyglot: resign (illegal engine move by white: c4d5)}

which is kind of spoiling the fun. Stockfish has won all its white games. Tomorrow I will post the results, and I will also do a match from the position

Code: Select all

rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w Qk - 0 1

Incidentally, cutechess gui ignores the castling restriction. Luckily I am able to run the match in xboard, which always works as claimed. One of the best programs along with pgn-extract. Top demais!

With cutechess or Chessbase GUIs, one workaround is to start the game with a move sequence that forfeits castling for the proper sides. It would be more informative if the two sides were evenly matched, for example SF vs Leela if you have a good GPU or else give Berserk double or perhaps triple time odds to make it close. If your goal is to reduce but not eliminate draws, then instead of Armageddon you can have the castling rights reversed, i.e. White can only castle long and Black can only castle short. In this version, White and Black chances are roughly equalized (at least Black's edge should be less than White's edge in normal chess), so draws are reduced just by the opposite castling and not by giving one side a large advantage and the other draw odds. The corresponding rule in normal chess would be that you can't castle on the same side as the opponent, but White can't castle before Black or before move 6 (or whatever number equalizes the chances).

[pohl4711]

I am now running a match between Stockfish 16.1 and Berserk 13, and so far no draws. The starting position is

Code: Select all

rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w Kq - 0 1

i.e., white can only castle short, black can only castle long. So far no draws. Mostly white wins, but as I reported in another thread, Berserk lost many games with white in a good position with the error

Code: Select all

{polyglot: resign (illegal engine move by white: c4d5)}

which is kind of spoiling the fun. Stockfish has won all its white games. Tomorrow I will post the results, and I will also do a match from the position

Code: Select all

rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w Qk - 0 1

Incidentally, cutechess gui ignores the castling restriction. Luckily I am able to run the match in xboard, which always works as claimed. One of the best programs along with pgn-extract. Top demais!

Have you considered my AntiDraw-openings download?
https://www.sp-cc.de/anti-draw-openings.htm
https://www.sp-cc.de/files/antidraw_v2.1.7z

In there, you find (besides a lot of other antidraw-concepts) NBSC/NBC openings (=No Black Short Castling V2, No Black Castling V2): Openings, where black is not allowed to castle short (NBSC) or not allowed to castle long or short (NBC).
From the ReadMe (about NBSC):

"There are 3 pgn/epd opening sets in this download-package:
NBSC_Noomen_3mvs: Built from the Noomen 3mvs opening set (deleted double endpositions and opening-lines, which are not good for computerchess IMHO (Dutch, King's gambit etc). 250 lines remaining.
NBSC_V2_4mvs: 12910 lines. Out of the Megabase 2020. All endpositions calculated by KomodoDragon 1.0 (Quadcore, 15''/move) All endpositions in an eval interval of [-0.29/+0.49] (before the NBSC-line was added, of course).
NBSC_V2_6mvs: 46582 lines. Out of the Megabase 2020. All endpositions calculated by KomodoDragon 1.0 (Quadcore, 15''/move) All endpositions in an eval interval of [-0.29/+0.49] (before the NBSC-line was added, of course)."

[jefk]

Ras wrote

Pickup Go or something

actually not such a bad idea, searched for correspondence Go,
and it can be played here
https://online-go.com/
But use of Ai/comps isn't allowed so this isn't what i'm looking for.
As for Hex, as chesskobra suggested, this is an interesting game in game theory
(network theory with topology.
https://arxiv.org/pdf/2102.02621
My tentative proof with network theory which i'm exploring for checkers
is a bit more simple, and when i'm done i'll extend it for draughts and chess.
As for Hex in practical play, there's a strong advantage for the first
mover so you need to play two games, play online here:
https://lazymuttgames.com/games/Hex/
or here
https://playhex.org/
doesn't seem to be so serious so i presume some 'cheating'
(computer use) is allowed.
Or otherwise infinite Hex may be more interesting, claimed to solved (draw):
https://www.semanticscholar.org/paper/I ... 0333417c9e
Good luck in proving that it's a draw with number crunching and tablebases.

My heart still is in conventional chess, it's the most popular boad game nowadays with
chess.com and all the twitch streaming and playing rapid on Lichess keeps me
sharp although i hit a plateau i admit as human who started late in life not
being in the near-top like in correspondence chess. Meanwhile i did register
again for a three phase ICCF tourn, mainly out of curiosity :
https://iccf.com/EventAnnouncement.aspx?id=105583
In the final there are (dollar) prices, but curiosity is to see if initially i still can
beat the newbies in the preliminary rounds (guess i shouldn't
tell all my secrets about openings and so on here anymore

The chess variant with opposite castling may be interesting but won't
be accepted any time soon by the ICCF i expect (just like Chess960
took a long time, and still isn't as popular as normal chess);
https://iccf.com/EventAnnouncement.aspx?id=25888
NB if it's so easy to test the opposite castling variant with Cutechess
then i presume some of the other proposals (Nickel, Kaufman) etc.
also can be tested (with an SF variant or so).

[chesskobra]

Have you considered my AntiDraw-openings download?
https://www.sp-cc.de/anti-draw-openings.htm
https://www.sp-cc.de/files/antidraw_v2.1.7z

Thanks Stefan, I will take a look at these openings. But I would really like to test "no same side castling", which will require a small code change in some engines.

[chesskobra]

With cutechess or Chessbase GUIs, one workaround is to start the game with a move sequence that forfeits castling for the proper sides.

This won't work because that would mean that the engines started with normal chess in mind and at some point we interfered and said now no more castling on this side. But as it is if an epd is given, I don't know why cutechess does not respect the castling rights. (But I will investigate more before claiming that there is a bug in cutechess.)

I ran the test with castling right Kq with Stockfish 16.1 vs Obsidian dev-12.15, which SF won 14.5-4.5 (W-D-L = 10-9-1, with one loss of SF coming from black side and all 10 wins from white side). So this was pretty one-sided. On the other hand, in SF vs SF it just seems to repeat the game every time. I now plan to run a test with castling right Qk. But ideally testing "no same side castling" by code change would be the most interesting.

Obsidian losing so badly itself is a big surprise to me. Sometimes it seems that many top engines have flaws. They are mainly good at middlegame tactics. In unclear, balanced positions many engines may be clueless. They are not universal players so to speak. Apologies for these sweeping statements. (On a side note, I ran an endgame round robin of 16 engines without tablebases, and Stockfish 16.1 and Berserk 13 are significantly ahead. I don't have Komodo or Torch, though. It seems that many engines are not trained on endgames. If they want to make progress, they should improve their endgame play.)

[chesskobra]

As for Hex, as chesskobra suggested, this is an interesting game in game theory
(network theory with topology.
https://arxiv.org/pdf/2102.02621

Thanks for this reference.

[jefk]

ck wrote

Thanks for this reference.

ok but i actually (related to game theory) meant this reference:
https://math.uchicago.edu/~may/REU2019/ ... achner.pdf

but the game of Y also seems interesting, you can play it online, offline engine is here
https://www.zillionsofgames.com/cgi-bin ... how;id=959

[Chessqueen]

Have you considered my AntiDraw-openings download?
https://www.sp-cc.de/anti-draw-openings.htm
https://www.sp-cc.de/files/antidraw_v2.1.7z

Thanks Stefan, I will take a look at these openings. But I would really like to test "no same side castling", which will require a small code change in some engines.

There is the Cemenz Opening which does NOT immediately benefit the White side but give the Black side a chance to survive.
OOPs Sorry this is one of the most drawish Opening the sucure a draw with Black

[pgn][Date "2024.05.24"]
[Round "?"]
[White "Stockfish_24052319_x64_avx2"]
[Black "Stockfish_24052319_x64_avx2"]
[Result "*"]
[BlackElo "3655"]
[ECO "A00"]
[Opening "Clemenz (Mead-Basman-de Klerk-Welling) Opening"]
[Time "09:23:47"]
[WhiteElo "3655"]
[TimeControl "1800+5"]
[Termination "unterminated"]
[PlyCount "8"]
[WhiteType "program"]
[BlackType "program"]


1. h3 e5 2. e4 Nf6 3. Nc3 Bb4 4. Nge2 O-O {(O-O g2-g4 d7-d5 e4xd5 Nf6xd5
Bf1-g2 Bc8-e6 O-O Nb8-c6 d2-d4 e5xd4 Nc3-b5 Qd8-f6 Ne2xd4 Ra8-d8 c2-c3
Bb4-c5 Nd4xc6 b7xc6 Nb5-d4 Nd5-b4 Qd1-a4 Be6-c4 Rf1-d1 Bc4-d5 Bc1-e3 Bd5xg2
Kg1xg2 Nb4-d5 Rd1-e1 Bc5-b6 Re1-e2 Bb6xd4 Qa4xd4) +0.35/37 36} 5. g3
{(g2-g3 d7-d5 e4xd5 Nf6xd5 Bf1-g2 Bc8-e6 O-O Nb8-c6 d2-d4 e5xd4 Nc3-b5
Qd8-d7 Kg1-h2 Ra8-d8 Nb5xd4 Nc6xd4 Qd1xd4 Qd7-e7 Ne2-f4 Nd5xf4 Qd4xf4 b7-b6
Bc1-e3 a7-a5 a2-a3 Bb4-d6 Qf4-f3 Bd6-c5 Rf1-e1 Bc5xe3 Re1xe3 Qe7-c5 c2-c3
Qc5-b5 b2-b4 Qb5-a4 h3-h4 Rd8-d2 h4-h5) -0.11/41 86} d5 {(d7-d5 e4xd5
Nf6xd5 Bf1-g2 Bc8-e6 O-O Nb8-c6 d2-d4 e5xd4 Nc3-b5 Qd8-d7 Kg1-h2 Ra8-d8
Nb5xd4 Nc6xd4 Qd1xd4 Qd7-e7 Ne2-f4 Nd5xf4 Qd4xf4 Bb4-d6 Qf4-g5 b7-b6 Bc1-e3
h7-h6 Qg5xe7 Bd6xe7 b2-b3 Be7-g5 Be3xg5 h6xg5 Ra1-d1 g7-g6 c2-c4 Kg8-g7
Bg2-d5 Be6xd5 Rd1xd5 Rd8xd5 c4xd5 Rf8-d8 Rf1-c1 Rd8xd5 Rc1xc7 Rd5-d2 Rc7xa7
Rd2xf2+ Kh2-h1 Rf2-d2) +0.12/39 43} 6. exd5 {(e4xd5 Nf6xd5 Bf1-g2 Bc8-e6
O-O Nb8-c6 d2-d4 e5xd4 Nc3-b5 Qd8-d7 Kg1-h2 Ra8-d8 Nb5xd4 Nc6xd4 Qd1xd4
Qd7-e7 Ne2-f4 Nd5xf4 Qd4xf4 b7-b6 Bc1-e3 Bb4-c5 b2-b3 Bc5xe3 Qf4xe3 Qe7-f6
Ra1-d1 h7-h5 Qe3-f3 Be6-f5 c2-c4 g7-g6 Qf3-c6 Qf6xc6 Bg2xc6 a7-a5 f2-f4
Kg8-g7 Kh2-g2 Rf8-h8 Rf1-e1 Kg7-f6) -0.11/41 64} Nxd5 {(Nf6xd5 Bf1-g2
Bc8-e6 O-O Nb8-c6 d2-d4 e5xd4 Nc3-b5 Qd8-d7 Kg1-h2 Ra8-d8 Nb5xd4 Nc6xd4
Qd1xd4 Qd7-e7 Ne2-f4 Nd5xf4 Qd4xf4 b7-b6 Bc1-e3 Bb4-c5 Ra1-d1 Bc5xe3 Qf4xe3
Qe7-f6 b2-b3 g7-g6 g3-g4 Rf8-e8 Rf1-e1 Qf6-b2 Rd1xd8 Re8xd8 Bg2-e4 Rd8-d7
Kh2-g3 Qb2xa2) +0.14/38 21} 7. Bg2 {(Bf1-g2 Bc8-e6 O-O Nb8-c6 d2-d4 e5xd4
Nc3-b5 Qd8-d7 Kg1-h2 Ra8-d8 Nb5xd4 Nc6xd4 Qd1xd4 Rf8-e8 Ne2-f4 Nd5xf4
Qd4xd7 Rd8xd7 Bc1xf4 Bb4-d2 Bg2xb7 Bd2xf4 g3xf4 Re8-b8 Bb7-e4 Rb8xb2 f4-f5
Be6xa2 Ra1-d1 Rd7xd1 Rf1xd1 g7-g6 f5-f6 h7-h5 Rd1-d7 Rb2-b4 f2-f3 h5-h4
Rd7xc7 a7-a6 Rc7-c6 Rb4-a4 Kh2-g2 a6-a5 Rc6-c5 Ba2-e6 Be4-d5 Ra4-a3 Bd5xe6
f7xe6 Rc5-c4 a5-a4 Rc4xh4 Kg8-f7 Rh4-b4) -0.07/41 58} Be6 {(Bc8-e6 O-O
Nb8-c6 d2-d4 e5xd4 Nc3-b5 Qd8-d7 Kg1-h2 Ra8-d8 Nb5xd4 Nc6xd4 Qd1xd4 Qd7-e7
Ne2-f4 Nd5xf4 Qd4xf4 Bb4-d6 Qf4-g5 f7-f6 Qg5-h5 b7-b6 Bc1-e3 Bd6-e5 c2-c3
Qe7-f7 Qh5xf7+ Kg8xf7 h3-h4 h7-h6 a2-a4 g7-g5 h4xg5 h6xg5 a4-a5 Rd8-d7
Bg2-f3) +0.11/38 32} 8. O-O {(O-O Nb8-c6 d2-d4 e5xd4 Nc3-b5 Qd8-d7 Kg1-h2
Ra8-d8 Nb5xd4 Nc6xd4 Qd1xd4 Qd7-e7 Ne2-f4 Nd5xf4 Qd4xf4 Bb4-d6 Qf4-g5 f7-f6
Qg5-h5 b7-b6 Bc1-e3 Qe7-f7 Qh5xf7+ Kg8xf7 Ra1-d1 g7-g5 Bg2-f3 Kf7-g6 Kh2-g2
h7-h5 Bf3-c6 Rf8-h8 Be3-d4 g5-g4 h3-h4 Be6-c4 Rf1-g1 a7-a5) -0.03/40 53}
Nc6 {(Nb8-c6 d2-d4 e5xd4 Nc3-b5 Qd8-d7 Kg1-h2 Ra8-d8 Nb5xd4 Nc6xd4 Qd1xd4
Rf8-e8 Ne2-f4 Nd5xf4 Qd4xd7 Rd8xd7 Bc1xf4 Bb4-d2 Bg2xb7 Bd2xf4 g3xf4 Re8-b8
Bb7-e4 Rb8xb2 f4-f5 Be6xa2 Rf1-d1 Rd7xd1 Ra1xd1 g7-g6 f5-f6 h7-h5 Rd1-d7
Rb2-b4 f2-f3 h5-h4 Rd7xc7 Rb4-a4 Rc7-c5 Ba2-c4 Kh2-g2 Bc4-e6 Rc5-e5 Ra4-a2
Be4-d3 a7-a5 Re5-e4 a5-a4) +0.15/38 26} 9. d4 {(d2-d4 e5xd4 Nc3-b5 Qd8-d7
Kg1-h2 Ra8-d8 Nb5xd4 Nc6xd4 Qd1xd4 Qd7-e7 Ne2-f4 Nd5xf4 Qd4xf4 Bb4-d6
Qf4-g5 f7-f6 Qg5-h5 Be6-f7 Qh5-f3 b7-b6 Bc1-e3 Qe7-e5 b2-b3 Qe5-b2 c2-c4
Rf8-e8 Rf1-b1 Qb2-c3 Rb1-c1 Qc3-b2) 0.00/40 57} exd4 {(e5xd4 Nc3-b5 Qd8-d7
Kg1-h2 Ra8-d8 Nb5xd4 Nc6xd4 Qd1xd4 Qd7-e7 Ne2-f4 Nd5xf4 Qd4xf4 Bb4-d6
Qf4-g5 f7-f6 Qg5-h5 Be6-f7 Qh5-f5 b7-b6 Bc1-e3 Qe7-e5 Qf5xe5 Bd6xe5 c2-c3
g7-g5 Rf1-e1 Rf8-e8 Bg2-f3 a7-a5 Kh2-g2 a5-a4 a2-a3 Bf7-d5 Ra1-d1 Bd5xf3+
Kg2xf3 Kg8-f7 Rd1xd8 Re8xd8 Be3xg5 Be5xc3 b2xc3 f6xg5 Re1-e3 Rd8-d2 Kf3-g4)
+0.06/38 33} 10. Nb5 {(Nc3-b5 Qd8-d7 Kg1-h2 Ra8-d8 Nb5xd4 Nc6xd4 Qd1xd4
c7-c6 Bg2xd5 Qd7-e7 Bc1-g5 Qe7xg5 Ne2-f4 Be6xd5 Qd4xb4 b7-b6 Qb4-c3 c6-c5
h3-h4 Qg5-f5 Nf4xd5 Qf5xd5 Rf1-e1 Qd5-d2 Kh2-g2 Qd2xc3 b2xc3 g7-g6 Re1-d1
Rd8-e8 Kg2-f1 Re8-e4 a2-a4 a7-a5 Rd1-d6 Rf8-e8 Rd6xb6 Re4-e2 Rb6-b5 Re2xc2
Rb5xa5 Re8-e2) -0.01/36 27} Qd7 {(Qd8-d7 Kg1-h2 Ra8-d8 Nb5xd4 Nc6xd4 Qd1xd4
Qd7-e7 Ne2-f4 Nd5xf4 Qd4xf4 Bb4-d6 Qf4-g5 f7-f6 Qg5-h5 b7-b6 Bc1-e3 Be6-f7
Qh5-f5 Qe7-e5 Qf5xe5 Bd6xe5 c2-c3 g7-g5 Bg2-c6 h7-h5 f2-f4 g5xf4 g3xf4
Be5-d6 Rf1-g1+ Kg8-h7 Kh2-g2 Rf8-g8+ Kg2-f2 Bf7-e6 Bc6-e4+ f6-f5 Be4-c6
Bd6-e7 b2-b3 Rg8xg1 Ra1xg1 Be7-h4+ Kf2-e2 Bh4-f6) +0.12/41 42} 11. Kh2
{(Kg1-h2 Ra8-d8 Nb5xd4 Nc6xd4 Qd1xd4 Qd7-e7 Ne2-f4 Nd5xf4 Qd4xf4 b7-b6
Bc1-e3 Bb4-c5 Ra1-d1 Bc5xe3 Qf4xe3 h7-h5 b2-b3 g7-g6 Qe3-e5 Rd8xd1 Rf1xd1
Rf8-d8 Rd1xd8+ Qe7xd8 Bg2-e4 Qd8-d2 Kh2-g2 Qd2-e1 Qe5-d4 c7-c5 Qd4-e5
Qe1-e2 h3-h4 c5-c4 Qe5-d4 Kg8-h7 Be4-d5 c4xb3 a2xb3 Qe2xc2) -0.05/39 61}
Rad8 {(Ra8-d8 Nb5xd4 Nc6xd4 Qd1xd4 Qd7-e7 Ne2-f4 Nd5xf4 Qd4xf4 Bb4-d6
Qf4-g5 b7-b6 Bc1-e3 Qe7xg5 Be3xg5 f7-f6 Bg5-e3 a7-a5 Rf1-e1 Kg8-f7 b2-b3
Bd6-b4 Re1-d1 Bb4-c3 Ra1-c1 Bc3-b2 Rc1-b1 Bb2-e5 Bg2-e4 g7-g5 f2-f4 g5xf4
Be3xf4 h7-h5 a2-a3 Be5xf4 g3xf4) +0.04/37 35} 12. Nbxd4 {(Nb5xd4 Nc6xd4
Qd1xd4 Rf8-e8 Ne2-f4 Nd5xf4 Qd4xd7 Rd8xd7 Bc1xf4 Bb4-d2 Bg2xb7 Bd2xf4 g3xf4
Re8-b8 Bb7-e4 Rb8xb2 f4-f5 Be6xa2 Rf1-d1 Rd7xd1 Ra1xd1 g7-g6 f5-f6 h7-h5
Rd1-d8+ Kg8-h7 Rd8-a8 Rb2-b4 f2-f3 Rb4-a4 Kh2-g3 Ba2-e6 h3-h4 Ra4-a1 Kg3-f4
a7-a5 Be4-c6 Ra1-h1 Kf4-g5 Rh1-g1+ Kg5-f4) 0.00/37 31} Nxd4 {(Nc6xd4 Qd1xd4
Qd7-e7 Ne2-f4 Nd5xf4 Qd4xf4 b7-b6 Bc1-e3 Bb4-d6 Qf4-g5 Qe7xg5 Be3xg5 f7-f6
Bg5-e3 g7-g5 Bg2-f3 Kg8-g7 Ra1-d1 h7-h5 Kh2-g2 Rf8-h8 Rf1-e1 Be6-d7 c2-c4
g5-g4 h3xg4 h5xg4 Bf3-d5 Kg7-g6 b2-b3 Bd7-f5 Be3-f4 Bd6xf4 g3xf4) +0.14/36
25} 13. Qxd4 {(Qd1xd4 Rf8-e8 Ne2-f4 Nd5xf4 Qd4xd7 Rd8xd7 Bc1xf4 Bb4-d2
Bg2xb7 Bd2xf4 g3xf4 Re8-b8 Bb7-e4 Rb8xb2 f4-f5 Be6xa2 Rf1-d1 Rd7xd1 Ra1xd1
g7-g6 f5-f6 h7-h5 Rd1-d8+ Kg8-h7 Rd8-a8 h5-h4 Ra8xa7 Kh7-h6 Ra7xc7 Rb2-b4
Be4-f3 Rb4-f4 Kh2-g2 Kh6-g5 Rc7-c5+ Rf4-f5 Rc5-c7 Rf5-f4) 0.00/39 46} Qe7
{(Qd7-e7 Ne2-f4 Nd5xf4 Qd4xf4 b7-b6 Bc1-e3 Bb4-c5 Ra1-e1 Bc5xe3 Re1xe3
c7-c5 b2-b3 h7-h6 Rf1-e1 Rd8-d4 Re3-e4 Rf8-d8 Bg2-f1 Qe7-d6 Qf4xd6 Rd8xd6
Bf1-d3 Rd4xe4 Re1xe4 Kg8-f8 g3-g4 g7-g5 f2-f4 Be6-d5 Re4-e5 f7-f6 Re5-f5
Kf8-e7 f4xg5 h6xg5 Kh2-g3 Bd5-c6 Rf5-f2 Rd6-e6) +0.13/42 154} 14. Nf4
{(Ne2-f4 Nd5xf4 Qd4xf4 Bb4-d6 Qf4-g5 f7-f6 Qg5-h5 Bd6-e5 c2-c3 Be6-c4
Rf1-h1 Bc4-e2 Qh5-f5 Rf8-e8 Qf5-c2 Qe7-c5 Bc1-e3 Be5xg3+ Kh2xg3 Re8xe3+
f2xe3 Qc5xe3+ Kg3-h2 Qe3-f4+ Kh2-g1 Rd8-d1+ Ra1xd1 Qf4-e3+ Kg1-h2 Qe3-e5+
Kh2-g1 Qe5-e3+) 0.00/37 38} Bc5 {(Bb4-c5 Qd4-e4 Nd5xf4 Bc1xf4 b7-b6 b2-b3
g7-g5 Bf4-e3 Bc5xe3 Qe4xe3 Qe7-f6 Ra1-e1 Be6-f5 Qe3-e5 Qf6xe5 Re1xe5 Bf5xc2
Re5xg5+ Bc2-g6 f2-f4 f7-f6 Rg5-d5 Bg6-f7 Rd5-d1 c7-c5 Bg2-c6 c5-c4 b3xc4
Bf7xc4 Rd1xd8 Rf8xd8 Rf1-f2 h7-h6 Rf2-c2 Bc4-e6) +0.06/43 75} 15. Qe4
{(Qd4-e4 Nd5xf4 Bc1xf4 c7-c6 Ra1-e1 Rf8-e8 Qe4-f3 h7-h6 Qf3-h5 Bc5-d4 b2-b3
Qe7-c5 Qh5xc5 Bd4xc5 Re1-e2 Be6-f5 Rf1-e1 Kg8-f8 Bg2-f3 g7-g5 Bf4-c7 Re8xe2
Re1xe2 Rd8-e8 Re2-d2 Bc5-b4 Rd2-d4 Bb4-e1 Kh2-g2 Bf5xc2 Rd4-d7 Be1-b4
Bc7-d8 Bb4-c3 Bd8-c7 Kf8-g7 Bc7-d6 Re8-e6 Bd6-c5 b7-b6 Bc5-e3) -0.04/38 18}
Nxf4 {(Nd5xf4 Bc1xf4 c7-c6 Ra1-e1 Rf8-e8 Qe4-f3 h7-h6 Qf3-h5 Bc5-d4 b2-b3
Qe7-c5 Qh5xc5 Bd4xc5 Re1-e2 Be6-f5 Rf1-e1 Kg8-f8 Bg2-f3 g7-g5 Bf4-c7 Re8xe2
Re1xe2 Rd8-e8 Re2-d2 Re8-e1 c2-c3 Re1-c1 Bc7-e5 Rc1-c2 Rd2xc2 Bf5xc2 Kh2-g2
a7-a5 Be5-c7 a5-a4 b3xa4 Bc2xa4 Bf3-e4 Ba4-b5 Be4-c2 Bb5-c4 Bc2-b3 Bc4-e2
g3-g4 Be2-d3) +0.09/35 22} 16. Bxf4 {(Bc1xf4 c7-c6 Ra1-e1 Rf8-e8 Qe4-f3
h7-h6 b2-b3 Qe7-f6 Re1-e2 Be6-d5 Qf3-g4 Bd5-e6) 0.00/50 25} c6 {(c7-c6
Ra1-e1 Rf8-e8 Qe4-f3 h7-h6 Re1-e2 Qe7-f6 b2-b3 Re8-e7 Qf3-h5 Qf6-f5 Bg2-f3
Qf5xh5 Bf3xh5 Kg8-f8 h3-h4 a7-a5 Bh5-f3 Be6-f5 Rf1-e1 a5-a4 Re2xe7 Bc5xe7
Re1-e2 a4-a3 Bf3-e4 Bf5xe4 Re2xe4 Rd8-d1 Re4-a4 f7-f5 Ra4-a8+ Kf8-f7 Ra8-a7
Rd1-a1 Ra7xb7 Kf7-e6 Rb7-c7 Ra1xa2 Rc7xc6+ Ke6-d7) +0.08/38 35} 17. Rae1
{(Ra1-e1 Rf8-e8 Qe4-f3 h7-h6 Qf3-h5 Bc5-b6 b2-b3 Qe7-f6 Re1-e2 Be6-f5
Rf1-e1 Re8-e6 Re2xe6 Bf5xe6 Re1-e2 Qf6-a1 c2-c4 Qa1-d1 Qh5-e5 g7-g5 Bf4xg5
Bb6-d4 Qe5-a5 Bd4-b6) 0.00/40 79} Rfe8 {(Rf8-e8 Qe4-f3 h7-h6 Qf3-h5 Bc5-b6
b2-b3 Qe7-f6 Re1-e2 Be6-f5 Rf1-e1 Re8-e6 Re2xe6 Bf5xe6 Re1-e2 Qf6-a1 c2-c4
Qa1-d1 Bg2-f3 Qd1-f1 Bf3-g2 Qf1-d1) 0.00/45 23} 18. Qf3 {(Qe4-f3 h7-h6
Qf3-h5 a7-a5 a2-a4 Bc5-d4 b2-b3 Qe7-c5 Qh5xc5 Bd4xc5 g3-g4 g7-g5 Bf4-c1
f7-f5 Bg2-f3 f5-f4 h3-h4 Bc5-b4 Re1-d1 Be6-f7 h4-h5 c6-c5 Bc1-b2 c5-c4
b3xc4 Bf7xc4 Bb2-f6 Rd8xd1 Rf1xd1 Bc4-e2 Bf3xe2 Re8xe2 Rd1-d8+ Kg8-f7
Bf6-d4 Re2xc2 Rd8-h8 Rc2-d2 Rh8-h7+ Kf7-e8 Rh7-h8+ Ke8-d7 Rh8-h7+) 0.00/39
28} h6 {(h7-h6 Qf3-h5 Bc5-b6 b2-b3 Qe7-f6 Re1-e2 Be6-f5 Rf1-e1 Re8-e6
Re2xe6 Bf5xe6 Re1-e2 Qf6-a1 h3-h4 Qa1xa2 Bf4xh6 g7xh6 Re2xe6 f7xe6 Qh5-g6+
Kg8-f8 Qg6-f6+ Kf8-e8 Qf6xe6+ Ke8-f8 Qe6xh6+ Kf8-e7 Qh6-g5+ Ke7-f7 Qg5-f4+
Kf7-e8 Qf4-e4+ Ke8-f8 Qe4-f5+ Kf8-e7 Qf5-e5+ Ke7-f7) 0.00/48 31} 19. b3
{(b2-b3 Qe7-f6 Re1-e2 Be6-d5 Qf3-g4 Re8xe2 Qg4xe2 Bd5-e6 h3-h4 Rd8-e8
Rf1-e1 Be6-d7 Qe2-d2 Re8xe1 Qd2xe1 g7-g5 h4xg5 h6xg5 Bf4-e3 Qf6-e6 Qe1-a5
Bc5xe3 f2xe3 Qe6-g4 Qa5xa7 Qg4-h5+ Kh2-g1 Qh5-d1+ Kg1-f2 Qd1xc2+ Kf2-g1
Qc2-b1+ Kg1-f2 Bd7-e6 Qa7xb7 Qb1-c2+ Kf2-g1 Qc2-d1+ Kg1-h2 Qd1-h5+ Kh2-g1
Qh5-d1+) 0.00/48 123} Qf6 {(Qe7-f6 Re1-e2 Be6-d5 Qf3-g4 Re8xe2 Qg4xe2
Bd5-e6 h3-h4 Bc5-b6 Rf1-e1 Bb6-a5 Re1-d1 Rd8xd1 Qe2xd1 Ba5-b6 Qd1-d6 Bb6xf2
Bg2-e4 Bf2-b6 Qd6-b8+ Bb6-d8 Bf4-e3 Be6-d5 Be4xd5 c6xd5 Be3xa7 b7-b6 Kh2-g2
Kg8-h7 a2-a4 d5-d4 Qb8-c8 Qf6-e5 Qc8xd8 Qe5-e2+ Kg2-h3 Qe2-f1+ Kh3-h2
Qf1-f2+ Kh2-h1 Qf2-f3+ Kh1-h2) 0.00/42 22} *[/pgn]