opposite castling armageddon

Discussion of anything and everything relating to chess playing software and machines.

Moderators: Harvey Williamson, bob, hgm

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
lkaufman
Posts: 4339
Joined: Sun Jan 10, 2010 5:15 am
Location: Maryland USA
Contact:

Re: opposite castling armageddon

Post by lkaufman » Fri Jul 26, 2019 5:27 pm

Guenther wrote:
Fri Jul 26, 2019 4:30 pm
OneTrickPony wrote:
Fri Jul 26, 2019 12:01 pm
What about: white gets to make 2 first moves and gives draw odds.
Lc0 eval is similar to opposite castling armageddon proposed in OP but it's even simpler to explain.
I did a quick test with CFish20190717 vs. itself and surprisingly the draw rate was still very high > 70% at a fast tc of 40/40''.
used this set of positions:

Code: Select all

rnbqkbnr/pppppppp/8/8/3PP3/8/PPP2PPP/RNBQKBNR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/3P4/5N2/PPP1PPPP/RNBQKB1R b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/2PP4/8/PP2PPPP/RNBQKBNR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/2P5/2N5/PP1PPPPP/R1BQKBNR b KQkq - 0 1
kbnr/pppppppp/8/8/2P5/6P1/PP1PPP1P/RNBQKBNR b KQkq - 0 1
kbnr/pppppppp/8/8/2P5/5N2/PP1PPPPP/RNBQKB1R b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/4PP2/8/PPPP2PP/RNBQKBNR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/4P3/2N5/PPPP1PPP/R1BQKBNR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/8/1P6/PBPPPPPP/RN1QKBNR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/8/6P1/PPPPPPBP/RNBQK1NR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/5P2/5N2/PPPPP1PP/RNBQKB1R b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/4P3/5N2/PPPP1PPP/RNBQKB1R b KQkq - 0 1
Yes, two moves isn't enough. But three is too many. You could rerun the same test with WTM each position to confirm this.
Komodo rules!

pohl4711
Posts: 1375
Joined: Sat Sep 03, 2011 5:25 am
Location: Berlin, Germany
Contact:

Re: opposite castling armageddon

Post by pohl4711 » Fri Jul 26, 2019 5:45 pm

Guenther wrote:
Fri Jul 26, 2019 4:30 pm
OneTrickPony wrote:
Fri Jul 26, 2019 12:01 pm
What about: white gets to make 2 first moves and gives draw odds.
Lc0 eval is similar to opposite castling armageddon proposed in OP but it's even simpler to explain.
I did a quick test with CFish20190717 vs. itself and surprisingly the draw rate was still very high > 70% at a fast tc of 40/40''.
used this set of positions:
All you need is here:

https://www.sp-cc.de/drawkiller-openings.htm

With low draw-rates. And no need for new chess-rules, because all Drawkiller lines are starting at the normal classical chess starting position:

These are the two move-sequences, which create the „SALC“-like king- and rook-positions on the chessboard:
1. e3 d6 2. Nh3 Nc6 3. Bd3 Be6 4. O-O Qd7 5. Kh1 O-O-O 6. Re1 Kb8 7. Bf1 Ka8 8. Ng1 Qe8 9. Nf3 Bc8 10. Ng1 Nb8
1. d3 e6 2. Be3 Bd6 3. Nc3 Nh6 4. Qd2 O-O 5. O-O-O Kh8 6. Kb1 Re8 7. Ka1 Bf8 8. Qe1 Ng8 9. Bc1 Nf6 10. Nb1 Ng8
(and at the end of these lines, only some pawn-moves follow)...

Test results:
(asmFish 170426 vs. Komodo 10.4, 5'+3'' time-control, singlecore, no ponder, no endgame-bases, LittleBlitzerGUI, 1000 games each testrun(!) except Noomen Gambit-lines (only 246 positions, so 492 games were played) and Noomen TCEC Superfinal (only 100 positions, so 200 games were played))

Stockfish Framework standard 8 move openings: Score 60.3% – 39.7%, draws: 63.4%
FEOBOS v20 contempt 5 top 500 openings: Score 58.7% - 41.3%, draws: 64.1%
HERT 500 set: Score: 60.6% - 39.4%, draws: 60.4%
Noomen Gambit-Lines: Score 59.1% - 40.9%, draws: 59.3%
4 GM-moves short book: Score 60.5% - 39.5%, draws: 57.1%
Noomen TCEC Superfinal (Season 9+10): Score: 62.5% - 37.5%, draws: 50.0%
SALC V5 half-closed: Score 61.6% - 38.4%, draws: 49.2%
SALC V5 full-closed 500 positions: Score 66.5% - 33.5%, draws: 47.7%

NEW:
Drawkiller (normal set): Score: 65.3% - 34.7%, draws: 33.5%
Drawkiller (tournament set): Score: 65.3% - 34.7%, draws: 33.5%
(no mistake by me: the results of Drawkiller normal and tournament were exactly
the same after 1000 played games!)
Drawkiller (small 500 positions set): Score: 66.4% - 33.6%, draws 30.5%

NEW in V2.0:
Drawkiller balanced (small 500 positions set): Score: 69.3% - 30.7%, draws 35.2%
Drawkiller balanced: Score 69.4% - 30.6%, draws 36.4%
Drawkiller balanced big (15962 positions): Score 67.4% - 32.6%, draws 38.8%

NEW in V3.0:
Drawkiller EloZoom (small 500 positions set): Score: 72.0% - 28.0%, draws 34.6%
Drawkiller EloZoom: Score: 73.2% - 26.8%, draws 36.5%
Drawkiller EloZoom big (20043 positions): Score: 69.2% - 30.8%, draws 40.7%

As you can see, the Drawkiller openings are not just an improvement over my SALC openings, they are a breakthrough into another dimension! Never before any openings-set gave such low draw-rates without crunching the scores of the engines towards 50%, but instead pushing the scores away from 50%. The Drawkiller Normal- and Tournament sets nearly halve the draw-rate, compared to FEOBOS or the Stockfish Framework 8-move openings. And the small 500 set has more than a halved draw-rate compared to FEOBOS or the Stockfish Framework 8-move openings.

lkaufman
Posts: 4339
Joined: Sun Jan 10, 2010 5:15 am
Location: Maryland USA
Contact:

Re: opposite castling armageddon

Post by lkaufman » Fri Jul 26, 2019 6:21 pm

pohl4711 wrote:
Fri Jul 26, 2019 5:45 pm
Guenther wrote:
Fri Jul 26, 2019 4:30 pm
OneTrickPony wrote:
Fri Jul 26, 2019 12:01 pm
What about: white gets to make 2 first moves and gives draw odds.
Lc0 eval is similar to opposite castling armageddon proposed in OP but it's even simpler to explain.
I did a quick test with CFish20190717 vs. itself and surprisingly the draw rate was still very high > 70% at a fast tc of 40/40''.
used this set of positions:
All you need is here:

https://www.sp-cc.de/drawkiller-openings.htm

With low draw-rates. And no need for new chess-rules, because all Drawkiller lines are starting at the normal classical chess starting position:

These are the two move-sequences, which create the „SALC“-like king- and rook-positions on the chessboard:
1. e3 d6 2. Nh3 Nc6 3. Bd3 Be6 4. O-O Qd7 5. Kh1 O-O-O 6. Re1 Kb8 7. Bf1 Ka8 8. Ng1 Qe8 9. Nf3 Bc8 10. Ng1 Nb8
1. d3 e6 2. Be3 Bd6 3. Nc3 Nh6 4. Qd2 O-O 5. O-O-O Kh8 6. Kb1 Re8 7. Ka1 Bf8 8. Qe1 Ng8 9. Bc1 Nf6 10. Nb1 Ng8
(and at the end of these lines, only some pawn-moves follow)...

Test results:
(asmFish 170426 vs. Komodo 10.4, 5'+3'' time-control, singlecore, no ponder, no endgame-bases, LittleBlitzerGUI, 1000 games each testrun(!) except Noomen Gambit-lines (only 246 positions, so 492 games were played) and Noomen TCEC Superfinal (only 100 positions, so 200 games were played))

Stockfish Framework standard 8 move openings: Score 60.3% – 39.7%, draws: 63.4%
FEOBOS v20 contempt 5 top 500 openings: Score 58.7% - 41.3%, draws: 64.1%
HERT 500 set: Score: 60.6% - 39.4%, draws: 60.4%
Noomen Gambit-Lines: Score 59.1% - 40.9%, draws: 59.3%
4 GM-moves short book: Score 60.5% - 39.5%, draws: 57.1%
Noomen TCEC Superfinal (Season 9+10): Score: 62.5% - 37.5%, draws: 50.0%
SALC V5 half-closed: Score 61.6% - 38.4%, draws: 49.2%
SALC V5 full-closed 500 positions: Score 66.5% - 33.5%, draws: 47.7%

NEW:
Drawkiller (normal set): Score: 65.3% - 34.7%, draws: 33.5%
Drawkiller (tournament set): Score: 65.3% - 34.7%, draws: 33.5%
(no mistake by me: the results of Drawkiller normal and tournament were exactly
the same after 1000 played games!)
Drawkiller (small 500 positions set): Score: 66.4% - 33.6%, draws 30.5%

NEW in V2.0:
Drawkiller balanced (small 500 positions set): Score: 69.3% - 30.7%, draws 35.2%
Drawkiller balanced: Score 69.4% - 30.6%, draws 36.4%
Drawkiller balanced big (15962 positions): Score 67.4% - 32.6%, draws 38.8%

NEW in V3.0:
Drawkiller EloZoom (small 500 positions set): Score: 72.0% - 28.0%, draws 34.6%
Drawkiller EloZoom: Score: 73.2% - 26.8%, draws 36.5%
Drawkiller EloZoom big (20043 positions): Score: 69.2% - 30.8%, draws 40.7%

As you can see, the Drawkiller openings are not just an improvement over my SALC openings, they are a breakthrough into another dimension! Never before any openings-set gave such low draw-rates without crunching the scores of the engines towards 50%, but instead pushing the scores away from 50%. The Drawkiller Normal- and Tournament sets nearly halve the draw-rate, compared to FEOBOS or the Stockfish Framework 8-move openings. And the small 500 set has more than a halved draw-rate compared to FEOBOS or the Stockfish Framework 8-move openings.
You have achieved a big reduction in draw rate, but the game you are playing no longer resembles chess. Maybe half the game is about getting your king safe while getting your pieces developed well and securing space, and if you start the game with the kings already safely tucked away it's just a very different game. Also the topic here is how to eliminate draws so there is always a decisive result of each game, reducing the percentage doesn't accomplish that, although it is a desirable goal I agree.
Komodo rules!

Guenther
Posts: 3426
Joined: Wed Oct 01, 2008 4:33 am
Location: Regensburg, Germany
Full name: Guenther Simon
Contact:

Re: opposite castling armageddon

Post by Guenther » Fri Jul 26, 2019 6:23 pm

lkaufman wrote:
Fri Jul 26, 2019 5:27 pm
Guenther wrote:
Fri Jul 26, 2019 4:30 pm
OneTrickPony wrote:
Fri Jul 26, 2019 12:01 pm
What about: white gets to make 2 first moves and gives draw odds.
Lc0 eval is similar to opposite castling armageddon proposed in OP but it's even simpler to explain.
I did a quick test with CFish20190717 vs. itself and surprisingly the draw rate was still very high > 70% at a fast tc of 40/40''.
used this set of positions:

Code: Select all

rnbqkbnr/pppppppp/8/8/3PP3/8/PPP2PPP/RNBQKBNR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/3P4/5N2/PPP1PPPP/RNBQKB1R b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/2PP4/8/PP2PPPP/RNBQKBNR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/2P5/2N5/PP1PPPPP/R1BQKBNR b KQkq - 0 1
kbnr/pppppppp/8/8/2P5/6P1/PP1PPP1P/RNBQKBNR b KQkq - 0 1
kbnr/pppppppp/8/8/2P5/5N2/PP1PPPPP/RNBQKB1R b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/4PP2/8/PPPP2PP/RNBQKBNR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/4P3/2N5/PPPP1PPP/R1BQKBNR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/8/1P6/PBPPPPPP/RN1QKBNR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/8/6P1/PPPPPPBP/RNBQK1NR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/5P2/5N2/PPPPP1PP/RNBQKB1R b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/4P3/5N2/PPPP1PPP/RNBQKB1R b KQkq - 0 1
Yes, two moves isn't enough. But three is too many. You could rerun the same test with WTM each position to confirm this.
By far not enough games but also surprisingly interim result of 11 White wins and 13 draws after 24 games with another epd set:
(only diff between CFish1 and CFish2 being contempt 9 vs. 10 - regrettably far too hot here for more games currently.)

Code: Select all

rnbqkbnr/pppppppp/8/8/2PPP3/8/PP3PPP/RNBQKBNR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/2P5/2N3P1/PP1PPP1P/R1BQKBNR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/3P4/5NP1/PPP1PP1P/RNBQKB1R b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/2B1P3/5N2/PPPP1PPP/RNBQK2R b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/4PP2/5N2/PPPP2PP/RNBQKB1R b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/4PP2/2N5/PPPP2PP/R1BQKBNR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/2PP4/2N5/PP2PPPP/R1BQKBNR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/3P4/1P3N2/P1P1PPPP/RNBQKB1R b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/3P4/1P3N2/P1P1PPPP/RNBQKB1R b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/3PP3/5N2/PPP2PPP/RNBQKB1R b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/3PP3/2N5/PPP2PPP/R1BQKBNR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/3P4/6P1/PPP1PPBP/RNBQK1NR b KQkq - 0 1
https://rwbc-chess.de
Greg Strong@ovyron wrote: What I think is that you have a very, very, VERY strong desire to post, and very little self-control.
All the time. In every conversation. You must post. Something. It's like you just can't help yourself.

Guenther
Posts: 3426
Joined: Wed Oct 01, 2008 4:33 am
Location: Regensburg, Germany
Full name: Guenther Simon
Contact:

Re: opposite castling armageddon

Post by Guenther » Fri Jul 26, 2019 8:02 pm

Guenther wrote:
Fri Jul 26, 2019 6:23 pm

By far not enough games but also surprisingly interim result of 11 White wins and 13 draws after 24 games with another epd set:
(only diff between CFish1 and CFish2 being contempt 9 vs. 10 - regrettably far too hot here for more games currently.)

Code: Select all

rnbqkbnr/pppppppp/8/8/2PPP3/8/PP3PPP/RNBQKBNR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/2P5/2N3P1/PP1PPP1P/R1BQKBNR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/3P4/5NP1/PPP1PP1P/RNBQKB1R b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/2B1P3/5N2/PPPP1PPP/RNBQK2R b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/4PP2/5N2/PPPP2PP/RNBQKB1R b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/4PP2/2N5/PPPP2PP/R1BQKBNR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/2PP4/2N5/PP2PPPP/R1BQKBNR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/3P4/1P3N2/P1P1PPPP/RNBQKB1R b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/3P4/1P3N2/P1P1PPPP/RNBQKB1R b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/3PP3/5N2/PPP2PPP/RNBQKB1R b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/3PP3/2N5/PPP2PPP/R1BQKBNR b KQkq - 0 1
rnbqkbnr/pppppppp/8/8/3P4/6P1/PPP1PPBP/RNBQK1NR b KQkq - 0 1
Noticed that in my adhoc created epd for 3 moves by accident one position was twice...and this position produced 4 draws.
https://rwbc-chess.de
Greg Strong@ovyron wrote: What I think is that you have a very, very, VERY strong desire to post, and very little self-control.
All the time. In every conversation. You must post. Something. It's like you just can't help yourself.

Joerg Oster
Posts: 789
Joined: Fri Mar 10, 2006 3:29 pm
Location: Germany

Re: opposite castling armageddon

Post by Joerg Oster » Fri Jul 26, 2019 8:49 pm

lkaufman wrote:
Fri Jul 26, 2019 5:03 am
I have wondered what would be the simplest rule change to chess that would make it a two-result game (win or loss) while still being roughly fair to both players. In the real chess world this is sometimes done by giving White more time and Black draw odds, but with a few exceptions this has been done only at ten minute or faster time limits. Let's say we won't use time in the rules, equal time is assumed. Here is what I think might be an original, simple, and fair enough solution.
White is not allowed to castle long, Black is not allowed to castle short, Black wins draws. That's it. It is a subset of normal chess; if you stipulate the starting sequence 1.Na3 Nh6 2.Rb1 Rg8 3.Ra1 Rh8 4.Nb1 Ng8 you get the above castling status. Since castling short is easier to accomplish than castling long and doesn't require an extra move to "finish" the castle (Kb8) White's advantage is obviously larger than normal, despite the symmetry of the rule, and the fact that same side castling never happens makes the draw odds less significant. But it's far from clear which side has the better chances or the theoretically won position. Most people would rather play White, because it's more "fun", but whether White would score more than 50% in either human vs human or engine vs engine play is not obvious. Recent strong versions of Lc0 give White about 68-70% win exp. with draws counting as a half, which is close to the point where White would win half the games outright. Stockfish and Komodo give White a bit more of an advantage than what I would normally call the line between a draw and a win (about .7 for Komodo, about 1 for SF). I had ten different Lc0 networks play it out using the "Shootout" feature on Fritz, at 2' + 1" on rtx 2080, and White won four, drew six. My hunch is that White will come out slightly ahead in human GM play, perhaps about even or a bit behind in engine play, but of course it's very hard to predict, and ten games with similar engines means little.
So who would you bet on, White or Black. First case top human GMs playing, second case top engines. And who benefits from longer time controls, White or Black? Is it roughly fair? It doesn't have to be perfectly fair, as players alternate colors and chess isn't at all fair in that sense now.
Here are two games played by a modified Stockfish, time control was 30"+0.3".
Note: it recognizes draws by repetition, 50-move rule und stalemates as wins for Black,
but not draws by insufficient material and no draw endgame rules have been changed.

Jörg Oster

pohl4711
Posts: 1375
Joined: Sat Sep 03, 2011 5:25 am
Location: Berlin, Germany
Contact:

Re: opposite castling armageddon

Post by pohl4711 » Sat Jul 27, 2019 2:27 am

lkaufman wrote:
Fri Jul 26, 2019 6:21 pm
You have achieved a big reduction in draw rate, but the game you are playing no longer resembles chess. Maybe half the game is about getting your king safe while getting your pieces developed well and securing space, and if you start the game with the kings already safely tucked away it's just a very different game.
The low-draw rates of Drawkiller prove, that you are completely wrong: moving the kings to opposite side of the board does NOT make them safe. Otherwise the draw-rates must be much higher (mention, that all Drawkiller endpositions have a balanced Komodo-eval). So classical openings make kings save (perhaps), but definitly not Drawkiller.
The pieces are NOT getting “developed well“ with Drawkiller: 11 of 16 non-pawn pieces are on the same position as in the classical chess starting position and all 16 non-pawn pieces are on the both baselines (1 and 8). So classical openings develop pieces (perhaps), but definitly not Drawkiller.

lkaufman
Posts: 4339
Joined: Sun Jan 10, 2010 5:15 am
Location: Maryland USA
Contact:

Re: opposite castling armageddon

Post by lkaufman » Sat Jul 27, 2019 4:08 am

pohl4711 wrote:
Sat Jul 27, 2019 2:27 am
lkaufman wrote:
Fri Jul 26, 2019 6:21 pm
You have achieved a big reduction in draw rate, but the game you are playing no longer resembles chess. Maybe half the game is about getting your king safe while getting your pieces developed well and securing space, and if you start the game with the kings already safely tucked away it's just a very different game.
The low-draw rates of Drawkiller prove, that you are completely wrong: moving the kings to opposite side of the board does NOT make them safe. Otherwise the draw-rates must be much higher (mention, that all Drawkiller endpositions have a balanced Komodo-eval). So classical openings make kings save (perhaps), but definitly not Drawkiller.
The pieces are NOT getting “developed well“ with Drawkiller: 11 of 16 non-pawn pieces are on the same position as in the classical chess starting position and all 16 non-pawn pieces are on the both baselines (1 and 8). So classical openings develop pieces (perhaps), but definitly not Drawkiller.
Yes, I should have said that your start positions make the kings about as safe as they can be given the opposite-side castling stipulation. My point is that part of chess is deciding where to place the king and how to develop in such a way as to make it possible to get the king to whichever corner is desired. Your start positions just skip over this very fundamental part of chess. Now I don't dispute that your drawkiller set is useful for testing engines, although it won't test their ability to play the opening properly. I suppose for people who only use engines starting around move 15 or so when the opening is pretty much finished, your set is extremely valuable. But I don't think it is a serious candidate for human use, or for testing how engines play the openings. Too much of the game has already been set in stone for that purpose. I much prefer chess960 for human use, as all the decisions of how to develop and safeguard the king remain to be made. But chess960 is only useful for engines to the extent that it becomes accepted by human players.
Komodo rules!

Ferdy
Posts: 4295
Joined: Sun Aug 10, 2008 1:15 pm
Location: Philippines

Re: opposite castling armageddon

Post by Ferdy » Sat Jul 27, 2019 7:45 am

lkaufman wrote:
Fri Jul 26, 2019 5:03 am
I have wondered what would be the simplest rule change to chess that would make it a two-result game (win or loss) while still being roughly fair to both players. In the real chess world this is sometimes done by giving White more time and Black draw odds, but with a few exceptions this has been done only at ten minute or faster time limits. Let's say we won't use time in the rules, equal time is assumed. Here is what I think might be an original, simple, and fair enough solution.
White is not allowed to castle long, Black is not allowed to castle short, Black wins draws. That's it. It is a subset of normal chess; if you stipulate the starting sequence 1.Na3 Nh6 2.Rb1 Rg8 3.Ra1 Rh8 4.Nb1 Ng8 you get the above castling status. Since castling short is easier to accomplish than castling long and doesn't require an extra move to "finish" the castle (Kb8) White's advantage is obviously larger than normal, despite the symmetry of the rule, and the fact that same side castling never happens makes the draw odds less significant. But it's far from clear which side has the better chances or the theoretically won position. Most people would rather play White, because it's more "fun", but whether White would score more than 50% in either human vs human or engine vs engine play is not obvious. Recent strong versions of Lc0 give White about 68-70% win exp. with draws counting as a half, which is close to the point where White would win half the games outright. Stockfish and Komodo give White a bit more of an advantage than what I would normally call the line between a draw and a win (about .7 for Komodo, about 1 for SF). I had ten different Lc0 networks play it out using the "Shootout" feature on Fritz, at 2' + 1" on rtx 2080, and White won four, drew six. My hunch is that White will come out slightly ahead in human GM play, perhaps about even or a bit behind in engine play, but of course it's very hard to predict, and ten games with similar engines means little.
So who would you bet on, White or Black. First case top human GMs playing, second case top engines. And who benefits from longer time controls, White or Black? Is it roughly fair? It doesn't have to be perfectly fair, as players alternate colors and chess isn't at all fair in that sense now.
Implemented Armageddon in Deuterium (around 3000 on ccrl 40/4), test it at dissimilar TC, white has 5 minutes and black has 4 minutes. It used a 4-ply noomen opening suite run in cutechess-cli randomly.

Result after 50 games.

Code: Select all

GameVariant: armageddon
WhiteTC: 300, BlackTC: 240
                                   Engine  Total  Win  Loss  Draw  Score  Score%
 Deuterium v2019.2.37.62 armageddon_white     50   17     6    27     17    34.0
 Deuterium v2019.2.37.62 armageddon_black     50    6    17    27     33    66.0
Games are at https://drive.google.com/file/d/1wNwtwh ... sp=sharing
Games with draw results were already corrected see Termination tag in the game.


Also implemented a variant with your idea and call it lkaufman. TC is the same as in armageddon that is white has 5 minutes and black has 4 minutes. I would like to see the impact of the castling this is why I use dissimilar TC, similar to armageddon test. Also used the 4-ply noomen test suite.

Result after 50 games.

Code: Select all

GameVariant: lkaufman
WhiteTC: 300, BlackTC: 240
                                 Engine  Total  Win  Loss  Draw  Score  Score%
 Deuterium v2019.2.37.63 Lkaufman_white     50   21    10    19     21    42.0
 Deuterium v2019.2.37.63 Lkaufman_black     50   10    21    19     29    58.0
Games are at https://drive.google.com/file/d/1FQzXvb ... sp=sharing

So there is improvement on the white performance (34% to 42%) from armageddon to lkaufman.

pohl4711
Posts: 1375
Joined: Sat Sep 03, 2011 5:25 am
Location: Berlin, Germany
Contact:

Re: opposite castling armageddon

Post by pohl4711 » Sat Jul 27, 2019 9:45 am

lkaufman wrote:
Fri Jul 26, 2019 5:03 am
I have wondered what would be the simplest rule change to chess that would make it a two-result game (win or loss) while still being roughly fair to both players. In the real chess world this is sometimes done by giving White more time and Black draw odds, but with a few exceptions this has been done only at ten minute or faster time limits. Let's say we won't use time in the rules, equal time is assumed. Here is what I think might be an original, simple, and fair enough solution.
White is not allowed to castle long, Black is not allowed to castle short, Black wins draws. That's it. It is a subset of normal chess; if you stipulate the starting sequence 1.Na3 Nh6 2.Rb1 Rg8 3.Ra1 Rh8 4.Nb1 Ng8 you get the above castling status. Since castling short is easier to accomplish than castling long and doesn't require an extra move to "finish" the castle (Kb8) White's advantage is obviously larger than normal, despite the symmetry of the rule, and the fact that same side castling never happens makes the draw odds less significant. But it's far from clear which side has the better chances or the theoretically won position. Most people would rather play White, because it's more "fun", but whether White would score more than 50% in either human vs human or engine vs engine play is not obvious. Recent strong versions of Lc0 give White about 68-70% win exp. with draws counting as a half, which is close to the point where White would win half the games outright. Stockfish and Komodo give White a bit more of an advantage than what I would normally call the line between a draw and a win (about .7 for Komodo, about 1 for SF). I had ten different Lc0 networks play it out using the "Shootout" feature on Fritz, at 2' + 1" on rtx 2080, and White won four, drew six. My hunch is that White will come out slightly ahead in human GM play, perhaps about even or a bit behind in engine play, but of course it's very hard to predict, and ten games with similar engines means little.
So who would you bet on, White or Black. First case top human GMs playing, second case top engines. And who benefits from longer time controls, White or Black? Is it roughly fair? It doesn't have to be perfectly fair, as players alternate colors and chess isn't at all fair in that sense now.
I find the idea interesting. So, I took the Stockfish Framework 8move-openings v3 (34700 games) and added the line 1.Na3 Nh6 2.Rb1 Rg8 3.Ra1 Rh8 4.Nb1 Ng8 in the front of each opening and deleted all lines, which included illegal move sequences, now (short castlings of Black for example).
20616 opening lines remaining.

I added a smaller set with only 4 moves (8 plies) played (plus the Armageddon line). Unique endpositions only:
3614 opening lines remaining.
I added a smaller set with only 2.5 moves (5 plies) played (plus the Armageddon line). Unique endpositions only:
871 opening lines remaining.

You find the PGN and the EPD-files in the folder. You can use them for engine-play. When finished, just set all drawn-games to a win for Black - done!

Download on my website or right here:
https://www.sp-cc.de/files/lk_armageddon.zip

Stefan (SPCC)

Post Reply