Capablanca Chess and Bird's Chess question(Is the initial position win for white?)

[Uri Blass]

From https://www.chess.com/terms/gothic-chess

"Trice's improvement over Capablanca Chess and Bird's Chess variants lies in the initial setup of the pieces. In Gothic Chess, none of the pawns are undefended at the beginning of the game. According to Trice, this flaw is present in all other variants of this kind, unbalancing the game in White's favor."

I do not know if it is correct that the game is unbalanced in Capablanca Chess or Bird's Chess variants.

If this opinion is not based on engine-engine games then I do not trust it because humans are weak at chess.

I would like to know if there is some evidence that white is winning in Capablanca Chess or Bird's Chess variants.
It is enough for me if most of computer-computer games that are based on NN engines for that games are win for white and not a draw.

Did somebody try to make NN or NNUE engines for these games?(note that if you do not use Ed Trice initial position then Ed Trice does not have a patent about the game and you do not need to ask him for permission).

[Ajedrecista]

Hello Uri:

I guess that these variants are way less studied than normal chess and we simply can not do a formal proof that normal chess is a draw, let alone in these variants. All signs point to draw in normal chess, of course, and top engines playing LTC without one-sided openings draw more than 90% (?) of games, which is enough for you ('It is enough for me if most of computer-computer games [...]').

Was not around there Fairy-Stockfish a while ago, which was very strong in different variants? I also remember Bihasa by Ferdinand Mosca being very strong back in the day, which plays Capablanca Chess and Gothic Chess according to CPW. Ferd coded strong engines for other chess variants IIRC. However, I do not know the current rating status of engines of these variants, to be fair.

------------

The idea of take an already successful, well stablished game and introduce a different piece arrangement at the start is fairly old. When learning chess, who had not put the rooks upside down to be queens and play from there?

Well, I want to share a known-to-me old example of other board game exploiting the idea: Spanish draughts. There is a book published in 1597 called Libro del ivego de las damas, por otro nombre el marro de pvnta, diuidido en tres tratados (Old Castillian language), freely translated as Book of the game of draughts, also named marro de punta, divided into three treatises. Spanish draughts checkerboard has got 8×8 squares (like chess) and each player starts with 12 men in his/her first three rows in the normal game. This book curiously covers a variant in the third treatise that replaces for each side a man of the kingsrow (where the major pieces start in chess) with a king, freely to choose each player his/her man to replace (numbered checkerboard at folio 36r and tip at folio 36v).

The tip is to put the kings on squares 3 (d1) and 30 (e8), also saying that 'when playing with kings from the start, it is best to play first'. So, there is an example of the late 16th century claiming a first-move advantage in a Spanish draughts variant. I guess that d1 is preferred over b1, f1 and h1 (the same for e8 over a8, c8 and g8) because if we count the number of available squares for the king (long king in Spanish draughts, more less like a bishop in chess) in an empty board, counting left and right, are 1+6 (b1), 3+4 (d1), 5+2 (f1) and 7+0 (h1), each adding 7, but the highest minimum is 3 for d1 (squares c2, b3 and a4; or 7, 12 and 16 on the numbered checkerboard). If you do not find the numbering scheme logical, just turn the checkerboard around 180º and you will get 1-2-3-4;5-6-7-8;...;29-30-31-32.

Spanish draughts were very popular in Spain centuries ago, but is almost dissapeared nowadays, sadly. OTOH, Spain has been for many years the country with most chess tournaments IIRC. It is said that the queen piece became more poweful in Spain circa the last third of the 15th century (please read my post and the links in it), as well as other old books, the Ruy López opening (the Spanish opening); and we proudly hosted the WCC 1987 (please read my summary here), many Linares International Chess Tournament (where Shirov played the famous Bh3!! and where Kasparov retired) and the Candidates Tournament 2022, which I was tempted to go and finally discarded the idea.

Regards from Spain.

Ajedrecista.

[jefk]

do not know if it is correct that the game is unbalanced in Capablanca Chess or Bird's Chess variants.

it appears to be true, but only slightly unbalanced (mr Trice, the inventor of Gothic chess)
once was exaggerating, maybe to promote his patented variant, improved Capablanca).
https://www.perplexity.ai/search/what-a ... gJQ9MNvw#0

In Fairy-Stockfish, a Nnue was developed for the Capablanca variant, which
also can be applied for Gothic chess.
https://fairy-stockfish.github.io/nnue/

The reasoning that an undefended pawn in the initial position leads to unbalanced
games is rather far-fetched for me, but apparently seems to be true, with for
Capablanca-chess the initial advantage for the first move being about 0.4 (pawn value).
During the game play this advantage will decrease with best play thus it is *not* a
forced win(*); not surprising for me btw, it's not a sharp/forcing as eg. in Losing-chess.

Gothic chess seems to be an improvement, and also less dependent on opening knowledge
(although learning the best openings for Capablanca chess would take some effort i think).
That being said, it's still not to everyone's taste:

Chess960 clearly is more popular nowadays (and imo can be further improved).

(*) even although the Gothic inventor, mr Trice once claimed that on a yt vid.

[hgm]

I think the 'proof' by Ed Trice was based on an assumption of an erroneous piece value for the Archbishop, namely 6.5-7 instead of 8.75. He showed that there was an opening line that forced a C for A trade, (the uncontroversial value of the Chancellor being 9), and assumed the imagined 2-2.5 material advantage would then guarantee a win. But in reality the material advantage is only a quarter Pawn, and this is largely offset by the fact that the variation gave black a half-open b-file.

The role of unprotected Pawns in the opening position is often exaggerated. For one, pieces get developed, and Pawns protected by a single piece might become unprotected when you develop that piece. Refraining from development just to keep a Pawn protected is a losing tradeoff. And Pawns that initially are unprotected might get protected by developing a piece. E.g. suppose a variant had swapped the Rooks and Bishops, so that a7 and h7 are unprotected, this is not a big deal, because black intends to play Nc6 and Nf6 anyway, which would protect them.

So it all depends how easy it is to attack the Pawns vs protecting them. IMO if an unprotected Pawn can be protected by a solid developing move, it is no realnweakness.

[Nordlandia]

Ed's assigned value to the arcbishop has to be taken with grain of salt. In actuality, the piece is closer to 8 than 7. By the way,
I stumbled upon Conservative Capablanca Chess which resembles the usual layout of the pieces, while the new heavyweights are in each corner.

http://ftp.chessvariants.com/rules/cons ... anca-chess

I'm into Capablanca Chess and Trice's Chess (new name) which can be played on Chess.com variant server and on pychess. The latter do offer randomized setups.

[Ajedrecista]

Hello:

Just by chance, I got aware about the family of tafl games yesterday, where both the size of the board (tipically 11×11 and 7×7) and the initial arrangement of the pieces can play a major role in the outcome of the game. A short, incomplete list of variants according to Wikipedia is:

Code: Select all

K: king.

============================================
            SIZE OF 
NAME       THE BOARD   ATTACKERS   DEFENDERS
============================================
Brandubh      7×7           8         4+K
Ard Rí        7×7          16         8+K
Tablut        9×9          16         8+K
Hnefatafl    11×11         24        12+K
Tawlbwrdd    11×11         24        12+K

There is even a list of game balances depending on some initial arrangements and board sizes after analyzing the oucome of some games, because the nature of tafl games is asymmetric sides (attackers vs. defenders), where some variants might favour attackers and other variants might favour defenders. If a variant is widely considered to be a win for a certain side, there is a possibility of bid a maximum number of moves to win (somethink like a moves odd equivalent to the draw odds at Armaggedon chess); or play each side once, and supposing the stronger side wins both games (for example the defenders), the tiebreak would be the winner player is the player that won in less moves, otherwise the match is drawn.

Brandubh variant looks reasonable to start with at computer level, due to the reduced size board (7×7) and the low number of pieces (8 vs. 5), though an upper bound of the state space complexity of Brandubh was estimated at circa 1.04e+14 in a paper (link to the paper got thanks to a Reddit thread). It looks like there are some engines that can play tafl games, just searching "tafl game engine" on Google.

------------

As a side note, I remember that I played a casual game of a variant of Spanish draughts in the vicinity of the Castle of Sigüenza (a highly recommended visit!) at around the summer of 2008 or the summer of 2009. Rules of Spanish draughts applied, with the differences that the board was not 8×8 but smaller (I do not remember if it was 5×5 or 5×4), there were less pieces on the checkerboard and also asymmetric sides like in tafl games (something like 4 vs. 3 or 5 vs. 4, with the player with the extra man moving first), with draw odds like Armaggedon chess. The context was pretty similar to tafl games, this case with the concept of siege, where the extra man side would be the attackers out of the castle that start the offensive and must win; and the man down size must resist and an eventual shuffling of pieces (the attackers can not convert a endgame with usually more pieces than the defenders) is understood as a fail of the siege, thus the win of the defenders. I do not know if this variant is heavily unbalanced, but just wanted to share old memories.

Regards from Spain.

Ajedrecista.

[Nordlandia]

As I understand it, there are several reasons why Capablanca's version of chess did not catch on 100 years ago, nor decades later. The board is rectangular after all and the weaker pieces lose strength more or less on larger boards.

At the same time, many people wonder why Seirawan Chess (S-Chess or Sharper Chess) has not attracted more strong players, given that the board layout is the same and most of the game closely resembles traditional chess. In my view, S-Chess should be considered a fully viable format alongside Chess960 in tournament play. However, it would be unrealistic to expect players to adopt it enthusiastically from the outset. Regardless of how effectively it is promoted online by enthusiasts and supporters, it cannot compete with the chess community’s tendency to gravitate toward conventional chess and, to a lesser extent, Chess960.

https://mats-winther.github.io/bg/seirawanchess.htm