Engine playing musketeer chess variant

[musketeerchess]

Hi
I'm looking for programmers to help me develop an engine playing musketeer chess variant.

It could be based on Stockfish, a work from scratch or based on any other existing commercial or free chess engines.

I'm willing to donate helping developing these engines. Only request is that the engine among the top 10.

If you know someone or your self already participated in developing engines let me know and send me a private message to musketeerchess (at) gmail.com

best regards
Zied

PS: my pieces are going commercial beginning October on HOUSEOFSTAUNTON.COM

you can see them on www.musketeerchess.com (not updated) or on my FB (Zied HADDAD)

[hgm]

So an engine that is merely of GM strength is not good enough for you, you really must have one with an Elo above 3400?

I can imagine the House of Staunton's interest in marketing the pieces, and how existence of software might help to popularize a game. But I don't understand their policy about software. They (or at least their director) are in possession of a version of Komodo that can play Seirawan Chess. (I still have a copy of that, as I was involved in its testing.) But they seem to be just sitting on it, for years already.

Having more different piece types will no doubt make Chess-variant afficionados very happy. I doubt they will use them for Musketeer Chess much, however. It seems the Seirawan pieces (Elephant and Hawk) are also mostly used for other games than S-Chess.

Of course you can do with your money whatever pleases you. I should warn you, though, that IMO you should not expect (m)any returns from an investment in Musketeer Chess (be it equipment or software). Of the many Chess variants I have seen I would not rate it much above average, (but that of course is just my opinion), and attempts with other good variants like Gothic Chess have not proved viable. But marketing is of course everything in such an endeavor, as the Omega-Chess case shows.

It is a sort of universal phenomenon that inventors of Chess variants are highly biased towards thinking their variant is the greatest invention since chocolate Chess pieces, while in practice even very good designs are often hardly played by anyone but the inventor and some of his close friends.

[musketeerchess]

Hi
I'm happy to share your thoughts. I find them smart and clever.

I assume the fact that my chess variant is a "personal" variant. But i also think that they are in the continuum of chess spirit. The pieces are very well designed like Elephant And Hawk.

My new chess pieces will certainly be used for other chess variants which will be very good to help Musketeer Chess become widespread.

The development of a specific engine will also be one of the best ways to make it widespread. That's why if one of the top engines could be involved in the development will be exciting for both of us.

But having an engine GM Level is for me sufficient. But in terms of marketing having a top ten is a warranty of quality.

[hgm]

So how about Ferdinand Mosca's engine? I recall he converted his Seirawan-Chess engine to do Musketeer. Have you tried it?

Houdini and Komodo are commercial engines developed by very small teams, and they will need all their time to keep their engine competitive for normal Chess. They won't risk falling behind there because they are side-tracked working on Musketeer, unless you would pay them really big bucks for it. And it is very unlikely they would be willing to hand their code to someone else who could then convert it. Other top engines often have a variety of authors, and some public license on the source code. Although anyone would be allowed to convert them to Musketeer, the license demands the source code must always remain public, so that you could never commercialize it. I don't know if that would be a problem for you.

Another concern is that most top engines nowadays use bitboards, with move-generating algorithms that are very specifically tailored to doing Rook-like and Bishop-like moves, and that some of your pieces, in particular the Fortress, would not fall in this category, and might require extensive modifications to the engine.

[Ferdy]

So how about Ferdinand Mosca's engine? I recall he converted his Seirawan-Chess engine to do Musketeer. Have you tried it?

I have not released it (also not all pieces are supported) I think it was only leopard and cannon that are implemented, not interested to continue its development as there are no possible opponents.
This man has also a surprisingly high demand for strength, he seemed interested on development based on existing strong top normal chess engines .

[musketeerchess]

Hi
If Stockfish or any other top free engines want to work on Musketeer Chess, i see no problem about the source code remaining free because i respect the spirit of GNU Licence.
I haven't tried Mosca. I'll be interested to try it if Ferdinand dares to send me his engine. Is it on winboard?

[hgm]

I don't know anything about the Stockfish code, but I can imagine it could be quite hard to extend a highly advanced engine to tasks beyond what it was designed to handle. E.g. increasing the number of piece types or pieces might require a change in the number of bits required to indicate the piece, which might require a change in the format of data that packs piece type in it (e.g. the move representation, if it contains the type of the captured piece), which then might not fit anymore in the basic data type that was used for it... Move encoding would have to be extended to handle gating moves, which might already require extra space in itself.

The only experience I have in this game is that I once made an attempt to convert Fruit 2.1 to play Seirawan Chess (in order to have a better-matched opponent for Komodo-S). Fruit is a comparatively straightforward engine, but I was a bit dismayed by how many problems I had to solve before it started working. (And I never got the gating at the Rook-square of a castling working properly; after I tried to implement that the engine became prone to crashing, while before it worked perfectly (apart from that it did not accept such gatings).)

My experience was that increasing the number of pieces, increasing the number of piece types, introducing new piece-type classes (like mixed slider/leaper), adding new move steps (like direct leaps to 2nd or 3rd square on a ray, or bent slider trajectories), adding new special moves (like gating or rifle capture) are all problematic, and tend to get to become more problematic if the engine is more advanced. (Because it does what is basically the same thing in a number of different ways for speed optimalization, e.g. generating moves for the purpose of searching them, for the purpose of determining whether you are in check, for determining if pieces can be safely captured, etc.)

[musketeerchess]

Hi
Stockfish is an engine "much easier" to tune than Fruit IMHO.

When you look at the code, and you compare it to the strength you are astonished how "simple" it is.

The pieces are "implemented" each one separately. Which can help implementing new pieces because they have their own programmation. The most difficult part is to make the necessary changes sothat the rules of Musketeer Chess could be added.

The program will crush many times, which is not a programming failure but a part of the normal process when you program. When i was younger i used to be a programmer. But this is far far away now.

I don't have enough time to learn again how to program in C or C++. But if i cannot find help (and i'm intending to help those programmers by donating and it doesn't matter if the source code remains free. All i want is that musketeer chess become a popular chess variant) i'll certainly learn again how to program and try to do it by my self.

If someone knows the mails of Shay Bushinsky, Amir Ban (Junior team), Hiarcs Team, Houdini Team please send me a personal message.

I'm also willing to make regularly a tournament between engines playing Musketeer Chess (i'll give the initial position and the chosen pieces the day of the tournament sothat it's the pure strength and creativity of the programmers which will be tested). But for that there must be at least 4-5 engines.

[hgm]

All i want is that musketeer chess become a popular chess variant) i'll certainly learn again how to program and try to do it by my self.

OK, I see. If Stockfish is easy, (and I believe your word for that, as I have never seen the source code myself), then this would seem a good way to go.

Btw, I am a bit skeptical that a program that would totally crush all human opposition would contribute anything to making a Chess variant popular. People typically use such engines to analyse their games against other strong players, to know the 'absolute truth' about them, and see what their most expensive mistakes were. But they won't have any such games to analyze, and won't be motivated to analyze any games they might have played, when they are not already hooked to the game and start to consider it a seriously.

To contribute to the popularity of a variant people should have the feeling that they might be able to beat the computer, if they pay attention. So the computer must play recognizably stupid in their eyes, and despite of that it will still punish them, as humans never stay fully alert during the entire game, and the engine will simply grab the material they so generously offer when their attention falters for a moment. The idea that they could have won "if only they had not made that stupid mistake that is way below their level" is what makes them come back game after game. Being crushed without a chance by a 3400 Elo monster typically only scares them away from the game, after one or two tries.

[musketeerchess]

Hi
Having a very strong engine can permit scaling different levels of strength.

Also a very strong engine could help making spectacular games.