Free online Chess server. Play Chess now in a clean interface. No registration, no ads, no plugin required. Play Chess with the computer, friends or random opponents.
Sign in

Installing antichess engine on Lichess.

Asked by Chesstroll_Ingot
Tags engine antichess feature variant
Activity Viewed 9038 times, last updated
I know developers are working now on atomic chess engine. And I tried to play against it and I found it very strong.
But there is nothing said about antichess engine.
I know:
1) There is a site with many solved lines:
2) The engine which is made by Fritz. It's Fritz Giveaway engine. The only insignificant difference it has is ability to castle (in Lichess antichess there is no castling). This is very strong engine.
It would be awesome if here would be possibilities to play against computer in atomich and anti chess, and also analyse positions.
But, as I said, I heard nothing about plans to install antichess engine.
The problem that may be, is that this could be illegal to use that engine. But what if to take algorithms from it? I don't know much about programming. But I'm just interested in that why there are no thoughts about implementing antichess engine to the site. Because there is a whole group for atomic engine development.
3) Also there is, as I understood, free engine "nilatac".
I think it's possible to use it.
One answer
Answered by Unihedron
To quote from a similar previous forum post (

"[...] The thing about shipping software is that once it's deployed, it's generally considered a binary implementation and is unlikely to change, so implementing things the best way possible should be done on the first try." - Unihedron

Allow me to quote from a discussion I had between Toadofsky, the developer working on teaching Stockfish how to play atomic:

"[...] I have asked Thibault about the same question (why not have more than one engine) and if I remember correctly he explained that having multiple engines would significantly complicate AI cluster configuration, deployment, and load balancing. [...]

"I have an additional concern: the way the AI cluster performs game analysis is fast but error-prone (since Stockfish instances can't share a cache and each position is analysed with the same amount of effort) and I'm afraid to make any change (such as adding a new engine; [...]) which might make it more difficult to address that issue." - Dan

While existing engines exist and we don't mind uses of it outside the scope of cheating (You can play with an engine to improve, not to use it to play for you), you can consider it an official stance that you'll unlikely see a non-Lichess-multi-variant-stockfish-build engine very soon.

There are more engines than Nilatac, some so strong that they can be considered reference quality for base of teaching stockfish of it; If you or any future readers would like to send any variant engines our way that you recommend (please, test to make sure it is of outstanding quality first), do message me or the developers relevant!
Chesstroll_Ingot commented :
Thank you for the answer!
cat_person commented :
What about adding the solved lines by Watkins and Steenhuis? That will be some GBs....4-piece endgame database is not large. 5-piece can be huge (60 GB), the solved lines themselves should be less than 30GB after decompression..
cat_person commented :
When these known lines are added the quality will be way better than most known engines..BTW that 5-piece database is not necessary.
Unihedron commented :
Opening books are indeed a helpful implementation detail, but in terms of computer analysis it is likely going to be Stockfish taught with variant knowledge for sure.
cat_person commented :
Of course...:-) We should listen to Thibault....hope the Stockfish for more variants come soon!:-)

Only registered members with one week of lichess activity can contribute to the Q&A.