The Universal Solver

[Vinvin]

The Universal Chess Solver

What we need : an engine that didn't miss direct good moves. This engine should have way less blind spots (= when very good moves are wrongly pruned) than the current top engines.
The strength of play of this engine should not be (approx) 50 Elo under Stockfish 12.

The ingredients :

3 main ingredients :
Crystal 3.1 : for his forteress detection, non-progress detection and sacrifice-for-attack detection. SugaR-AI-ICCF-1.0 and ShashChess 15.1 are close variations.
BlueFish 12 : Seems to be different from Crystal but also finds a lot of deep tactics.
Sting-SF 27 : even better at fortress detection, piece trapped, special endgames understanding and deep mates detection. But this code is based on the old Stockfish 2.1.1.

+3 seasonings :
Corchess : it seems to have a very clean search
McCain-v2 : has a special ability to report deep mates in Syzygy search.
Black-Diamond : Very good to find special moves but weak in play


The recipe :
Take the best of all this 6 engines and put them together efficiently.
An improvement in all these engines would be the ability to find a shorter mate when it discovers a mate.


The idea is to create an engine able to solve all (or about all) the 114 positions (in less than 300 seconds on 1 core or 20 seconds on 16 cores) here : http://talkchess.com/forum3/viewtopic.p ... 36#p882236 (The Hard-Talkchess-2020, revision 2021)
But this engine should be strong compare to current standards, 50 Elo under SF12 is a good target. As you can see here : http://talkchess.com/forum3/viewtopic.p ... 78#p878878, Bluefish, ShashChess, Crystal and SugaR AI ICCF already reached this level of play.

Any developer(s) to carry out this project ?

[Jouni]

Solver don't need to be strong in game play. It's not any problem if it can't play at all. It only solves.

[Vinvin]

Solver don't need to be strong in game play. It's not any problem if it can't play at all. It only solves.

I don't agree. A high level of play is a nice plus.
Some positions win a small advantage, so if the engine is weak it will not even win the game after it found the winning move.
And as I wrote, the current "solvers" are already at this strength.