To set some test positions you need experience: chess player (ELO 2000+) + study composer (enthusiast). And welcome: programmer (heuristics or/and NN) + HAI (human autonomous intelligence).
It is very important to know the real value of the tested positions (valid or statistical).
Sting (based on the very old Stockfish) is intended only to demonstrate the capabilities of heuristic algorithms for AI (autonomous intelligence), not to prove tactical abilities in statistical positions.
The “IQ” test aims to show the engines’ ability to manage with “skepticism and intuition” – two real important AI problems.
Is it possible to create a solid (safe), common AI algorithm (choice), free from the individual logical properties of problems? The chess zone can be very helpful.
Sting SF Black Hole 10
Moderator: Ras
-
- Posts: 1169
- Joined: Sun Feb 14, 2010 10:02 pm
Re: Sting SF Black Hole 10
Maybe, I can't be friendly, but let me be useful.
-
- Posts: 1169
- Joined: Sun Feb 14, 2010 10:02 pm
Re: Sting SF Black Hole 10
I replaced the Sting-sf-bh-10x version with a new one (same link). One change, a new UCI option is added: OldSting. If true, Sting works without the Black Hole idea. Instead, the old idea is restored to neutralize the bad effect of null-move.lech wrote: ↑Sun Apr 06, 2025 1:37 pm There is a bug and it needs to be fixed: Sting SF Black Hole 10x.
---
download: https://www.mediafire.com/file/lf76ia96 ... x.zip/file
Sometimes it can be very useful. Especially when the Black Hole idea doesn't work within its limitations.
Maybe, I can't be friendly, but let me be useful.