Back to Chess Programming after 20 years

[CaptainCheckmate]

Hello everyone,

I just joined, thought I'd say hello and introduce myself.

I wrote my first chess engine 19-20 years ago as a teenager. (Still available for download here: http://hamedahmadi.com/ai/HamedChess.zip)

I didn't know about chess engine UI's so I just put something together in visual studio (images borrowed from chessmaster). I estimate it plays around maybe 1400 ELO, with no opening book or endgame tables, so anyone with the discipline not to hang pieces will win. It would probably be stronger if it searched deeper, but I wrote this on a 200mhz computer from 1997.

I'm looking into writing version 2.0, 20 years later, mostly as an opportunity to play around with some of this newfangled neural network stuff. I'm also curious about the applications of engines toward opening theory, and for teaching/training. (e.g. I think it would be interesting to investigate if one could produce an engine that can concisely explain/demonstrate "why is this move a bad move?") (My other "hobby" in high school was automated theorem proving so this idea brings 2 disciplines together)

If anyone has any thoughts, please feel free to share.

H.

(Funny story: if you search "introduction to game algorithms" the article I wrote about alpha beta search in 2002 still shows up on the first page of google. I don't know why, it's not even a good article by today's standards...)

[Graham Banks]

Welcome back.

[maksimKorzh]

Welcome back)

To play with NNUE you need really fast search. Stockfish NNUE without optimizations strips down a few plies from the search. So you need at least 7-9 plies deep search in blitz - after adding SF NNUE it would be 4-5 plies if no optimizations are used.

[Ckappe]

It's exciting times for AI and the advances the last few years have really been mindblowing... Welcome back