Correspondence chess in the age of NNUE

Since some on CCC want to use my data to propagate clearly wrong assumptions on the strength of NNUE, and what this means for Chess. And I have banned myself from CCC for refusing the change my Signature. I will address this here.

1. NNUE is a stupid type B search.
2. Because NNUE can not beat itself even with a 10x speed advantage. This does not mean chess is a proven draw, or that NNUE plays near perfect chess.  
3. I have shown that even given hours per move on very fast hardware. NNUE makes many mistakes that can be proven against perfect play that we can prove. Meaning 7 man table bases.

Conclusion. Because NNUE and humans in correspondence time controls on very fast hardware can not beat themselves. This tells us nothing. As a type B search is not perfect play, and perfect play clearly exposes the flaws in NNUE. We know for a fact that NNUE is not perfect play in chess. So all this shows us is that NNUE is better then humans, and NNUE is weaker then perfect play.

mwyoung wrote:

And I have banned myself from CCC for refusing the change my Signature.

Who pushed you?

BTW, NNUE replaces the evaluation function of a chess program, nothing more. And because of that search will make better (pruning) decisions, 2 flies in one catch.

I think the key questions are:

1. is a particular computer unbeatable?

This can be disproved by finding a way to beat that computer.

2. Is chess a draw?

For me, the evidence strongly suggests that it is (e.g. there is no known way to force the win of material from the starting position), but it's not proven.

3. How close are we to chess computers being unbeatable?

We can't say for sure unless we can prove that chess is drawn!  However, the rising level of draws at the top level of correspondence chess, to the point where this kind of competition seems to be losing its viability, seems to imply that we're close to "almost unbeatable". There is also a rising number of draws in unrestricted* tournament computer chess.

*include opening books and tablebases.

I doubt if strong engines are usefull. For instance there are variants that are playable according to stockfish but not playable for humans for they are tactically too complicated.

I remember Smyslov not playing a variation he invented himself. He abandoned it for it got too complicated.

I think it was Smyslov variation of the Grunfeld.

Admin wrote:

mwyoung wrote:

And I have banned myself from CCC for refusing the change my Signature.

Who pushed you?

BTW, NNUE replaces the evaluation function of a chess program, nothing more. And because of that search will make better (pruning) decisions, 2 flies in one catch.

I do not know, but after weeks and weeks with the same signature. I was banned, and that is good with me. As I am not changing it, as it was 100% true.

And I did not contact anyone at CCC, as I am done with them.

"Information
You have been banned from this board until Mon Feb 01, 2021 9:43 am.

Please contact the Board Administrator for more information.

Reason given for ban: quarrelsome signature

A ban has been issued on your username."

Here is my signature "The worst thing that can happen to a forum is a running wild attacking moderator(HGM) who is not corrected by the community." - Ed Schröder
But my words like silent raindrops fell. And echoed in the wells of silence."