The current strongest chess engine

[avo_k]

If Alphazero or Muzero were to rematch against Stockfish 13 or even the last Leela nowadays, would they be crushed?

Is it difficult to estimate the strenght of A0 against today's engines? Can we still say that A0 is the strongest chess engine even if its last games are several years old?

Thanks in advance for your answers.

[amanjpro]

Alpha Zero doesn't even have a time management... I doubt it stands a chance if the rematch happened today.

Even in the first match, TC was move/second instead of normal TC formats


Edit scrap what I just said... The second match actually used a proper time control

[lucasart]

If Alphazero or Muzero were to rematch against Stockfish 13 or even the last Leela nowadays, would they be crushed?

Is it difficult to estimate the strenght of A0 against today's engines? Can we still say that A0 is the strongest chess engine even if its last games are several years old?

Thanks in advance for your answers.

There never was a match between A0 and SF.

A0 was never the strongest chess engine, because it's never been a chess engine. It was only a proof of concept.

For A0 to be a chess engine, a version that can run on consumer hardware would have to be released, so that both programs can compete on the same machine, and so that the match can be run independently. And a bunch of small trivial details like protocol interface and time management (btw everybody seems to focus on that which is missing the point).

The only thing that we have is a paper (really a PR stunt) by Deepmind, which supposedly reports the result of a match between their proto-engine, and an old SF, using a completely ridiculous hardware advantage for A0 (TPU vs CPU…), and very artificial / biaised testing conditions.