Perfect play

[mmt]

The perfect player does not necessarily have to just choose a random move from TBs. I wasn't thinking about random moves in the OP, I was thinking more about how the best chess program from year 100,000 might do. It will try to maximize its score against weaker players but never lose. What's the percentage then if it chooses positions that make it hard for white?

Note that once the losing position happens, it's over. The engines might be able to save a losing position that they see as -1.0 against each other but not here. If we switch sides, the safety margin becomes quite a bit less for black.

[Spliffjiffer]

i dont want to entighten any prospects towards solving sth but i think that we are not yet there to enlight whether if this or that is consequent here...chess is highly too difficult to get truths in a stadium where we are at 2. g5?!, or sth else....its absolutely fine for me to discuss these things but to come to conlusions will take still a lot of investigations and time and this will not be solved by running an opening expert like Leela for 7 days imo...all we have until now is to believe in one or the other thing

[mmt]

Well no, we won't find out exactly. But an educated guess is interesting and more fun to argue about on message boards than if we knew exactly anyway.

[Ovyron]

I was thinking more about how the best chess program from year 100,000 might do.

It's possible that by then you can beat any fallible opponent 100% of the time, by simulating the relevant parts of the universe where this entity exists, and just play random games until one beats them, then replay those random moves in reality.

You might want to go back to year 3000 or an earlier point where simulating universes isn't possible yet.