Idea For A Way To Win Today

[towforce]

This obviously falls into the category of "easy to say, difficult to do", and if it does work, it won't work forever (death by draw is coming), but it seems to me that it might work right now.

Today's chess programs:

1. Don't know everything

2. Cannot search deeply enough to cover all the gaps in their knowledge

So... to exploit these weaknesses, we need to find positions that are not well known, and are difficult to search from. That screams "complex". Get programs to prefer positions where the branching factor is high, to throttle off search. It also seems likely that such positions will be unusual enough that other programs won't know enough about them to be able to play them without making mistakes.

I remember in the 1990s the keywords for humans playing computers were "blocked centre", to kill the computers' search advantage. Maybe today, the keyword can be "complex".

[Chessqueen]

Today the key is to always find an engine that is 50 rating below you

[towforce]

That's a good plan, and I'm hoping that the idea of going for complexity could help somebody achieve that!

[Werewolf]

It's getting nearly impossible if the engine has decent hardware, decent time and agency over its opening choices.

In terms of complexity, I don't know of anything more complex than the KID or the Frankenstein Dracula, both of which the engine plays well now.

If I was seriously trying to beat an engine I still think a blocked centre would be the way to go.