What Are the New Ideas That Chess Engines Introduced in Chess Theory?

[AntonisCh]

Reading articles, books or posts about chess engines, I usually stumble upon two main punch lines:
a. "Yes, engines are good but still they lack of fantasy".
b. "Chess engines have changed the way we play chess because they can see deeper and introduce new ideas in chess".

As a custom, the above chess mantras are enriched with some very spectacular examples-diagrams from Tal's famous sacrifices or some Lc0 vs SF12 games to prove the a or b.

But my question is this:
What are specifically the "new ideas" introduced from chess engines? Nimzowitsch had a couple of new ideas, Fischer too, Petrosian gave his two cents but engines?
Can you name one or two really modern chess ideas coming from chess engines?

And if you really can do that, can you still accuse engines for lacking of fantasy? New ideas cannot come from a fantasy scarce world.

[gaard]

Reading articles, books or posts about chess engines, I usually stumble upon two main punch lines:
a. "Yes, engines are good but still they lack of fantasy".
b. "Chess engines have changed the way we play chess because they can see deeper and introduce new ideas in chess".

As a custom, the above chess mantras are enriched with some very spectacular examples-diagrams from Tal's famous sacrifices or some Lc0 vs SF12 games to prove the a or b.

But my question is this:
What are specifically the "new ideas" introduced from chess engines? Nimzowitsch had a couple of new ideas, Fischer too, Petrosian gave his two cents but engines?
Can you name one or two really modern chess ideas coming from chess engines?

And if you really can do that, can you still accuse engines for lacking of fantasy? New ideas cannot come from a fantasy scarce world.

I don't know how one would begin to separate the advances in chess theory over the past (say) 20 years from chess engine usage. My understanding is that most, if not all, 2700+ GMs use chess engines as part of their preparation. If you mean something more general than opening theory and preparation than I'd need to know what you consider as new ideas from Petrosian, Nimzowitsch and Fischer.

[JohnW]

What is an example of a "fantasy" in regards to chess?

[mclane]

Chess Engines Do Not play chess.
They solve test positions.

So there is no fantasy and no plan.

LC0 sometimes plays like a human. As if it sees an idea.
When the alpha beta have seen it too, they are most often dead.

When Kasparov played dedicated chess computers in a simultan in 1985 he said the dedicated chess computers have no plan.
He was right.
This was 36 years ago.
But it is still true.

[h1a8]

Chess Engines Do Not play chess.
They solve test positions.

So there is no fantasy and no plan.

LC0 sometimes plays like a human. As if it sees an idea.
When the alpha beta have seen it too, they are most often dead.

When Kasparov played dedicated chess computers in a simultan in 1985 he said the dedicated chess computers have no plan.
He was right.
This was 36 years ago.
But it is still true.

I call BS. Clearly alpha zero and leela demonstated in many games moves that had significant plans behind them. These games were analyzed very carefully. These moves were so deep that it was impossible for those engines to reach the final depth of all the variations leading to a massive advantage.


To answer the above question. The theory behind thorn pawns was improved upon if not a completely new idea. Some ideas with pawn breaks were discovered.

[jdart]

There was an interview with Magnus recently about a game in which he'd played an early g4, a novelty. He said, basically, I always check out the possibility of g4. Alpha Zero and similar NN engines have a penchant for pawn attacks, and the GMs have picked up on that.

But that's a general principle. When most players talk about "theory," they mean the state of the art knowledge of particular opening systems. For this general principles don't help you much. However every top player nowadays is analyzing with engines and these are constantly suggesting new moves and variations. I think it's fair to say most of the novelties you'll see in pubs like New in Chess or Chess Informant are computer-derived.

[Madeleine Birchfield]

Chess engines are indispensible in modern opening theory; it is one of the reasons why the Spanish is now considered to be a draw at the highest levels of grandmaster otb chess, due to the Spanish Berlin and Spanish Marshall that computers have shown to be superior for black than the traditional Closed Spanish which gives white a slight advantage.

[jp]

There was an interview with Magnus recently about a game in which he'd played an early g4, a novelty. He said, basically, I always check out the possibility of g4. Alpha Zero and similar NN engines have a penchant for pawn attacks, and the GMs have picked up on that.

But g4 (and h4) are very old, human ideas. At best, engines have emboldened players at the highest level.

[jp]

The theory behind thorn pawns was improved upon if not a completely new idea. Some ideas with pawn breaks were discovered.

What ideas with pawn breaks? Can you give some examples?

[jp]

And if you really can do that, can you still accuse engines for lacking of fantasy? New ideas cannot come from a fantasy scarce world.

There are different ways to skin a cat. The same result produced by a human might involve "fantasy" but produced by an engine is simply the product of mass calculation.