Is all chess POSITIONAL?

I dont understand what you mean here. My statement was correct, if your opponent is threatening a fork, you cant make a slow improving move whilst ignoring your opponent

What I mean is when it comes to your examples you are correct, but that doesn't prove these words couldn't be some what correctly used in some other scenarios, which I described.

MCO formal Chess Opening for every move there is a line to follow. Most of the opening found in that book are "formal," formal chess. Pieces follow pieces (specific ones at that) as soon as you get out of the book for a preferred position then: doesn't that become positional?

And isn't this purely a chess theory question?

Wouldn't GM Dubov be someone that more often that not prefers to deviate and emigrate to certain positions in order to surprise or challenge the opponent?

I have more questions...

There's no embarassment here. I simply view you as a player who has much to learn about the game.

The irony is that you've chosen to show a match where Komodo defeated a human GM, as if this proves the insignificance of positional understanding.

This is ironic because Komodo is well-known to be one of the most "positional" engines there is, both in playing style, and in evaluation preference.

About Komodo Monte Carlo:

"Komodo gains an edge over conventional brute force engines because of its positional style of play and the fact that it relies on position evaluation over depth ..."

About Komodo Dragon:

"The added NNUE technology allows Komodo's engine to integrate the deeper positional understanding possessed by neural network engines such as AlphaZero and Leela Chess Zero."

Source: https://www.chess.com/terms/komodo-chess-engine

All you've done is support my original example, where I pointed out that AlphaZero (the original self-taught NNUE chess entity) defeated Stockfish using its self-learned positional intuition.

Stockfish's calculating abilities (80-million positions per second) faltered against AlphaZero's self-taught positional understanding, despite the fact that Stockfish's calculating abilities were 1,000 times stronger than A0.

Stockfish could calculate far deeper than AlphaZero. But AlphaZero didn't need such calculation powers - it relied, instead, on its own positional understanding, from millions of self-play games. It understood, from past experience, what kinds of positions were desirable or not, without needing the deep calculation that Stockfish relied on.

The strength of such positional understanding was not lost on Stockfish developers. They scrambled to add their own NNUE evaluation to Stockfish, soon after the AlphaZero match.

They recognized a superior chess approach when they saw it.

When they finally succeeded in adding NNUE capabilities to Stockfish (in 2020), they announced a large strength improvement in the engine - as a result of the improved positional understanding that NNUE technology provides ...

They even used an exclamation mark when announcing this improvement, from how excited they were about the development.

Oh, but I couldn't possibly understand "Higher Chess".

I'm lower species...

Didn't notice this comment until now. Okay, let's continue.

I actually have no need to Google AlphaZero. One of the many chess books on my shelves is Game Changer by Matthew Sadler and Natasha Regan, which I've read cover to cover several times.

I'm not an expert on the subject by any means, but I know a thing or two about it.

From what I know: the strength of AlphaZero came not from its calculation abilities (which were significantly weaker than Stockfish's), but from its extensively trained neural network.

The hardware differences between SF and A0 were decreased in the subsequent 1,000-game rematch, which was played in time conditions that mimicked the TCEC. SF had 44 CPUs. A0 had 44 CPUs and 4 TPUs (the 4 TPUs were necessary to efficiently handle its monolithic, deep-learning neural network).

Again, I'll mention: Stockfish developers scrambled to add NNUE technology to their engine, following the AlphaZero match. If we ask ourselves why this is, we might start to appreciate the positional value that neural networks provide.

The whole point of neural network technology is improved positional evaluation and insight. It's not about calculation - it's about enabling the engine to consult a pre-trained network to identify positional patterns and long-term strategic ideas, based on its huge dataset of completed games.

After Stockfish added NNUE technology to its engine, its performance leapt up by around 100 Elo. This is noteworthy because it reflects a similar Elo disparity between AlphaZero and SF in their matches.

This demonstrates that it wasn't hardware that proved the difference - but the improved positional insight that comes from a neural network-aided approach. Once SF adopted similar technology to A0, its own strength of play improved, accordingly.

The word "positional" comes up again and again ...

As far as the match between Komodo and Nakamura goes: that was a match between a 3400 player and a 2700 player. That's a 700 point difference in strength. 2-pawn odds aren't enough to compensate for a strength differential of that size, IMO.

As mentioned earlier, 2-pawn odds are, arguably, the rough equivalent of anywhere between 60 and 200 Elo.

Pit me, a 2300, against an accurately-rated 1600, with 2-pawn odds given - and I'll crush the 1600 by a similar margin. (Or vice versa, a 3000-rated player would demolish me at a similar ratio.) It wouldn't really prove much, either way. A stronger player is going to defeat a weaker player, regardless.

All in all, when it comes down to it, we're probably not going to reach an agreement here. I believe positional understanding matters. You seem to be convinced of the opposite. Neither of us are going to change each other's minds, so it's probably best to move on ...

For a human the big picture is positional and the small picture tactical. And for a human the positional part have to be learned and much of the tactics trained. That is what we are dealing with here at Chess.com. What the engines does is not very interesting when you play another human being; it is a world of itself. I know because I have been playing correspondence chess with engines as long as they have been around and for a long time at a decent level.

I would have been interested in the ideas. Maybe I would have thought they were rubbish but equally they could have been thought provoking. Best not to go in like the proverbial ton of bricks? A tonne is slightly less than a ton.

I've responded to all your points, including your example of a match won by an engine over a human, involving an engine that is well-known to be a positional engine.

I've also shown:

1) AlphaZero defeated Stockfish in two seperate matches, using the first entirely self-taught neural network approach to chess.

2) Stockfish developers added NNUE capabilities to their engine, following the A0 matches, proving that Stockfish developers recognized the power of AlphaZero's neural network approach to chess.

(It's interesting, to me, that Stockfish supporters are so quick to dismiss AlphaZero's arrival - meanwhile, Stockfish's own developers were so eager to adopt their own version of A0's approach.)

3) I've talked, in multiple posts, about the positional benefits that a neural network approach provides. One main benefit of a pre-trained neural network is the holistic positional view that it offers. With a huge dataset of high-level games to reference, the NN is able to identify positional patterns and strategic motifs that relate to the current position. This helps the engine identify promising candidate moves and lines, and can even enable it to generate its own positional insights and move preferences, accordingly.

It's an entirely different, positional-based approach to evaluating, compared to the traditional, hand-tuned evaluation algorithm - and you should know this as a computer chess expert.

4) AlphaZero's strength is directly relative to its neural network. Its play reflected the level that could be reached without any external knowledge or input. It's only knowledge of the game came from self-play, and nothing else. It wouldn't know what Stockfish is, nor what Stockfish's games looks like - as the games it played against Stockfish were not part of its training set.

One could speculate how strong AlphaZero (which is now obselete and was since replaced by the even more impressive MuZero) would be, were it to train itself by specifically playing against Stockfish as a direct opponent.

Could A0 eventually teach itself to defeat today's Stockfish? In theory, yes - it would eventually learn to defeat any modern engine you trained it against.

But that would defeat the whole purpose of A0 in the first place - which was to show what a self-learning algorithm could accomplish, without any external knowledge or input at all.

Stockfish, by contrast, is a Frankenstein-esque engine that has been continually tweaked and modified by countless contributors and developers over the years. This is part of its tremendous success - but also what makes it a completely different kind of chess entity to A0. They're almost incomparable, in their individual approaches.

Stockfish has always been a juggernaut, but it didn't truly "level up" until it adopted its current, AlphaZero-inspired NNUE approach to chess, as described in point #3.

Anyway ... consider this my last post on the subject. You're getting contentious, so there's no point in continuing the discussion. It'll only serve to aggravate us both.

Dismiss me as a player who "knows nothing", if you like. Adieu.