Opening, Middle-game, or Endgame, Which is most critical for improving your chess and why?

There are AI engines ( Leela) and Traditional engines ( Stockfish) already.

Tests? 

Leela had 320 millions games trained and Stockfish had more than 2 billions games tested.

Strength and weakness

Stockfish is weaker in opening and early middle game(weakest) and stronger in late middle game and endgame.(strongest). Leela is strongest in opening and weakest in endgame.

How do you know? And how can I know?

Unless you see above testing data, you will hardly notice the difference. Assume, both are 3600, they might be playing 3500 in weaker phase and 3700 in stronger phase.

Interesting @drmrboss

If the test is done with two different engines, i think the result might be ambiguous. Because the two engines are programmed very differently. 

Better to use the same engine, just to change and set the programming for the test

Stockfish White: opening 3200, rest of the game 3500

Stockfish black: endgame 3200, rest of the game 3500

I think this would produce some appropriate results.

What do you think?

When I started on chess.com, I was previously on hiatus OTB for two years. I had no idea about any openings, but was able to cheese my opponents with knight forks (and that's literally it), so I stayed in 1200 range.

Then, I started to learn a variety of openings, usually with many traps but still lead to a reasonable position if all traps are avoided, and my rating slowly climbed up. I feel like many people say "oh, just learn opening ideas, you'll be fine".

No, you will not. I feel like many people under-emphasize the opening... I know that when people played 1.d4  before my 2 year break (I was 800 OTB then), I would play 1...d5 without hesitation because "what am I supposed to do, play Nf6? What does that do? pffft". 

Now I have a fat endgame book, but am still mainly working on how to apply principles of chess and how to think ( like "I played sound developing moves, so why should I be defending on move 10?" kind of things). 

Of course, the OP has a high enough rating that he/she won't really need to grind openings....

1...d5 is a GM level move, I don't know what you're talking about

For years, in bullet and blitz, I premoved 1...d5

For a while that was also my OTB tournament move against everything.

Even if there are two machines, you can have comparable identical strength.( after testing thousands of games).

For example, in chess.com machines, GPU for leela and CPU for stockfish are in almost equal strength.( within 5 elo)

Testing against different program is practically easier than testing against self.

You can test Stockfish's weakness by Stockfish if you give the mentor Stockfish  by giving x50 long time or x50 cpu. 

I meant that I played d5 because I had no idea what I was supposed to do. like play a6?

Im not sure, but i think 1. Nf6 is a more common GM move than 1. d5

After studying openings, realised that nf6 is better. 

Didnt have any knowledge of nf6 before studying openings.

Also, there are a lot of positional yhemes in the opening.

I think thats the point @SNUDOO is trying to make

I think

Yes, in 1.d4 games I play 1...Nf6 now, but it hardly matters. You could walk into a world championship match armed with 1...d5 as you primary response against 1.d4.

@drmrboss

Even though their ratings are almost same, i think the result in such a case would be ambiguous.

Because the two engines have a different logical thinking process, due to the different way they are programmed.

But sure, I think it's useful for even very new players to memorize the first 5 moves or so of the mainlines of the openings they play.

"Understanding" the opening is important, but it's for relatively more advanced players. In the beginning you don't know much, so just memorize... but only a few moves, because your 800 rated opponents aren't going to go deep in to book anyway 

Yeah i guess it follows the logic-

Memorize the theory before the exam if you dont understand it.

You'd have to know the openings, though. Like what if white plays... Catalan? colle? London formation? Barry attack? (That's just a random name I remembered)

What will you play? Slav? (theory) QGD? what kind of declined? Colle-Zukertort formation? Tarrasch?

You will need to know what you want to do in the opening.

1...d5, e6, Nf6, Be7, 0-0 and then when it seems you can get away with it play c5.

That's something you can do almost as a system against anything, not just 1.d4

True, but there is a small dedicated army of catalan players XD

I know, right? Not so hard. Just remember it. Nothing to it at all...

Is the catalan somehow good against that?

I'm not trying to be coy, I'm just saying.

For example, this is completely legitimate, AFAIK

.

I was making fun about how "nobody plays the catalan"

fun fact, Hikaru actually said that on stream; he said, "The catalan? who plays the catalan?" lol