1...e5 is better than 1...e6 and 1...c5, prove me wrong

1... e5 in no way prevents white from playing d4, google Center Game

Proving in a different way that the OP could be wrong.
1. ... d5 is best;
1. ... c5 is second best;
1. ... e5 is in third place.

(-0.3) Stockfish 14+NNUE, Depth 40 CLOUD

(-0.3) 1...d5 2.Nf3 g6 3.d4 Bg7 4.Bf4 Nf6 5.e3 O-O 6.h3
(-0.2) 1...c5 2.e4 g6 3.d4 cxd4 4.Qxd4 Nf6 5.Nc3 Nc6 6.Qd2
(-0.1) 1...e5 2.c4 Nf6 3.Nc3 c6 4.e3 d5 5.d4 exd4 6.exd4
(-0.1) 1...Nf6 2.d4 d5 3.e3 c5 4.Nf3 cxd4 5.exd4 Bf5 6.Bd3
(-0.1) 1...e6 2.Nf3 d5 3.e3 b6 4.d4 Bb7 5.Nbd2 Nd7 6.Bd3
1 ... How did I use the Lichess analysis to generate the above? Simple ...
Solution: Set-up is Black plays first, and then press analysis board. @Alcadeias

lichess.org/analysis/rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR_b_KQkq_-_0_1?color=black

Toggle local evaluation on to see the results. Post if your's are different.

I played black first and rewrote the moves as if it was white that was playing first. So you now see it converted to white in the study. 1. ...e5 was not played on the first PV engine line. It was played on the 2nd engine PV 2 line.

PV1: 1. d4 Nf6
PV2: 1. c4 e5
PV3: 1. e4 c6
PV4: 1. Nf3 c5
PV5: 1. e3 d5 Best Accuracy 99%

After each move 5, there is a near perfect game to view from the GM database.

I have reached my daily maximum for links in forum posts. So just copy past to see the study:
/study/3X7g3ZLt/eCi8WsHQ

@Toscani said in #28:
> I think so, but will the lichess analysis present more than 9 moves? What's it's limit?

I have been discussing this before in some thread. While SF can display in command line output all the plies (I think even those with extended quiescence searches) in its output, because of some notion of instability of deep searches beyond depth 20 or something like that (there is a recent github issue with such a statement answering that clearly, but i don't recall the exact depth).

Lichess as I understand, because of that instability of the PV at deep depths, but apparently not of the score attributed at current position being analysied by engine, has chose a maximal PV depth of 16 half-moves.

And if deeper searches than the local engine does by default are being uploaded to the lichess CLOUD, then those would superceded local engine analysis, unless asking for more depth. And as they superceded, for some reason the PVs depth displayed becomes even shorter than 16 plies.. sometimes 10. (trigger local engine deepening would bring back the 16 plies, but some kind one increasing work upon reloads.

It would be nice to find actually data analysis for such uncertain deep search behavior, with many subtree searches, for SF, so that lichess would have better data than some accepted statement of "instability" at depth.

precision and accuracy notions could be investigated.

But it does bring the question of using engine scores for very early positions. Is the stability of the score upon deepening equivalent to accuracy increase? What is a 0,1 pawn or centipawn value, is that within the margin of errors (which would need to be measure one day, over some database of search trees over many positions covering big enough chess).

c5 is a cheap trick (Larsen).

"... [1 e4 c5 2 Nf3 d6] Almost everybody plays [3 d4]. But isn't this a positional error? I am not joking. I like my center pawns, and I like a [d-pawn] better than a [c-pawn]! I know that sometimes White sacrifices a Knight on [d5] or [e6] and smashes Black before he can castle, but in those games where this has been done, haven't improvements always been found for Black afterwards? Well then, isn't [3 d4] something like a cheap trap? I know it can be combined with purely strategical ideas, but I find it easier to discuss strategy when I have an extra center pawn! I cannot explain where I first got this idea, but I had it before I was Danish Champion and regularly playing [3 Bb5+]. ..." - GM Bent Larsen (1974)

@Alcadeias said in #1:
> Both 1...e5 and 1...e6 2.d4 d5 accomplish the following three things:
> - occupy the center with a Pawn
> - open diagonals for the Queen and the King's Bishop, help to develop these pieces
> - get Black one move closer to castling
>
> But the difference is that 1...e6 2.d4 d5 accomplishes these three things in two tempos instead of one, so it makes Black lose one tempo. Why would you do something in two tempos when you can do it in one?
>
> Other differences between 1...e5 and 1...e6:
> - 1...e5 prevents White from playing d4
> - 1...e5 allows Black to develop his King's Knight to it's optimal square, the f6 square, without fear that White will kick it away by advancing his e4-Pawn to e5
> - 1...e6 blocks the Queen's Bishop, turning it into a bad Bishop
>
> 1...e5 is better than 1...e6 in every single way.
>
> 1...c5
> - does not occupy the center with a Pawn
> - does not help develop any piece (except the Queen, but there's an opening principle that says "don't bring your Queen out too early")
> - does not get Black any closer to castling
If e5 is strictly better than e6, then why do GMs play it? Are they stupid or is it maybe that an opening cannot be analyzed so easily?

Nah, GMs are just stupid. They can't even get a win with scholars mate.

@PopeonDope said in #39:
> Nah, GMs are just stupid. They can't even get a win with scholars mate.
Riiiight...