You are thinking in terms of strategy, by applying positional principles and your own experience.
Stockfish is "thinking" in terms of brute force calculation, by looking at 1.3 million positions per second.
Why would you expect the two methods of analysis to come up with the same suggested move?
On higher depth it shows that Nxd5 is better. +0.1 vs +0.4 for exd5 (if I wait until depth 30).
Thank you for your answers. Even though that ...Nxd5 might be better on higher depth, it still shows that ...exd5 is ok (really a minor difference). However in my analysis ...exd5 is a positional blunder. That's why I created this topic and why I seek to understand the merit of that move.
Maybe I asked a wrong question. Is ...exd5 a computer move justified by deep brute force calculation (e.g., in 30 next moves White could not capitalize on IQP and Black would hold the draw in all variants) and no good human would ever play it? In that case it would be ok for me and I would not think about it further. But if a stronger player than me would indeed play such move, I would really like to know why.
Well... I've got about 400 rating points on you (2350), and I consider this to be a "brute force vs positional principles" issue.
The only thing that the IQP has going for it in this position is the fact that White has the wrong Bishop for restraining the Pawn. That's the plus side. The minus side is that so many pieces have already been exchanged off that the position lacks dynamicism.
Like you, I would avoid the IQP if possible in this position, on the grounds that there isn't enough wood left on the board for Black to make proper use of an IQP.
@1
"He who fears an isolated Queen's Pawn should give up Chess." - Tarrasch
In this case white's light square bishop does not control the blockade square d4,
so black can always push d4, chasing or trading any blockers of square d4.
latest Stockfish does not think exd4 is better, from first iteration NNUE sees Nxd4 at 0.14 (getting better to 0.04) and exd4 at 0.35 (getting worse to ~0.5). Only SF classical sees exd4 slightly better for first 3 iterations.
exd5 does not affect black's structure. It can be defended likely, and opens up a file for rooks. Of course, you can just play Nxd5
@1
"He who fears an isolated Queen's Pawn should give up Chess." - Tarrasch
In this case white's light square bishop does not control the blockade square d4,
so black can always push d4, chasing or trading any blockers of square d4.
It seems I do not truly understand this concept of blockading the isolated pawn. It makes sense to me in positions where the pawn could actually move but in this specific position it cannot - it would be immediately taken by White's c3-pawn. True, White would then have his IQP, which I could easily blockade with my Knight, focus all my attention to it and maybe eventually capture it but that would just equalize material, not gain a plus. So is the aspect of White not being able to blockade my IQP really relevant in this position?
latest Stockfish does not think exd4 is better, from first iteration NNUE sees Nxd4 at 0.14 (getting better to 0.04) and exd4 at 0.35 (getting worse to ~0.5). Only SF classical sees exd4 slightly better for first 3 iterations.
I am sorry but this is actually not important to me at all. These are all small nuances and it could play a role in a game of 2 GMs or 2 engines (where even the slightest inaccuracies count). From my point of view the ...exd5 was a positional blunder, which would get me from equal position to a losing position (my reasons are listed in the original post). Stockfish told me that this was not the case and ...exd5 is similar in strength to ...Nxd5. Which one is sligthly stronger or sligthly worse is irrelvant to me. I wanted to understand why ...exd5 is not an absolutely bad move.
@9
"So is the aspect of White not being able to blockade my IQP really relevant in this position?"
++ That is the key to play against the isolated queen's pawn, as elaborated by Nimzovich in his My System. If black can advance d4 and trade off, then he is rid of the weak pawn.
If white had a knight instead of Bc2, then white could blockade the pawn d5 with a knight at d4 and after many trades blockade the pawn d5 with his king at d4 and then win the pawn.
exd5 does not affect black's structure. It can be defended likely, and opens up a file for rooks. Of course, you can just play Nxd5
I am really grateful to all of you guys for taking a look at this position and helping me understand it. I really appreciate it.
However this is also something I do not truly understand. I always thought that open files for Rooks are not important by themselves (just for the sake of allowing mobility for Rooks), but rather as means for Rooks to invade opponent's position.
In this specific position d-file is already opened for my Rooks and also c-file is semi-opened for my Rooks (so they could potentially apply presure on White's c-pawn and thus prevent his b-pawn from moving).
Why would I want to open e-file when the White Rook would already control it (and could invade to e7)? Wouldn't that actually help my opponent and not me?
We have been assuming that you would want to arrange your pieces so as to play d5-d4 trading off your IQP for the c3-Pawn. This should be possible if you're careful (ie: don't play Ne6 in a position where White can reply Bxg6), since you have a Knight that can cover the square, while White's Bishop cannot. That's why we described White's Bishop as "the wrong Bishop"... it can't cover d4, while your Knight can.
Why would you want to disolve the IQP? Because too many pieces have already been traded off. Counter-play with an IQP involves play on the two adjacent file with your Rooks, and creating a Knight outpost on one of the dominated squares (typically e4). This should generate some piece activity... which won't be worth much if too many pieces have ALREADY been exchanged.
More pieces on the board => favors the IQP
Fewer pieces in the board => favors the opponent.
1...Nxd5 is almost mechanical for any player rated between 700 and 2700.
1...exd5 is based on the fact that the Ng3 is misplaced, but this is only temporary, and I would not take the slight risk of creating an IQP out of nowhere.
And last, but not least: if your engine advocated 1...exd5, it should be high on something evil.
In my own analysis during the game I almost immediately chose (and played) 21...Nxd5, which is good according to Stockfish, but slightly better is 21...exd5, which I rejected during the game almost immediately for the following reasons:
Still, according to Stockfish 21...exd5 is a slightly better move than 21...Nxd5. I am profoundly confused and feeling like I hit my strategic boundary preventing me from improving. This is something I just cannot understand. Could someone please explain the strength of this move? Either by abstract concepts or specific analysis.