How to realize this +2 evaluation in the Ruy Lopez, Steinitz defense

Question

I played a game as White using the Ruy Lopez: Steinitz defense against a lower-rated player (1594 vs. 1500 on chess.com). While I won the game due to a blunder by my opponent in a drawn endgame, the engine gives an almost +2 evaluation in the opening (+1.7 by move 10). But my advantage slowly decreased until it was a dead drawn 0.0 by around move 15. This was a 10 minute game, in case that is relevant.

[Event "Live Chess"]
[Site "Chess.com"]
[FEN ""]
[Date "2021.05.21"]
[Round "?"]
[White "Me"]
[Black "NN"]
[Result "1-0"]
[ECO "C62"]
[WhiteElo "1594"]
[BlackElo "1500"]
[TimeControl "600"]
[EndTime "22:27:44 PDT"]
[Termination "White won by resignation"]

1. e4 e5 2. Nf3 Nc6 3. Bb5 d6 4. d4 exd4 5. Nxd4 Bd7 6. O-O Be7 7. Re1 Nxd4 8.Bxd7+ Qxd7 9. Qxd4 Nf6 10. c4 c5 11. Qd3 O-O 12. Bg5 Rad8 13. Bxf6 Bxf6 14. Nc3 Bxc3 15. Qxc3 b5 16. Re3 Rfe8 17. Rg3 g6 18. f3 bxc4 19. Qxc4 d5 20. Qxc5 dxe4 21. fxe4 Rxe4 22. Re3 Qd4 23. Qxd4 Rdxd4 24. Rxe4 Rxe4 25. Kf1 a5 26. Rc1 Re5 27. Rc7 Kg7 28. Kf2 h5 29. Ra7 Kf6 30. Ra6 Kf5 31. Ra7 f6 32. Rh7 Re4 33. Ra7 a4 34. Ra5+ Ke6 35. b3 axb3 36. axb3 Rb4 37. Ra3 Kd5 38. Ra5+ Kc6 39. Ra6+ Rb6 40. Rxb6+ Kxb6 41. Ke3 Kb5 42. Kd4 Kb4 43. Kd5 Kxb3 44. Ke6 g5 45. Kxf6 g4 46. Kg5 h4 47. Kxh4 g3 48. Kxg3 Kc4 49. Kf4 Kd5 50. Kf5 Kd6 51. Kf6 Kd7 52. g4 Ke8 53. Kg7 1-0

My justification behind trading the dark-squared bishop was to have the d5 square for my knight. However, it pretty clearly backfired quickly as my opponent just made the same trade as soon as it was possible. Then we were back to a fairly symmetrical position. I'm assuming the evaluation came from the d5 square for my knight, but otherwise, I don't see how to make use of this advantage. By playing out some alternative moves with the engine, my opponent's dark-squared bishop makes it to d4 and they get their own great piece.

What was the correct plan here for White to keep this advantage?

Answer 1

First things first, I don't believe thinking in terms of computer evaluation is useful at all for humans. We don't look at a position and go "Well, I guess that's about a +0.54".

Instead, I'd focus on the key elements of the position. By move 15, Black has a weak pawn on d6 that will probably stay weak for the entire game, while the d5 square is a great outpost for the White pieces. This reasonning leads us to what went wrong. Instead of keeping control of the semi-open file and weak square, you went on an attack on g7. That's okay, but you should have consolidated your position first. Moves like Rad1 and b3 would have been useful.

By move 21 your position has transformed. You're a pawn up but your pieces are poorly coordinated and Black controls the newly open files. Most endgames you will get into will be drawn with correct play and if anyone has a chance to create an attack, it's definitely Black.

Answer 2

Your idea was completely correct: Exchange on f6 and implant a monster on d5. Unfortunately, that knight only says "Boo." and doesn't guarantee you a quick win at all. And if you let Black trade it (you still could have played Re2 instead of Nc3, and try to play N-a3-c2-e3-d5, possibly prepared by g3 and f4 if the B tries to annoy you from g5), your advantage reduces to the backward pawn - since Black can easily bring as many defenders as you attackers, this is still only a theoretical advantage.

In both cases, traditional wisdom says in both positions you must open a second front somewhere else. Black, bound to d6 or restricted by Nd5, can't regroup his pieces as quickly as you.

That said, I think already Bg5 might have been a bit premature on the grounds that Nc3 (instead of Bxf6) can be answered by Ne8 followed by Bf6 if you refuse the trade, and Nd5 fails to the hanging Pb2.

Additional suggestion: Find other (LiChess?) games with this position where White brought his advantage home (without Black blundering away material). Study them.

Answer 3

Although I in principle agree with other answers, I'd like to double down on one particular idea which did not work in your game. Clearly, after c5 you played Bg5 and Bxf6 to establish a knight on d5. This looks very reasonable. Yet, you first take on f6 and then play Nc3, allowing Bxc3 after which you have no knight left, and the weakness of the d5 square cannot be exploited so easily. It might be smarter to play Nc3 first and then try to see if you can exchange your bishop for the black knight.

Please note, I am not saying that the position after, for example, 14. ... Bd4 (instead of Bxc3) 15. Nd5 is won for white (although it is nice to play). Just wanted to highlight that you had a reasonable plan and unnecessarily allowed your opponent to stop you from executing it.

Answer 4

OK, I disagree on two counts :-) First, I do think computer evaluations are useful, provided you use them sensibly and accurately.

Sensibly: agonising over +0.54 for this move versus +0.61 for that move is pointless. But if the evaluation has gone from +2 to =0, that's a pretty significant change, and it's a clear sign that something went wrong between those two points.

Accurately: are you by any chance using the evaluations from chess.com's automatic analysis, which spends only a fraction of a second on each move? When I run your game through an engine on my desktop computer at 20 seconds per move, I get quite different answers.

This leads to my second point: your advantage didn't "slowly decrease". With the more accurate evaluations, it more or less holds steady and then suddenly drops at each of 12.Bg5 and 13.Bxf6. And you didn't actually have a +2 advantage to start with.

To be specific, the numbers I'm seeing are:

• After 10.c4, +0.5.
• After 10...b5, +1.5. Black made a mistake here.
• After 12.Bg5, +0.9
• After 13.Bxf6, +0.1
• After 15.Qxc3, +0.3. Stockfish isn't keen on black's Bxc3, but it's not a terrible move either. The position at move 15 is pretty much level, but not a "dead draw", there's still plenty of play here.

Running the engine on different hardware or for different amounts of time might change the exact numbers, but the basic story is the same. And others have already give good explanations of why certain moves were not the best. My point is that the engine can give you some useful pointers (if not the full story), but you could think more critically about how you're using the engine and interpreting the numbers.