Annotated Game #228: Offsides knight, or what was the plan again?

This next game continued the downward tournament trajectory, which had started with a relatively clean game and win in the first round (Annotated Game #225), followed by a shaky win, followed by a shaky draw. The below is a rapid loss in a Classical Caro-Kann, which really shouldn't happen, but is instructive to see. White plays solidly and deviates in the opening, a situation highlighted in "Common opening repertoire pitfalls" - in fact, I had this game in mind as a prime example.

I do well in responding as Black through move 12, finding appropriate development for my pieces that both takes into account and takes advantage of White's play. Once I am faced with finding a plan on move 13, however, instead of the thematic ...c5 break I choose to simply exchange off my opponent's pieces. Mindless exchanging is a common error of Class players and in this case it simply improves my opponent's position while denuding my kingside of defenders. The most egregious fault, however, is placing my knight on b6, where it stays offsides for the remainder of the (short) game. I think that Caro-Kann players need to be very careful about any ...Nb6 ideas, as it can be seductive to think about repositioning it to the more central d5, but White often can either kick it with c2-c4 or just keep it shut off from the action. Props to my opponent for quickly taking advantage of my deviations from what the position demanded.

[Event "?"] [Site "?"] [Date "????.??.??"] [Round "?"] [White "Class B"] [Black "ChessAdmin"] [Result "1-0"] [ECO "B18"] [Annotator "ChessAdmin/Komodo 11.2"] [PlyCount "43"] {[%mdl 8192] B18: Classical Caro-Kann: 4...Bf5 sidelines} 1. e4 c6 2. d4 d5 3. Nd2 dxe4 4. Nxe4 Bf5 5. Ng3 Bg6 6. Nf3 Nf6 {this is a move-order finesse, taking advantage of White's 6. Nf3 instead of the standard h2-h4. Now Black has the option of playing ...Nh5 to block the h-pawn advance and exchange the Ng3.} 7. Bd3 {a solid move that is very common (second choice in the database), just behind the more theoretical h2-h4.} e6 {in positions with a Black pawn on h6 (typically played as a response to h2-h4), it is essentially obligatory to exchange on d3. Here it is not, which is one of the main positional differences. Instead of helping develop White's queen with the exchange, the freeing pawn move is played. Incidentally, almost all of the top-level games in the database feature the text move.} 8. O-O Bd6 {another difference is that the bishop can be developed here rather than e7, preventing White's bishop from going to f4 and getting on the more active h2-b8 diagonal.} 9. Re1 O-O 10. Bxg6 {the most common plan here. The White bishop will get no better and exchanging itself for its counterpart on g6 allows the infliction of a minor weakness on the h-file for Black. This should not be a real problem, but it ends up being a losing one for me, as we shall see.} hxg6 11. Bg5 Qc7 {this takes advantage of the doubled g-pawns, as White exchanging on f6 will no longer inflict positional damage. In fact, it would help Black by controlling e5 with a pawn and allowing for ...Kg7 followed by putting a rook on the h-file.} 12. c4 Nbd7 {developing the last minor piece.} 13. Rc1 (13. c5 Bf4 $11 {poses no issues for Black.}) 13... Bf4 {this is not a bad decision, but the follow-up is poor. It is unnecessary to force the issue of the Bg5 right now, and Black could gain more activity with the thematic ...c5 break.} (13... c5 14. dxc5 Bxc5 $11) 14. Rc2 (14. Bxf4 Qxf4 15. Ne2 Qc7 $11) 14... Nb6 $6 { this actually worsens the position of the knight, which no longer influences e5 and c5 or supports the Nf6.} (14... Bxg5) 15. Ne4 $14 {my opponent finds the best way to take advantage of the offsides knight.} Nxe4 {this worsens the situation by removing all of Black's minor piece defenders on the kingside and helping White's piece activity.} (15... Bxg5 16. Nexg5 Nbd7 $14) 16. Rxe4 Bxg5 17. Nxg5 Rfe8 {not admitting the mistake with the knight, as it would be best to bring it back immediately. The text move has the idea of supporting e6 and giving the king a square on f8, but too much of Black's army is still sidelined.} (17... Nd7 $5 $14) 18. Qf3 $16 Qe7 19. Qe3 {supporting the Ng5 and increasing pressure on the e-file.} Rad8 $2 (19... Nd7 {is necessary to defend. I underestimated the danger to the king, which now immediately manifests itself.}) 20. Rh4 $18 Qf6 $4 {a blunder in a bad position, comments Komodo via the Fritz interface. I saw that the king had no squares and would need an escape route from the h-file threat, but this places the queen on a vulnerable square.} (20... Qc5 {is the engine line, an active defense based on White's back-rank vulnerability. White is still much better, however.} 21. Rc1 Qf5 22. g4 Qa5 23. a3 $18) 21. Qh3 Kf8 22. Nh7+ (22. Nh7+ Ke7 23. Qa3+ Kd7 24. Nxf6+ gxf6 25. Rh7 $18) 1-0