Here's the game (I was white):
lichess.org/B5HPzcoNAccording to the computer, I had a winning advantage from approximately moves 17 to 34 (peaking at +5.1 at move 29). However, I made a lot of mistakes and inaccuracies, and ended up lost. I tried looking at them, and I don't really understand most of them. Is there some kind of common theme I am missing? Can anybody explain why my moves were bad?
Definitely some subtleties in the attack, but the most glaring mistakes I noticed involved your rooks.
1) 20. Rhg1 taking your rook off of a potentially good square to defend the g-pawn, when you should just play h4! right away (him taking g4 helps accelerate your attack).
2) 26. Rf1 is too passive, when a move like Nd5 accomplishes the same thing but brings another piece into play, or even a move like Rdg1 instead, preparing to play h5.
3) 30. Rgg1?! leaves your rooks uncoordinated, when a move like Rg2! gives you the ability to double rooks on the g-file with Rfg1, and you have a powerful attack. Maybe you wanted to defend f1 so you were threatening to take e5, but that type of thinking is short-sighted.
Overall, you were too worried about minor things when instead you should want to bring your pieces onto good squares and open up lines.
32.Nf6 looks pretty powerful to me
In reply to #2:
1) I didn't play 20.h4 be cause I missed this line: 20.h4 Qxg4 21.h5 Qxf4 22.hxg6 Qxe3 23.Rxh7#. If I saw this line, I would have played 20.h4.
2) I didn't play 26.Nd5 because I was worried about 26...Qa4, when I thought black gets some counterplay. But looking at the computer now, seems like black has nothing. White doesn't even have to protect the c4 pawn, although 27.Rc1 also leads to a large advantage. (I ended up having to play Nd5 anyway, though.)
3) I don't understand how 30.Rgg1 leaves my rooks uncoordinated. Rooks next to each other seem natural enough to me. Although I admit that I played that move rather quickly, since it seemed rather obvious to me that gaining a tempo chasing the strong knight on e5 was good, and opening up the long diagonal seemed strong. I didn't seriously consider other options. I'm not sure why it is short-sighted though. Also, maybe I was a bit worried about black playing c6 and capturing f4 at some point.
Interestingly, according to Lichess's "chess insights" feature, the average centipawn loss when I move a rook is higher than when I move any other piece. I am not sure why. Maybe I might sometimes have trouble using them effectively. Maybe I might assign a defensive role to rooks too often?
Also, you said that I was too worried about minor things. During this game, I thought a lot about preventing counterplay (I think that's called prophylaxis or something). Isn't that a good thing?
In reply to #3:
Yes, I considered 32.Nf6 during the game. However, I dismissed it because I wasn't sure how to proceed after that. The only way I saw to proceed after that was to get Nh5 in somehow. However, the h5 square is currently protected, and I was worried about Qxe4+ in some lines.
Prophylaxis is not always a good thing, and this is especially true when attacking. Sometimes, you just have to get there before the opponent either has a chance to generate counterplay or bolster his defenses. Notice most of what you missed was due to calculatiion. Intuition can help tell you what the right move is (for instance, my instinct was to play h4 when I first saw the game, and I noticed the line you gave with Rxh7# after a few seconds), but calculation allows you to trust your instinct.
As for 3), at that point in the game, it's clear the g-file is the only file that matters, so you deprive yourself from taking full advantage by boxing your f-rook out from the file. Especially when attacking, rooks prefer files, not ranks.
Agree with all of the above.
"Is there some kind of common theme I am missing?"
I see 2 common themes:
1) passive rook play: 20 Rhg1?, 26 Rf1, 30 Rgg1, 34 Rf5?, 40 Rg1?, 47 Re1?, 47 Rf3?
2) indecisiveness with your knight: 6 Nc3, 14 Nd5?, 16 Nc3, 27 Nd5, 34 Nf4?, 37 Nh5?, 41 Ng3, 45 Nh5? It is obvious that the knight belongs on d5, but 14 Nd5 was premature. The rerouting 34 Nf4, 37 Nh5 is erroneous: the central d5 is a much better square than h5, knight on the rim = dim. From both squares it controls square f6, but from d5 it controls more and more important squares for both attack and defence than from h5.
In short: rooks on open files, knight in the centre!
@theme 33.b4 seems to take care of that problem (the rook eyeballing h5),then you can play Nh5.
but yeah...I definitely agree with the others the common theme here is that you mismanage your rooks,and over-calculating...just trust your instincts more.
Reply to #6
@amazingoid :
Thanks for you advise on prophylaxis. As for the other issue, I guess the real issue was that it never occurred to me that the g file was important. Maybe it's just a blind spot I had. I was also overly worried about some lines where black might move the queen somewhere (so capturing on b6 doesn't come with tempo), play c6, forcing my knight to move, and then capture on f4. In hindsight, that plan is much too slow on black. I guess I focused way too much on preventing counterplay.
----------
Reply to #7
@tpr:
1) Note to self: Don't maneuver rooks like they are knights.
2) I'm not completely sure how the knight on d5 helps with the kingside attack other than protecting the f6 pawn. My idea with 35.Nf4 and 37.Nh5 was that I was planning to put either put a knight on g7 before switching to the h file, or put a rook on g7, which I thought is crushing. I'm not entirely sure why my plan was bad.
As for 14.Nd5, I initially thought about 14.Qd2 or 14.Qe3, preparing f4, but I didn't like the idea of black playing Ng6-e5-c4, when I had to waste another tempo moving the queen again. Therefore, I decided to block the e6 bishop. I don't quite understand why 14.Nd5 was premature.
----------
Reply to #8
@CerebralAssassin19:
According to the computer, after 32.Nf6 Qf7 33.b4, black has 33...Rxa3 34.Qxa3 Qxf6, after which white's advantage is reduced. I guess the actual problem, though, was that I was overly worried about e4 hanging.
14 Nd5 is premature. You realised this yourself as you played 16 Nc3 i.e. you just donated 2 moves to your opponent. You could have played 2 useful moves and only then move Nd5. 14 Qe3 was good. On 14...Ne5 15 f4 Nc4 16 Qf3 yes you have moved your queen twice, but on f3 behind pawn f4 it stands better than on e2, as it has cleared the path for Bf1. Moving 14 Nd5 and 16 Nc3 has accomplished nothing, only loss of 2 tempi. I guess you looked at 14 Qe3, decided you did not like it and then played 14 Nd5 without giving it as much thought. If you had looked at 14 Nd5 more closely, then you probably would have realised that you had to pull it back 2 moves later with 16 Nc3.
So 35 Nf4 and 36 Nh5 was intended to prepare switching to the h-file. You could have switched to the h-file right away: 35 Rh5. Your plan Nf4-Nh5 was bad, as you came to realise yourself when you played 41 Ng3 and then back 45 Nh5. The knight stands poorly on h5, while on d5 it dominates the whole board. On d5 it is no weaker than a rook. On h5 it blocks the h-file you want to switch to. Square h5 was good for the rook, but bad for the knight. Do not move knights as if it were rooks.