Here's the game
lichess.org/75I92G5G/black#45Instead of 23...Rf6 (which I played in the game), the computer suggested 23...Qh4+ 24.Kg1 Rfc8, which looks very bizarre, pointless, and slow. However, the computer evaluates the resulting position as equal (evaluation of 0 at depth 26/26).
Can anybody explain the idea behind Rfc8 and the reason the position is equal? It seems to me that white's attack is about to crush black already, while it's hard for black to get more pieces into the attack, especially since Rfc8 moves a rook away from the kingside.
You need to be able to generate counterplay, so ...Rf6, blocking your Queen's access to h4, can't be right. Qh4+ gives the rook a safe square to go to so that you can get your counterplay going. Various ideas include Qh1+ and Qh4, with a repetition, or even things like Bh3, Bxg2 and Nf5, sacrificing two minor pieces for a perpetual check with Qg3-h3.
Also notice that even Qh4+ Kg1 Qh1+ Kf2 Qh4 is playable, but white can freely take the f8 rook since you don't have any fantastic discoveries. Your best bet in that situation would be to recapture on f8 and then play Nf5+--->d4 and have compensation that way.
My idea with 23...Rf6 was to bring the rook over to h6 or g6, and leaving h4 open for the knight, solving the problem of how to bring other pieces into the attack which I found in the 23...Qh4+ line. The computer says that if white didn't play 24.Ne6 or 24.Ne8 (which I somehow managed to miss), then black would have been at least OK, or even slightly better.
My issue with 24...Qh1+ 25.Kf2 Qh4 was that I didn't see how a discovered check would help. After reading your comment and looking at the computer lines, I realized that the knight at d4 prevents both rooks, the bishop, and the queen from guarding g2 after moves like Qg3 and Nh4. Given how powerful that knight is, it makes sense to me now why black has compensation. (Interestingly, the computer main line leads to a bishop+knight vs rook+pawn endgame: 24...Qh1+ 25.Kf2 Qh4 26.Bxf8 Rxf8 27.Qb6 Nf5+ 28.Kg1 Nd4 29.Ne6 Bxe6 30.dxe6 Qg3 31.e7 Nh4 32.exf8=Q+ Bxf8 33.Nxf4 Nxe2+ 34.Kf1 Nxf4 35.Qf2 Qxg2+ 36.Qxg2 Nhxg2 37.Rc8 Nxe1 38.Kxe1)
I still don't understand Rfc8 though. Does it accomplish anything other than save the rook? The position doesn't seem like one that black can waste time on. Sacrificing the exchange makes more sense to me than wasting a tempo, and I don't understand why Rfc8 is so much better than Rfd8.
Yeah, it just turns out tactically you didn't have the time to activate your rook because you had too many pieces in the way.
Rfc8 is preferable because it gives you the resource Ra8! For example, after Qh4+ Kg1 Rfc8, if white plays Nf2, you have Ra8! If Nxa8, Nxe2! nets you the rook back with interest. These positions are ultra-sharp, so tactical considerations will take precedence over anything strategic.
I saw Ra8 in the computer variations, but I didn't look too deeply into it. I didn't realize that it's connected to Rfc8, which pins the c7 knight. I kind of like this pinning idea. Also, I looked deeper in to the line, and it seems like the point of Ra8 is still to activate the rook, except from the other way around. It's kind of crazy this idea works at all.
Anyway, thank you.
18. ... g3?! probably this is bad.
thanks amazingoid for clearing up the game
24... Rfc8. 25 Ne6 Bf6! :) nice attack. I wish i could see those :)