Hello,
Usually we recommend to castle early to avoid trouble later on in the game. This is what I did in the following game at move 15:
However the computer sees that as an inacurracy (not a big mistake) and proposes a very obscure line to me. The first move is okay, but all the part with the rook sacrifices and creating weaknesses in the pawn structure, I do not get the point. What are your thought about it? Would any human player at high level consider this line to be the most natural following or is it only some computer line that does not deserve analysis?
It's completely okay for a human to castle :)
15...b6 attacking the center is more active than 15...O-O which allows white to consolidate with 16 Bg2 threatening 17 f4, 18 f5. It is generally recommended to prepare castling as soon as possible, but to delay castling itself if there is a more active move available.
Ra4 looks strange at first but can be easily explained : it is a prophylactic move against White's only active plan : that is f4-f5 trying to play against the Bg6 and open some lines on the kingside since Black is better on the queenside already. Then f4 is met by Rxd4 when Qxd4 Bc5 loses the queen. But a human would probably play 0-0 instead of Ra4 and wait to see another move, because Black has several plans available on the queenside : he can play b5 then gang up on the a3 pawn, or play c5 to strike in the center. So it is useful to have more information and make a necessary move meanwhile before choosing a plan.
b6 is a completely natural move, you don't want White fixing your pawn on b7 by tripling on the b file. Before going 0-0, it is necessary to check all the pawn breaks and see whether they work : that is b6, f6 and h5. Here only b6 really works and it may be better to play it right now to fight for initiative, although the difference with 0-0 is probably not great anyway. White can try to use the extra tempo to speed up the play on the kingside, but his position has too many weaknesses for this to be really successful.
Thanks everyone! It is much more clear now! In fact I considered b6 because it seemed also natural to me but I did not think it would achieve anything in the near future. It was just a short game but I am not sure I would have calculated this rook move only to anticipate f4 even in classical or correspondence, with one pawn seemingly blocking the attack on d4. I think this a a hard move to see!! But it could make the game more interesting than castle. :)
You can look at it another way. O-O is a strong move as it brings your king into safety and as it connects your rooks. In the present position your king is not in any imminent danger and there are no open files for your rooks. So it is safe to postpone castling for the time being and with ...b6 attack the center and open the a-file for your Ra8.
Wow