Castling



I was taught that to castle early in a game was good practice
I like to prepare to castle as rapidly as possible, but to delay the actual move until it's clear whether O-O, O-O-O, or not castling is best. Here's an example from yesterday where I delayed until move 17... a seemingly perfect moment.

Thank you likesforests for showing that. I tend to try and do the same, get in a position where I can castle if needed, but only do it when it suits me, not when I have other moves I want to make first. When I was first being taught chess by my dad he taught me to castle early and I think this is a good way to teach kids very early on, otherwise I used to just leave the king completely open in the middle of the board!

There's a great article on chess.com called "Breaking the King's Indian Attack" which illustrats how tactical castling can be. Look at what black does.
http://www.chess.com/forum/view/game-showcase/breaking-the-kings-indian-attack

It depends on the opening, but the basic idea, as Reb said, is to look at the center. If it's locked, reserve castling. You may never need it. If it's open, castle now! IM Jeremy Silman says "If you're safely castled and your opponent isn't, you should be trying like mad to rip open the game." Obviously, this means that not only should you castle out of open games, but the opponent will try to mold the open/closed according to you (and you should to)!
As for castling queenside or kingside, look at the pawns. Did you play Bxf6 gxf6? The opponen'ts not going to castle onto the open file, and that H rook is going to go to g8, guaranteed. Don't castle right underneath his guns! Go queenside! Vice versa is true too.
I was taught that to castle early in a game was good practice, King safe behind pawns, Rook in centre of board, etc. Recently I've lost games after castling by attacks down the a & h files by Rook and Queen also Bishop supporting Queen on long diagonal.
Has theory and practice changed/evolved ? Whats your opinion? Comments welcome.