Imagine if it would be possible to, optionally, make a three square jump with the king, when castling. Arguably, this would enlarge the opening tree, as many more opening variations would become practicable. The long king jump could prove quite useful in most variants with queenside castle. At kingside castle, it is sometimes worthwhile to place the king on a more protected square. This ought to be good, for instance, in King's gambit variations. More information, and Zillions program here:

http://hem.passagen.se/melki9/castlechess.htm

Play chess with extended castle, online, here:

http://www.chessvariants.org/index/msdisplay.php?itemid=MPcastlechess

/Mats

This is maybe the first genuinely sensible suggestion for a variation that I have seen in a long time.  It would certainly be a game-changer, as when deciding whether to castle one's king into an open file, much rests on whether or not the file can be adequately defended to pinning or mating attacks without losing tempi, at least it does at my level...  For that reason, it would probably only benefit the defender and improve middlegame king safety, which one might say would lead to more draws over time.  However, if one's opponent's attack is based on a double-edged book opening that offers chances along that opening file with opposite side castling, then it would swing the game very much in the quuenside-castled defender's favour.  The open c-file motif is with hanging a-pawn is a common theme of many positions, and I see that castling queenside with the king going to b2/b8 would nullify, or at least reduce the tension of a lot of these.

Does anyone a lot better than me have any other thoughts? 

I am not certain that it would generally benefit the defender. Take Caro-Kann, main variation, as an example. Black certainly benefits from the long king jump, but so does White. So nothing has changed. In many variations in the King's gambit, White would benefit from a king jump to the corner, as the diagonal to g1 is open. So the extended castle could get some life into the King's gambit again. In certain open Sicilian variations, White attacks with the g-pawn. To have the king placed on h1 is probably better.

It would become more popular to castle queenside, since you gain a whole move. But queenside castle is generally associated with attacking variations, for example, as White in the Sicilian Dragon. A worriment is that the tempo gain would kill the Sicilian Dragon. But perhaps Black can benefit from the king jump to the corner(?).

The Dutch defence is a problem child. But the extended castle could perhaps make it viable again.

/Mats