I don't think I'm saying anything that hasn't already been mentioned in this thread. Just my $0.02.
Castles don't need to be as good as towns. They just need to be a little different, and provide something that towns don't provide...or provide it differently. I think castles could jump in value a lot by providing "normal" stuff also available elsewhere, but somewhat owner- or faction-restricted. I can see them playing a niche role, being useful during vassal or ruler stages of the game, while remaining pretty much useless as they are now during early-game.
I'd like to see:
- Noble troop recruitment...only open to the castle's owner or perhaps only members of his/her faction. So, for example, if I'm a Battanian vassal, I can recruit Fian-line nobles at, say, Ab Comer castle, but I cannot ride down the valley and also get Banner Knights from Talivel Castle. Maybe the additional benefit for ownership is that the troops are relations-locked in higher slots just as everywhere else, and only the owner automatically has access to all slots, whereas other Battanian nobles are limited by whatever their relations are with me. And if I'm a mercenary or otherwise an unaffiliated adventurer, then no castle-recruited noble troops at all, I have only the villages like now.
- A smithy. It's pretty obvious that any self-respecting castle would have weapon- and armorsmithing capacity. But it's also pretty likely that they're not just going to let any old flea-bitten adventurer waltz in and take over their forge. It would be reserved for the owner's use and possibly others within the faction. I'd like for this castle smithy to have something that differs from the current town facilities, although I don't have any firm idea on what that would be. Maybe stamina depletes more slowly in one's own castle smithy, reflecting that the work area is arranged and stocked just how the castle lord likes, and is more efficient? Or maybe there's always a few extra smithing materials there, like, it spawns a few Fine or Thamaskene Steel occasionally?
- Strategic significance. I don't think every castle needs to have some paramount strategic value, but a few should. Garontor and Tubilis castles on opposite sides of that western-desert land bridge are an obvious example. Requiring possession of both castles for an army or war party to pass might be a bit much...but preventing passage of enemy caravans seems realistic. So, for example, I can go north with an Aserai army and take Ortysia, but if I don't also take Garontor, then no Aserai caravans are going to be able to transit back and forth...gimping Ortysia's economy and not contributing anything to the larger Aserai economy either. (yes, I'm aware I'm just restating a very similar idea already posted further up the thread).
Game mechanic-wise, could be implemented as some sort of "zone of control"; at a certain distance, caravans move away from a hostile castle just like they currently do from a hostile field force. For many castles that are just out-of-the way or in the middle of easily traversible territory, it won't mean much; caravans just aren't in the area anyway or will find a way around. But castles at chokepoints would have very significant value. Perhaps the ownership benefit of such a chokepoint-castle could be a peacetime passage toll for all caravans not of the owner's own faction?