I'm apparently on the other side of an invisible divide that runs through the OSR and, perhaps, the gaming community in general. Most of the old school renaissance strays to the Picaresque when it comes to worldbuilding, a fact highlighted by the enormously enlightening Tom Fitzgerald at Middenmurk. If there is one blogger I aspire to be, it is Mr. Fitzgerald, who routinely posts extremely well-researched and enlightening topics and allows filler to clog his writing as, unfortunately, I sometimes do here.
The so-called "full build" stands in stark contrast to the Vancian partial build. I love the work of many pulp authors who create a universe only large enough to support the story and the characters in it. The successful ones manage to hint at a world beyond the borders without actually fleshing it out. But that's not the kind of man (or DM) I am. Having taken Tolkien into my lifesblood so early, my work in both roleplaying games and in writing has been to overprepare. The world itself is not easily reducible to its appearances, so (my logic dictates) I am required by the law of verisimilitude to know as much as I need to in order to sufficiently prepare all eventualities. This means I do not fly blind into design, but rather prepare things in layered stages so that the final work is really only a surface encrustation of all the work that's come before.
This strikes at the root of arbitrariness. It requires a superhuman amount of preparation, as well, which is not always for the good. For example, my games would never need to halt for lack of notes on a region if I were content with creating the regional details on the fly. Returning to the Tolkiensien task of building from some deep subterranean level upwards, this has not been my style since I was very young. I must know not only the terrain I'm dealing with, but its deep history as well.
What does this mean for my games and my settings? How does the full build stand apart from the Picaresque approach to worldbuilding? I must first set aside the thoughts that my games are "high" fantasy in any sense, or that epic stories concerning the fate of the world are the norm. No, indeed. The gameplay is much more picaresque than the building. Tales grow organically as various eventualities emerge. The story of the players is much less the narrative construct of a grand novel and much more the collation of various shorts and slices-of-life.
But the building... what benefit can there be for a single group of players (more or less) to a statically developed setting? Here, then, are the benefits as I see them. In contrast to the vast audience posited by Tom at Middenmurk, my own full build setting has rarely been used by DMs other than myself. True, there are times when others have taken the setting in hand and run games... but those times are rare, the exception and not the rule.
Secrecy. The first power of the full build is that of secrets. This, of course, requires the DM and designer to maintain a firm grip on his own tendency (or her own, as the case may be) to run his mouth and reveal things. Secrets are encrusted into the very heart of the setting when it is built up in layers. Their effects can be seen, but not necessarily predicted, by characters (and players) observing the surface layer. These secrets move things, motivate them, and push them for years of game time when used properly. In many cases, things you don't even think or intend to be secret serve as motivating secrets. Many are simply pieces of history that are lost to time or where not otherwise recorded.
The presence of buried carbuncles of secrecy like this can provide a similar (though not the same) drive toward endless newness experienced in the picaresque view.
Consistency. Consistency is the bigger of the two benefits. Under consistency, the same group of players will find that their character's actions have long-term and game-to-game effects on the setting. Retiring a slew of characters introduces a slew of new NPCs into the political life of the setting. Destroying kingdoms, founding empires, and slaying important folks all has a direct effect on the games of the future. This, I find, is one of the greatest rewards for the full build. Consistency of setting means that you and your players build together a future-history that they have a vested interest in, and delight in returning to.
Dealing with the Problem of "growing used to" the setting. The biggest problem with this, of course, is the end of the reveal. The cysts of secrecy that inhabit the deepest layers of the setting cannot provide the endless newness of a fresh Vance-style reveal. New cities don't generally exist just beyond the horizon. Players know that using acid on trolls kills them. What do to with this?
The way I deal with it is thusly: metaknowledge of that nature incorporates itself into the general knowledge-base available to all communities in the game. Most folks in the 10th Age know, for example, that acid kills trolls. Why? Because every player knows that. It would be asking an unrealistic feat of mental gymnastics for them to pretend they don't. This makes sense in a setting like this, where adventuring is a recognized phenomenon.
In conclusion: I fall outside the OSR's normative model. I knew that already as a proponent of 2e. I have my own model, and it has served me well. Perhaps it will serve others, though there is a requirement of intense dedication to the craft of writing notes that will literally never be seen. Maybe that's a waste. Maybe not.