I think my italics went wide. 
Thanks for the explanation of your distinction between "build" and "pick". I think it's a useful way of thinking about characters. My term "develop" isn't so helpful.
What I'm opposed to is something shared by 3.X and 4e - that a player can sit down and draw out their character a priori, without a die being rolled in anger. (Indeed, it's often the case that they can do this without a die being rolled at all.) This seems like the complement of the ref sitting down and drawing out their story a priori, which I also oppose.
Let's see if that analogy will run back the other way:
In a pulp adventure game, I think the ref should lay out a simple sketch of the initial setting (typically a village, or quarter of a city), plus an initial adventure site (usually The Dungeon), with some brief notes on what lies beyond. Doing more than that is a bad idea because it may be a waste of time (the game may be stillborn, or the characters might go off in an unanticipated direction).
So, for the player, they need a simple sketch of the initial character (name, physical appearance, a defining character trait or two) and an initial adventure mechanic (fighting man, magic user, etc.). It's not helpful to plan out another 19 or 39 levels of experience, because the game might die, or go in an unexpected direction.
The big problem is: if you want to keep character optimization around (and it is an accepted part of the hobby now), there have to be development choices. Finding a good place for it in the rules, that still reflects the D&D heritage - that's a good design puzzle.

Thanks for the explanation of your distinction between "build" and "pick". I think it's a useful way of thinking about characters. My term "develop" isn't so helpful.
What I'm opposed to is something shared by 3.X and 4e - that a player can sit down and draw out their character a priori, without a die being rolled in anger. (Indeed, it's often the case that they can do this without a die being rolled at all.) This seems like the complement of the ref sitting down and drawing out their story a priori, which I also oppose.
Let's see if that analogy will run back the other way:
In a pulp adventure game, I think the ref should lay out a simple sketch of the initial setting (typically a village, or quarter of a city), plus an initial adventure site (usually The Dungeon), with some brief notes on what lies beyond. Doing more than that is a bad idea because it may be a waste of time (the game may be stillborn, or the characters might go off in an unanticipated direction).
So, for the player, they need a simple sketch of the initial character (name, physical appearance, a defining character trait or two) and an initial adventure mechanic (fighting man, magic user, etc.). It's not helpful to plan out another 19 or 39 levels of experience, because the game might die, or go in an unexpected direction.
The big problem is: if you want to keep character optimization around (and it is an accepted part of the hobby now), there have to be development choices. Finding a good place for it in the rules, that still reflects the D&D heritage - that's a good design puzzle.