When I was younger and just getting into RPGs, I read the hell out of the original Monster Manual, and not just for the naughty pictures. For the unfamiliar, monster entries used to be pretty short - an illustration, stat block and a very short descriptive block, often shorter than the stat block, and usually composed of a few sentences of description, and notes on special attacks and tactics.
That may sound a little dry, but I found it inspiring. Those few sentences of description were often quite colorful and suggested much more than they said. Each monster provided an invitation to flesh out the details surrounding it in your own imagination. And people did: the now-famous "ecology of..." line of articles was based almost entirely around the idea of taking that seed idea and expanding it extensively. Today's monster books - illustrated by the fantastic work in the recent Monstrous Compendiums - come down somewhere between those two points, filling in enough detail to flesh thing sout without quite going to the extremes of writing 6 pages on the ecology of the darkmantle.
This range seems like a snapshot of one of the big questions of gaming, that of legos versus action figures. That is, should the game provide you the bits with which you can play the game in your head (legos, old monster manual) or should it provide you everything you need for the complete game (action figures, fleshed out monsters). Now, before you instinctively leap to legos as the better choice (as nerds are wont to do) I want to poin totu that while the Forgotten Realms may be an example of an action-figure style of play, so are many story games. It may seem odd to lump Fiasco in with Dungeon Crawling, but in this case they're in the same bucket. Both provide the structure (in VASTLY different ways) rather than the parts. This, I hope, is a good illustration of why both approaches are fruitful and full of goodness. Anyone who as ever played with action figures knows there's no shortage of imagination applied to the play, it's just within the bounds (dare I say, creative constraints) of the form.
Hopefully that makes it clear that I don't think the question is one of one versus the other. Both approaches have a place in gaming, and that's good, because there are plenty of places where the distinction is hard to make. Take GURPS - the game is designed to be a lego, but any given setting book tends to be deep and rich enough to be all about the action figures. Sure, someone might deconstruct one for parts, but more likely, they'll just mix them up (combining the G.I. Joes with the My Little Ponies, as it were). Rather, the question on my mind is what makes for a more broadly useful _product_.
Evidence points towards action figures. Fiction and structure make money, and at first glance, legos aren't much fun for the uninitiated, but there's a bit of a hidden trick to it. If the potential customer has something in their own imagination which they wish to capture, legos are the tool for it, and this is something that the sparse, almost accidental lego-ness of early games managed to capture. If you had an idea in mind, such as swinging a sword on adventures, then this gave you the tools to manifest that idea.
The curiosity, of course, is that those ideas came from other action figures - that is to say, fiction - as people built ideas based on wanting to play something inspired by their impressions of Lord of the Rings, Interview With a Vampire, Star Wars and anything else. On some level, that created a weird cycle of fiction --> Ideas --> Game built with legos --> Spinoff Fiction (novels, gamebooks) --> new Ideas. I admit that in this context, I become a little more patient with the infinitely-milked settings (Forgotten Realms, Star Wars Extended universe) but at the same time I cement more clearly why it's not for me. I like my action figures, but when they become full on figurines, I'm out.