Creating a world together is definitely a great way to do this, and doing the sort of cinematic, storytelling reveal is also great.
I think, though, that you run into problems with games set in worlds with established lore. Like, if you’re playing Legend of the Five Rings, you’re not going to be asking the players what the samurai code of honor requires, or what their clan is best known for. You’re playing L5R, dammit, and that means playing in a richly detailed world with all sorts of presumed lore.
And while just-in-time information and gentle corrections are good (probably the best, really), they still can’t handle assumed context very well. Like, imagine that you were 18th-century gentry playing in a “fictional” sci-fi world that was very similar to 21st century America. Your character sheets mention being able to buy all sorts of highly lethal, easily concealed guns. The story involves assassinating a local magistrate. So of course the PCs start to put a plan together that involves sneaking guns into the courthouse.
The GM knows that the courthouse has metal-detectors and armed guards. But the players are 18th-century gentry, not 21st century citizens, and so they don’t know (or have the context to anticipate) the metal detectors and armed security. So we could end up with the players making all sorts of plans based on known setting detail (guns!) only to have the GM be like “oh, hey… there are metal detectors and guards here, people queuing up to go through them, what do you do?”
Now, if the players had shared their plans with the GM, the GM hopefully would have mentioned the metal detectors and guards. But maybe not! If the GM was just running a scenario and the PCs zigged when the GM expected a zag, the GM might be flipping pages quickly and be just as surprised as the players by this bit of assumed lore.
I guess my point is: if you take it as a given that your game has rich, established lore, there need to be solutions for communicating that rich, established lore that doesn’t suck.
Is I write this, I find myself thinking of the Savvyhead workshop rules, and these two articles:
The idea being, have a procedure where the players are like “I want to do __” and the GM provides the requirements. The GM uses those requirements to establish setting lore in bite-sized, contextual pieces.
If my Victorian role-players said “I say, the best way to deal with that knave of a magistrate would be to sneak some of these remarkable repeating flintlocks into the very municipal building wherein he holds court, and bun him down publicly like the rabid dog that he is! GM, old chap, how might we accomplish this?” I might twirl my mustachio, suck on my pipe, and say “Primus, you must first acquire a suitable array of these repeating flintlocks. Secundus, And you must learn the magistrates schedule and the layout of the courthouse, so that you place yourselves in the correct time. And tertius, you’ll need a way to avoid the notice of the armed constabulary and slip past those marvelous arches of theirs that detect ferrous metals.”
“By Jove! How do legitimate citizens pass through these arches? Certainly they are not forced to empty their pockets of cigarette cases and remove their very belts and buckles in public. The indignity!”
“Quite so, good sir. It seems that this is a reality of the world in which they live! Perhaps it’s a natural cost to having such lethal armaments so freely available.”
“I say!”
(sorry, kind of got lost in that little reverie)
By having a procedure to say “I want to accomplish __,” and the GM says “that will require __, __, and __,” you build in a way to establish setting-specific lore just in time, before anyone goes too far down the road of making misinformed plans.