This is from Vornheim:
Once, demons ruled every universe, unchecked. Then came 12 sisters – medusae – they looked upon the demon kings and changed them to stone, and drove the rest away. The grey bones of this earth were hewn from the colossal statues of these demon kings. Or at least that’s what the 12 sisters will tell you.
Two of these sisters were in the city kit. Two others have appeared in my campaign already. One day I sat down and tried to decide where exactly the other 8 were. I figured they were all old and powerful and that it would probably help enrich and define the setting to figure out exactly what they were up to.
I knew one was an eel medusa, and at least one was a powerful sorceress, and...and after about 40 minutes I decided it was dumb idea and I didn't want to know. Or I didn't want to know yet.
It wasn't time to know.
____________
Imagine a very simple gameworld.
It has four great powers--mighty sovereigns. The GM (and maybe the players) know who they are and know they are the most powerful and setting-defining entities in the game. They are:
King Arthur, whose sphere of influence grows each time monsters are beaten.
Dread Cthulhu, whose forces becomes more powerful whenever major magics are unleashed.
Queen Isabelle, whose mercantile empire benefits whenever new lands are explored, and...
The Lich King, whose armies swell whenever war occurs and soldiers die in droves.
The idea is: whatever the PCs do in this world can cause one or the other power to gain in strength, and can, therefore, affect events at the "domain level".
In turn, once these domain level events occur, they can affect what kinds of emergencies, quests, and/or opportunities the PCs see before them in the course of the game and, therefore what happens to them.
This is not an unfamiliar idea, basically here I just want to lay it out as one end of a spectrum, Most gameworlds are or can be more complicated than this, but the basic principle is:
knowing the important forces in you gameworld is attractive because it allows PC actions to generate new plot hooks for themselves merely by doing what they'd normally do
...every time the PCs do something dramatic, it can affect the fortunes of some great power, who then must realign and send new things at the PCs and the whole thing is like a self-generating adventure-creating machine.
(I thought my Dozen Gorgon Sisters set-up would do the same thing. My PCs have already fucked with four of them--killing two, this could potentially set up events for sessions and session to come.)
Another nice thing about making the game work this way is the story follows the law of dramatic efficiency--what's glimpsed in the first act becomes relevant in the final act, all roads eventually return to the center, Darth Vader turns out to be Luke's father and then it is Darth who kills the emperor, etc.
You can see some examples of people coming at these ideas from different angles here, here, and, to some degree, here.
The same ideas can be ported to different levels of the game: the great powers in question can be gods, or factions in a city or whatever. God A loves Hercules, God B does not, thus: the 12 labors, etc. etc.
HOWEVER...
appealing as this idea is--you can indulge your world-building impulses while simultaneously doing "real work" by putting together a machine that will generate stories for you, there's an equally interesting notion about how to do things whose essential mechanism pulls in the opposite direction. That is: the aesthetic of exploration.
Exploration in two senses:
Sense One: the PC's job is to discover unknown places and things
Sense Two: the world is revealed to the players (and often to the GM as well) the way it is in a book. That is, the explanation of each new aspect of the world is made a part of the story.
The second sense is slightly more complicated, but examples are everywhere. Here's MJ Harrison...
"With the mist dispersed the village smelt of smoked fish and salt. Fulthor and his party stood outnumbered and uncertain at the centre of an unarmed crowd. Hornwrack had put up his knife. Like the survivors of some forgotten colonial war (desultory, expedient, never quite resolved) the occupants of Iron Chine drew round him: thin intelligent women, a few bare-limbed children. There were no young men present, only some old ones who stamped their feet and turned up their heavy collars, faded blue eyes watering in the cold wind. They stared up at him with a defiant incuriosity and he stared back embarrassed, although he could not have said precisely why. It was a mixed community; at the periphery of the crowd a handful of the Reborn hovered like strange, long-necked animals, their delicate features coarsened a little by an unrelenting deprivation. What had they left behind them in the Afternoon, what mad sophistications exchanged for the smell of dead fish?"
...in other words, the "PC"s just arriving at a place is a moment for storytelling. Even if the PCs theoretically already know about Iron Chine, the players don't, and so having it unfold in front of them is interesting.
The slow revelation, to readers (or, in this case, players), things that the characters supposedly take for granted is one of the great tools of storytelling.
It is also an important element in world-creation-through-play--the technique of building up details of a world the way pulp fantasy writers did--one short story at a time.
I didn't leave parts of Vornheim unfinished just so the GMs reading can have the joy of filling them in themselves--I also did it because having undefined-at-the-beginning is a fundamental part of how I see the city parts of a campaign unfolding. It's a cliche and a simplification to say "If I knew the whole setting I wouldn't want to run it"--what I mean is more precise: "If I had an exhaustive list of everything important about Vornheim, I'd feel my options for what I could do in any given session were limited--in an amorphous but palpable way".
________________
What I want to point out here is, the essential mechanisms that makes these two approaches--intrigues and exploration--interesting are opposed.
The appeal of the "Here are the powerful forces and everything is a result of them interacting" method is the satisfying feeling it creates that all the variety and complexity of the campaign is generated from a few comparatively simple moving parts that are known from the beginning--or at least near the beginning. If wholly new elements are constantly being introduced, then the feeling that each development is a natural consequence of previous actions is spoiled or obscured. It would be like a chess game where suddenly new pieces were randomly dropped on the board: interesting, yes, but it would change the nature of the appeal of chess altogether. The feeling of a tightly interacting mesh of intrigue and consequence is lost. And the feeling that, if the PC found a monkey wrench, s/he'd know where to throw it.
The appeal of the "revealed through play" method is that it allows the players--and usually the GM, too--to experience sudden revisions of their view of the whole gameworld at any time. However, when the PCs have this feeling of constant untapped potential, they don't necessarily feel very up to turning the setting to their advantage.
___________
So a game of intrigues and consequences and a game of exploration have fundamental differences, but it would be a trite oversimplification to say they're wholly incompatible.
It would also be trite and kinda not terribly helpful to say "If the world is too defined, then there's no room for surprises". The real question is to fix the boundaries at which one impulse genuinely interferes with the other.
Here's an example people know: Star Trek.
So in the beginning we have two powerful forces defined: The Federation (mostly earthlings) and (part of it? Perhaps it was unclear in the beginning) The Vulcans. They are on a mission of exploration.
So this'll be a show about exploration, not intrigue, right?
They go around to planets, meeting new things they didn't know were there. Important unconscious fact: most of the phenomena they discover are only relevant on a relatively small scale--one planet, one city, one family, one solar system.
However, eventually we are introduced to two new phrases: "Klingon Empire" and "Romulan Empire".
The important part is "Empire". According to Star Trek Expert Mandy, the Klingons at least were introduced as if they had always been there, and the Romulans too.
Suddenly, we have a paradigm shift in the way you look at the whole "gameworld". Space isn't just a disorganized jungle place that the forces of civilization (the Federation) hexcrawl through looking for unusual localized phenomena, it is also a (somewhat known--there's a "neutral z0ne") battleground between large political entities.
As time went on, more and paradigm-shifting revelations were made: Romulans used to be the same as Vulcans, Vulcans came to greet humans as soon as they discovered the humans had warp-capability, the Federation was formed after near-apocalyptic internecine wars on Earth, the Ferengi (another race that had apparently always been there) show up and they're despised by the Klingons, and on and on.
Now it's fairly safe to say that as of now we have a fairly defined set-up. It's hard to imagine a whole new universe-spanning empire or religion that had always been in the Alpha Quadrant and just never mentioned showing up by the early '90s. So what did they do? Advance into whatever quadrant the Founders were in and whatever one the Borg were in and into "fluidic space" where lurks Species 8472. Now we have three more political players introduced through exploration. Now every Star Trek political decision at the "domain" level theoretically takes into account all of these players.
And, during all this intrigue, the exploration--oh look the people on this little planet eat rocks and have antigravity shoes and hate all androids--keeps going. But it does so under a "ceiling" of how earth-shattering these revelations can be, in the scheme of things. In any Star Trek episode one has a definite feeling that whatever new thing's going on can only be so important unless it involves one of the Big Powers. Things are either new (and therefore novel but important) or important (and therefore involve the intrigues of familiar friends and familiar enemies).
This same dynamic is pretty clear in most sci-fi settings. In fact, it may be the only possible interpretation of the words "sci-fi setting". In the Warhammer 40k RPGs: we have intriguing powers--humanity, chaos, eldar, space orcs, and whatever other lead armies the WH40k people have put out and, in between and relatively quietly, we have all the novel weird, out-of-left-field stuff which you can surprise people with in the "exploration" game.
The Trek people eventually stylized Deep Space Nine and Voyager right along this split: DS9 spun endless plots by setting up a series of (mostly) pre-existing powers--The Cardassians, The Boringjorans, the Founders, The Federation, The Klingons, and the Ferengi--as moving parts, dropping in some crisis, and watching the balls bounce off each other for 40 minustes. Voyager stripped away all the familiar Star Trek intriguers and dropped the ship in the middle of a 3d hexmap full of the unknown.
_____________________
I guess this goes back to something I've said before. If the players know too little about the setting, then they can't effectively move the levers of power or even just enjoy the feeling of the watching the machine work and the feeling that that causes have effects and the gameworld has dramatic closure once in a while. On the other hand, if the players--and the GM--know too much, then they get to a point where you can't just casually reveal paradigm-shifting-eye-opening-hey-look-the-Earth's-actually-round-and there's-a-whole-continent-over-here type things because the known world is already fixed in the players' minds. And that's fun.
I hear people go "Oh I want to do a campaign where the PCs are pirates and the enemy is an island nation of mad wizards who rule the seas with snailships" and I think that sounds hella fun. Then I think: but what do you do if one day you want to do an adventure where they go to the center of a hollow earth and fight the avatar of the crystal spider god who spins the web of all men's fates? Just wait and do it next campaign? Far too impatient for that.
What I would like, I think, is to extend that "first season" feeling as long as possible. The point where you think you know how the world works, but you're not so experienced that the world has to work the way you think it does or else fail to make any sense. It's tricky--there's the risk, on the one hand, of PCs having no fucking clue what's going on or what to do about it, and, on the other hand, feeling like Oh this guy again.
And there are as many ways to do it as there are GMs, I suppose. Tricky every time though, unless your players are all the same, and all wholly dedicated to one thing or the other. For my part, mine are as moody as their GM.