Wow. This is a topic near & dear to my heart, but I'll try to avoid waxing nostalgic too much in this thread. A lot of what's been said already by peers I regularly see and respect, such as Spatzimaus and Nyaricus is very good. So I'll just add my wood to the fire as I sit down and regale you with where 21 years of world building has gotten me.
Yeah, serious. Started young, never quit. So, question one: is it worth it? Dude, it's worth exactly how much you decide it is. You've heard garbage in, garbage out? The same principal applies. If you invest your time, passion and effort into defining, refining, developing and building your world, it will become an extension of your psyche. It will take on a life of its own inside your mind, and you'll be able to spin, change, move and understand the ebb & flow of it without breaking out of it.
Rome wasn't built in a day. TRUE DAT. Getting to that point takes years. For me, it's much simpler because I have a series of ideas I continually refine and invest in a new place (such as my current HB campaign setting). I have stacks of rules and axioms I run my games by which keep everyone on the page I want them to be on in order to tell the story successfully.
Nyaricus brings up a good comparison with the GTA series; I have a Ravenloft campaign which runs under a similar 'sandbox' design; here's the world, here's some plotlines, pick what you like and run with it. Once you choose a direction, I can more effectively build a plot to go with it. This is an open-ended storyline; in this case I have an ultimate villain, and a group of heroes, but they're choosing the path they take to get from here to there. They choose which sites they see on the way, which battles they fight and what the world and its inhabitants mean to them.
You mention the "World-Saving Campaign" which ultimately is what a lot of D&D becomes because just hanging out in the local tavern doesn't get you real far; it goes against the grain of the game, which is about mighty heroes & horrendous villains (BoVD, how I do love thee). The demon plot you have is (in concept) similar to something I'm doing, except I have a much larger story built around the plot line; that's my fingerprint. The actual Story Arc.
- WHO is the villain? How did he come to this point? What need will he fulfill by executing his plan? Why is that plan important to him? When is it a good time for the PCs to encounter him? Personal favorite trick? Introduce the villain in the beginning of the campaign as an innocent NPC. Amusingly, I didn't do that this time and everyone suspects the Paladin at the head of their Order is the BBEG. Irony.
- WHY do the PCs care? When you mentioned you were content to let them sort of runamok and gain treasure, that's a measure sticking point for me. My games are the antithesis of the stereotypical D&D experience. Building the world is great; kingdoms, histories, etc. All great. The question all PCs ask (intentionally or not) is "Do I care?" You need to give them something and make it important to them. For example:
In the Ravenloft campaign, I let everyone do their own character gen, but with one caveat. In all of their character histories, they must include their Uncle Hammet, who raised them for varying periods of time in his ancestral mansion. This did two things instantly. First, it got everyone talking and defining -- on their own terms, with my help -- who Hammet was and what he meant to each of them, giving them common ground. Second, their characters were automatically invested in the concept of family. These are siblings, sometimes rivals, but all family. They aren't just "my adventuring party."
Then I killed Hammet. And that's how I started the game. I say this because during that process, they're asking questions. "Did we go on trips?" "Did we have servants, what are their names?" "Is there a major holiday? Birthday? What were those like?" All of those questions forced me to come up with answers, and build in more characters, more reasons, and solidify the construct of their reality, before going on and answering larger questions; Who killed Hammet? Why?
That's the reason I brought up this specific example; usually I start with a major story arc (a demonic invasion) and build it up into a fairly linear storyline. The PCs get choices as to where to go and what to do, but they keep their eyes on the prize. They're invested in the world and have bought into saving it as being a worthwhile and important thing. With the Ravenloft campaign, there's a much larger world with far less linearity. I started with a premise: Your uncle has been murdered.
No order to find the killer; they did that on their own. I gave them the mansion, X-Men style, to rebuild and improve as they saw fit. Or sell it. But since it contains their entire history as a family, they aren't getting rid of it. I made the basement a dungeon for a change of pace, and I put Uncle Hammet's Ghost in his study, along with how he was murdered and by what, if not whom. From there, the world was built expressly based on how the players approached it. They go to a tavern, I have one ready, with cast & crew standing by. They go through a cemetary, I have an encounter. When I wasn't ready for them to go somewhere, I told them so.
In this case, I've kept them in the loop. I'm doing what I call reactive writing, instead of proactive writing. Once they've settled on some basic plots I can expand the writing much more easily, because they're on a temporarily linear path. Then as they make decisions I can more easily anticipate where they'll go next.
These are the two basic methods I use to start developing my worlds. I call it a Ravenloft game, because that's the concrete I used to pour the foundation, but the rest of it I've written myself, with the occassional splat book for inspiration.
Mapmaking: don't even worry about this early on. My suggestion? Pick a shape. Add some funny bumps to it. Not big enough for how you imagine it? Add more shapes on your shape. Draw some lines through it (rivers). Get messy. Seriously. Maps are messy. Pick a size, and a general population. Nyaricus leans towards historical accuracy, I just pick numbers that sound good, then change them on the fly.
As you build the map (you'll see a theme here) write about each section as it occurs to you. Put a number on the map. Say, 1. And that number one becomes a set of pages in a binder (which is genius, listen to the person who said binder, I use notebooks and I'm always flipping around. Binder.) which then gain definition as you write about them. We'll say "1" is a capital city of a major kingdom. It houses people, like the King & Queen. Or just the King, who's just but gay, and struggles to his infidelities from his people. Or just a Queen, who's too good for her own good, and is running her kingdom into the ground because she's farming out recourses to help others. Or a Viscount with a horrendous reputation as he holds the throne for the future Queen who's still an infant, but is actually a major hero of the story, because he is Good and Just, but unapologetic about he runs the state and wants that Queen to inherit a worthwhile, prosperous kingdom.
And then '2'. And so on. I don't know if those suggestions help you or not; but the more you think outside the box, and the more you write, refine and invest in the project, the more your players will love it, because none of them - not one - love much more than a place they can believe in and be part of.
LCpt. Thia Halmades