In my teens and early twenties I made three sustained attempts to create worlds.
The first two were for AD&D games, the third was for a freeform – i.e. largely ruleless – role-playing campaign. Each of them ran into two basic problems – one in-game, one out-of-game. The in-game issue – we’ll call it the Balance Problem – is that a role-playing world needs to accommodate the kind of adventures players are likely to have: if the characters are powerful enough and the adventures wide enough in scope, it’ll most likely affect the status quo of the gameworld. The out-of-game issue – we’ll call it the Botherd Problem – is that different players will have different levels of commitment to caring about the game world – and almost all of them will care about it less than you.
The first time I world-built I got round the Botherd Problem by ignoring it and not bothering having any kind of background beyond a vague “long ago the world was high-tech” thing. I dealt with the Balance Problem by introducing a big overall quest involving moving east and simply having the countries become more and more lawless and dangerous the further east one went. Sharp-eyed readers of fantasy books might recognise this solution from, well, almost all of them published before the 1990s. It makes worldbuilding extremely easy: you have the good guys at one end and the bad spooky evil guys at the other. This kind of basic world-creation worked very well and my friends and I had a good time with it, but the truth is the good time was nothing to do with my background work.
The second time I world-built I took it more seriously, and took a lateral approach to the Balance Problem – I intended my players to get powerful enough to affect the destiny of their world. I was aiming for epic – a lot of adolescent referees do. I ran smack into the Botherd Problem, which was that I’d gone to some lengths to think about what each country on the map was like and who was allied to who, and my players didn’t give a monkeys. But in order for them to play a part in an epic they need to care about that stuff. And if they don’t, in plot terms you end up railroading them into caring – forcing the gameplay along certain lines. In which case what they’re essentially doing is reading out a fantasy book you’re writing, and to pull that off you need to be a better writer than I ever was.
(Tellingly, I can’t now remember any details AT ALL of the actual world I put together for that game.)
The third attempt was something different, and an approach I’m still pleased with. Being a freeform game, the characters didn’t undergo big rules-mandated jumps in power levels, which made the Balance Problem only as big or small as we wanted to make it. We were able to keep things low-key and have a group bumbling around and barely resolving anything, having no impact on their setting at all. Instead of experience points, players would be rewarded with “knowledge cards” – index cards I’d written or typed out which had some details about the world on: a local legend, some political rumours, a bit of history. Some or all of this might turn out to be relevant: I myself had only the barest idea of where things might be going, but that didn’t matter as long as the players thought there might be a design behind it. It meant players could discover the background in a drip-feed, rather than be burdened with info and crash into the Botherd Problem.
Freeform games have their own perils, and this one collapsed into chaos, as many do. A particularly bad mistake was to introduce a parallel psychic world with a tenuous link to the ‘real’ one, mostly as an opportunity to play while drinking and/or smoking weed: it included as antagonist Sauron-figure a thinly disguised analogue of Bono. But in general this was my most successful attempt at world-creation, because I kept things small.