I'm a proponent of starting with a map.  First off, I like maps, and secondly, if you're using anything like real-world physics, you're going to want your game world seasons and weather to align with what we would expect (ie, very high latitudes are cold while very low latitudes [equatorial] are hot.  If you don't have deserts where they would be expected (ie, rain shadows from mountains, horse latitudes), you should have an explanation.  Basically, I assume everything is 'real world sensible' unless it has a reason for not being that way.  A map is really helpful for that purpose.  

It'll also help direct 'building out'.  If you are looking for a place that's somewhat like England in weather and temperament, the map will help guide you to that.  Then looking around, you'll see what's nearby and that will give some ideas of what could reasonably make sense to exist.  

It's absolutely fine to take an Earth-like culture and imagine how they'd be different if they lived somewhere else.  This is also helpful for non-human cultures (ie, if resources are similar to what we'd find in our world, we'd expect some of the same types of crafts - if there are no sheep, there won't be any wool).  

With a general map, keep things vague.  If you know you have some cultures inspired by the real world, you'll want to indicate generally where they are, but the biggest mistake people make at this point is filling in too much detail.  For most game worlds to make sense, while the Kingdom of Kinglandia may claim a huge amount of territory, a fair amount of it will be wilderness.  The frontiers are not going to be well defined, nor will there be clear borders.  Instead of drawing a modern political map, you might want to think of it like a map of barbarian tribes - generally indicate the center of their population, but don't draw in borders - they're unlikely to be entirely fixed.  

Once you have the general map and the broad strokes, focus in on a small area to really get going.  Where the PCs play is going to need a lot more detail than the parts of the world that they won't see (at least for a while).  Think of what you know (or don't know) about China.  You may need to answer some pretty vague questions, but you're not going to have to go into a lot of specifics - especially if your world is not as well-connected as ours.  

Regarding pantheons, I personally recommend 'distant gods'.  Your deities should never jump in and start smiting people with their avatars.  Firstly, it's not really necessary, and secondly, it can really detract from the fun of your setting.  If the sun-god is mad at the PCs, he tells his worshipers in a dream to go kill the PCs.  Can anyone prove that he really said to do it?  Maybe the sun-god worshiper just really doesn't approve of what the PCs are doing.  This is going to also help your PCs feel like they matter.  You never need to justify why your God of Just Retribution isn't getting involved - you've basically made it from the outset that Gods need mortal beings to work their will in the world.  There's going to be someone that says 'well, the deities don't get involved because than the other deities will get involved'.  That usually doesn't work out well unless you have absolutely balanced pantheons between good and evil and it tends to make it much harder to justify multiple pantheons.  With distant gods, you can have the Greek Pantheon and the Norse pantheon and some people think those are the same gods under different names, others think that there are 'true gods' and 'false gods', etc, etc.  Basically, religion becomes a lot more like 'the real world' - with all the conflict that naturally arises.  

At this point, you have a map with some generally defined geography, some rough ideas of cultural groups and their locations, and a zoomed in area of a few hundred square miles (maybe larger) for a campaign.  As you have time and inspiration, you can expand the area.  If this campaign ends, you can set the next one in a different area of the same world and that will help expand the world with a lot of depth.  As you come across something that you think is interesting, decide if that can fit into a part of your world that hasn't been rigidly defined yet.

I want to thank everyone who has contributed to this thread.

I've always used manufactured settings (because it's easy) but now my imagination is sparking and I want to run with it before it's gone.

It was, in my mind, shaping up to be a daunting task but I feel I've enough ideas now that I can carry on.

Special thanks to Ravenswing for his blog posts addressing this issue.

Any time.

Quote from: S'mon;752488IME the best worlds grow organically in play...

Here's a MSPaint version of (most of) that map, from ca 1996:

Win!

Quote from: Thanos;752433I know all the things I don't want but don't know what to actually put in the damn thing. I have a few ideas but don't know how to pull them all together.

First: Do I draw a map? Build the nations/societies?

I'd like multiple pantheons (I think). How do you make something like the god(ess) of the harvest different from one pantheon to the next? Can you? Should you?

Maybe I'm just realizing this is going to take more effort than I thought, becoming more work and less fun.

Looks like there is going to have to be some research. I don't know. Just thinking out loud.

People found my "How to make a Fantasy Sandbox" helpful.

As for religions and societies, culture is a set of common behaviors that a group of individual centers around. For gaming purposes this means defining how a group commonly acts means you also defined their cultures.

The same for religions which can be looked at as the culture surrounding the worship of a deity or dieties. Have two gods of harvest, want to figure out how they are different. Figure out how their adherents behave. After then figure why they behave that way which will give you their history.

My biggest tip is 'make up some shit you think will be fun'.  My current world that I'm starting to build will include everything from Prester John's Kingdom from The Travels of Sir John Mandeville to Captain Shakespeare and his Lightning Pirates from the movie "Stardust."

Second tip... you don't have to outrun the Balrog, you just have to outrun Gandalf.  In other words, don't worry about drawing a 10,000 level dungeon... for now, just draw one level, the "town", and the "castle."

It was a year of playing in Gary Gygax's "Greyhawk" before the temples were called anything other than "The Lawful Temple," "The Neutral Temple," and "The Chaotic Temple."

I agree. I've always started with one hex and two dots: the town dot and the dungeon dot.

Just yesterday I found a nice tutorial for the opposite method: top-down building based on geology and meteorology.
http://www.dragonsfoot.org/forums/viewtopic.php?f=42&t=29425

This is a damn cool thread actually. My two cents:

When I was teenager, I used to start it all "top down", maps of continents, several pantheons of gods and their spheres of influence, a detailed five-age history of the entire world, with elder races ... blah blah blah ...

Now I feel it's much more fun, and more mysterious, to only detail a corner of the world and have the rest in broad strokes. For this, a continent map for the GM only is a good idea, and some names of exotic places to throw around from time to time, but I've only detailed two religions (that of the native race to the corner of the world I've worked up, and that of the invading colonists ... Already the two religions, and the fact that the "Old Religion" is now outlawed, for the most part, have given rise to loads of plot hooks and scenario ideas), and two cultures. The characters can choose to be either from the colonist or the native culture, and interaction between the two, and a brewing insurrection, are already proving to be the hook to hang the campaign on. For another series of games, my "Swords & Sorcery Anthology", I've set it in a Carribean with plantations type region of the same world, with a mix of cultures and religions, but only detailing the ones that the characters come into contact immediately with, and basically taking my cue from what the players want to play. They had the brief to come up with pirates from any Earth analogue culture they wanted, and we ended up with a "local" from the Siyana Archipelago (Malaysian in feel) and member of a grim and outlawed Shark Cult of Ravenashoor, a "European" pirate, from the slightly Renaissance Italy inflected Elerinn in Harronia (the world's White "Western European" region) who's gone in for a syncretic animism featuring local and imported spirits and demons, and a "Mongol", a nomad tactical thinker, exiled from his people, of the Varanghar tribe of the Jedekhan peoples of the Grey Plains, in the northern continent. Throw in some native animal totem animism, and hints of dark and strange gods all thrown together in the island melting pot pirate port they find themselves in, and we're good to go!

My own methodology runs a little contrary to what several here have suggested: people seem to think the best way is to start with a very small area, and then build the world up around that.
But for me, what has been by far more successful is to FIRST think of the big overall world or region in very general terms, and only AFTER that, focus on a more specific area and develop that area in depth as your campaign starting-point.

Yeah. You have to have some concept of the world before you start even if there's no map per se.

Quote from: RPGPundit;753669My own methodology runs a little contrary to what several here have suggested: people seem to think the best way is to start with a very small area, and then build the world up around that.
But for me, what has been by far more successful is to FIRST think of the big overall world or region in very general terms, and only AFTER that, focus on a more specific area and develop that area in depth as your campaign starting-point.

Mm, well, to a degree, I'm talking out of my ass: I started developing my setting decades ago, and I've layered a staggering amount of detail into it.  It's not merely that the giant seaport I've used for most of the last decade as a home base for the main party has eleven hundred businesses described with at least a large paragraph each, it's that I've done the kingdom's ten largest cities as well, and have similar treatments for every village within two days' ride of the capital.  I know where every temple of every faith in the world is, as well as the locations, numbers and Orders of every wizard of Master-class on up.

But I've been building this setting for approaching forty years now.  I've had the time not only to look at the big questions, but all the teensy ones as well.  I haven't had to create a completely new setting for a loooong time.

But if I was going to do it all over again from scratch?  No way.  No freaking way.  I'm not going to do months of work as a prerequisite to getting a game off the ground.  I'm going to want to tell people, "Hey, I'm starting up a new game, want to kick off a week from Saturday?  Great!"

So with that, I've quite enough work ahead of me: putting together a dozen key businesses for the small town they're starting in, putting together the first adventure, doing up some regional customs and practices, deciding the details of three or so key religions.  I'm not going to say I couldn't do that in a week and a half (along with scheduling character creation and, well, working and making sure my wife doesn't feel neglected), but I won't have broad-picture creation time to spare.

If I had 40 years to build a setting, I would do a good, solid six months of work on it. Swearsies.

Have we mentioned this yet?
http://www.welshpiper.com/hex-based-campaign-design-part-1/

Aside from the odd choice of five-mile hexes, it's an excellent way to get started on the big picture stuff.

Quote from: Thanos;752433I know all the things I don't want but don't know what to actually put in the damn thing. I have a few ideas but don't know how to pull them all together.

First: Do I draw a map? Build the nations/societies?

I'd like multiple pantheons (I think). How do you make something like the god(ess) of the harvest different from one pantheon to the next? Can you? Should you?

Maybe I'm just realizing this is going to take more effort than I thought, becoming more work and less fun.

Looks like there is going to have to be some research. I don't know. Just thinking out loud.

just imagine it in your head. Steal profusely and adapt.
Once it becomes a lot of work meh ...

The gods thing is more tricky you need to decide do you have a group of gods worshipped by different groups with different names or are the gods in competition or are some of the gods false etc .... You can do loads of work on theology but your players will probably not care and because they aren't familiar the gods won't mean shit to them. Its why I tend to steal gods from books and stuff so to be popular lift the gods from Game of thrones. There are the old gods , vague and hard to fathom, the Seven the state religion for most of the game world (using a well known pantheon means that PCs can start to use in game idioms like "by the Smith's Hammer that is a fine suit of mail" ) and the new fire god fromt eh east.  
So this means the PCs sit down and immediately understand your religion from a paragraph not 30 pages of detailed notes. And if they aren't familiar then that PC isn't very religious.

Quote from: RPGPundit;753669My own methodology runs a little contrary to what several here have suggested: people seem to think the best way is to start with a very small area, and then build the world up around that.
But for me, what has been by far more successful is to FIRST think of the big overall world or region in very general terms, and only AFTER that, focus on a more specific area and develop that area in depth as your campaign starting-point.

yes.  exactly what I said.  Sometimes, I do work like Pundit.

Well, better to do things right sometimes than not at all, I guess.