I'm not sure if that's fair to say that encounter development on the fly is the easiest to improvise. It might be at very low levels, but, the common complaint in campaigns over about 12th level is the sheer amount of fiddly bits you have to keep track of. While I know I could improvise an encounter with 1d6 dire rats, I don't think I could do the same with 1d6 elite fiendish venomous cryohydras.
But, yes, if you want to run a "sandbox" campaign, then the ideas I'm putting forth are certainly not the way to go. Honestly, from my own point of view, I don't think I would enjoy such a campaign. When I've played in that style of campaign, it invariably grinds to a slow, stuttering halt as the party wanders aimlessly from disconnected element to disconnected element. I know that it can be done right. I've just had very bad experience with playing and attempting to run them.
I realize that many people don't follow RAW exactly for xp, but, I did a poll a while back about how fast people leveled and it did appear that most people do follow RAW. Levelling in 2 or 3 sessions most of the time.
Bear with me for a second while I try to organize my own thoughts. I've kinda got three points in mind that need to be connected.
First off, we can be fairly sure that most campaigns only last between 1 and 1.5 years. Numerous polling shows this. While I know there are multiyear campaigns out there, for most of us mere mortals, 18 months seems to be the half life of a campaign.
Secondly, again according to polling I've seen on Enworld, most people shift campaign setting with each new campaign. Most people end an eight month Forgotten Realms camp and move on to a homebrew. When they finish that, they move over to Eberron. After that, maybe Kalamar or whatever. The point is, there is not all that much setting loyalty in many groups.
Thirdly, and this goes back to the xp bit, you can craft a 1st to 20th level campaign in about 8 adventures. By RAW it takes about 20-25 encounters to go up two levels. Assume for a moment that you overbuild on the assumption that the players aren't going to hit every room and we can ballpark about 40 encounters at the high end for a 2 level adventure. 40 encounters is a fairly doable number.
Now, combine all three of those points - an average 1-20 level campaign is only going to last about 18 months and then the players are going to move on to a different setting. Suddenly developing large swaths of setting isn't all that important.
Make 8 adventures plus maybe a half dozen one shots and you have a full, 20 level campaign.
Suddenly, campaign design looks a whole lot easier to me.
First off, I don't think it is possible to run a game in which world building elements do not accrete over time. No matter how much prep work you do, the players tend to take you on angles that you haven't thought of before, or even ask questions to which there is no prepared answer. So, I would say that allowing details to accrete is mandatory.
Agree 100%
RC said:
Better yet, my mapping the island made me include some shark-haunted reefs on a whim. Now that I am looking at that map, though, I have a great idea for an adventure that takes the PCs to the ruins beneath that reef, discovering exactly why sharks congregate there. So, I write a brief outline of that location. I do the same for several other locations that interested me when doing the map prep.
I guess my point is to instead detail out the shark reef as an adventure and then come up with a way to link that to the adventure before it and after it. Or, make the adventures recursive enough that you could do the adventure after it and come back to Shark Reef.
When I said not buying into the campaign, I was perhaps stating things too strongly. But, think of it this way, the original hook is heading into the jungle to investigate the disappearance of the Chieftain's daughter. The players turn to you and say, "Naw, we want to build a raft and sail around the island." To me, that's not buying in to the campaign.
Now, hopefully, the campaign arc would include more than these simple examples. Hopefully, as I mentioned before, you could make the adventures recursive so you could part of A then part of B, followed by going back to A then off to D, finding D too hard and moving back to C and so on and so forth.