I'm adding only a few sentences here because the thread is already full of great advice!
What would you expect from a sandbox that is not delivered by an AP, and how would you expect a published sandbox adventure to be laid out?
For me a sandbox is always like "a delimited area where nobody is telling you what you should do" (hence the name...).
That doesn't mean there are no possible "tracks" to follow, but rather that it's up to the players to
trigger something and choose to follow it, instead of the adventure track being presented to them on a silver tray and then presuming they will follow.
Because of this, what I expect from a sandbox is an area to explore freely, where the DM has "dropped" a series of "points of interest", which might be anything from a one-time encounter to a whole adventure ready to be triggered, e.g. dropping the whole Temple of Elemental Evil into your sandbox is always a lovely idea...
The difference might be a little like:
- Adenture Path: someone tells you there's a Temple of Elemental Evil that must.be.stopped.now, otherwise the whole world will end!
- Sandbox: you stumble upon the Temple of Elemental Evil accidentally while exploring an unmarked area on the map OR you track its location down following legends and rumors OR it was there on the map since the start and you just wanted to check it out. Then you decide it's your quest to stop it.
With that in mind, there is no such thing as a sandbox
adventure IMO.
The introduction: How railroady is too railroady? Is it ok to open with a framing story like "You guys are on your way to meet King Soandso who as a secret task for you. You were intrigued so here you are," or is that too contrived? I guess I'm asking, what is the best way to get the players moving in a direction, any direction? In previous sandboxes players have complained that they simply didn't know what to do next.
When not knowing what to do next, the players should just take another look at the map!
But since you can drop any adventure into a sandbox, nothing forbids you from dropping even a railroad... as long as you don't do it all the time, it can still be a sandbox. Eventually it might cease to be a sandbox if you give the PCs no choice but to accept the quest.
Player vs monster level: This is much less of a problem in 5E than in 4E, but how do you plan appropriately balanced encounters in a sandbox when you don't know when or how players will tackle them? Or is this a feature of the sandbox, that players will run into things that can TPK them? Or that they'll run into encounters they can steamroll? I don't like the Oblivion/Skyrim "monsters are always your level" play that 4E pretty much required (or you were forced to run linear adventures, which is what I want to step away from), but I don't have an answer for the level disparity problem (if, indeed, it even is a problem).
For me it's paramount for a sandbox game that the world does
not conform to the PCs' level.
But that doesn't mean you should place CRs totally randomly... and actually I think a MORPG like World of Warcraft (don't know the ones you mention) is pretty much a huge sandbox, but with the added feature that it's divided into regions of similar CRs. You always start in a low-level region, but as you progress further in the world, you can travel to higher-level locations. Narratively, it does make some sense: civilized people tend to live in reasonably safe areas (there aren't many permanent settlements built on top of the Himalayas, in the middle of Sahara, or inside volcanoes), while other areas might see a proliferation of dangerous species only, driving the weaker ones away.
When you put very high level threats in the starting area, just make sure there is a kind of barrier that restricts interaction between it and the common people.
The conclusion: When is a sandbox adventure over? My players are a big fan of free agency, but they are equally big fans of story, including the climax, the denouement, and the eventual end. But in a sandbox it seems like the ending is a lot more ambiguous, and more so the number of open threads you have going on. One thought I had was having all extant story threads funnel into one overarching epic, with all the foes they have fought along the way being pawns of one BBEG, but perhaps that's been done to death? Contrived? I don't know.
Well as I said, I don't think there's a sandbox
adventure, but the sandbox contains adventures. Each of them can work the same as any published adventure (although probably you don't want a series of
end-of-the-world-threat types), so they can have an end. The sandbox itself never ends.
Player paralysis: With no big sign saying "go here, do this" my players are apt to scratch their heads and say, "I dunnow." And I by no means blame my players. I don't really think I'm that great of a DM, so I am either laying too subtle clues or I'm over-complicating my stories. I think it was Angry DM who once said to me on Twitter that even having a story (and I might be grossly oversimplifying or wildly misinterpreting what he meant) meant I was tacitly railroading my players. If that's the case, is player paralysis a function of their expectation that I have a trail for them to always follow? How do you instigate player action that more organically generates adventure?
Just bring the map back on the table, or keep a list of interesting locations/characters/events/kingdoms/organizations that the PC have encountered already, and might want to go back and see how things are going with them.
Setting material: In no event has a player ever read any material I have ever written for an adventure. So is it a waste of my time? Is it still any good for internal consistency? Is internal consistency even necessary for player immersion? When you read a published campaign setting or sandbox adventure, do you as a DM actually read things like calendars, historical timelines, and exhaustive breakdowns of churches and factions? Do such things enhance sandboxes and where is the line that you've written too much?
Yes I think those examples are all enhancements, but do not require the players to 'learn' that stuff...
About history, it's probably best to just sketch it. After all, even in the real world we don't know the details of our history. Roll back to the middle ages (as that is the typical era which inspires most D&D settings) and try to imagine how incomplete the knowledge of history was at that time... Fuzzy knowledge is totally fine, and you can always let the PCs find out more themselves.
Churches and factions are great for building PC motivations, long-term ally/foes relations, and of course new adventures. I would rather spend more time detailing those, than locales and past events.
Finally, remember a couple of more things:
- keeping everything in mind is an impossible task, so don't set this as your standard
- there is not that huge difference in results between generating ideas while playing and pre-designing everything in advance
- the fantasy world itself changes, thus DM forgetting something just equals that something happened in the world