Over at Gnome Stew there is an interesting post about a forest-based megadungeon. I like the concept, although it's potentially going to be set up as "you're lost in this woods, now get out!" instead of "there is this crazy woods with stuff in it you want done, now get in there!"
Naturally I noticed this become of traffic referrals to my blog, which is awesome. Rather than comment there, because that would take some minor work of signing up and logging in, I just decided to write some comments here on the "escape the dungeon" approach.
I see some downsides to "escape the dungeon" or "locked in the dungeon" as a campaign.
First off, the incentives in the game are backwards. Instead of seeking adventure, adventure is an obstacle to your goal. Instead of accomplishing things, you're hoping to avoid entanglements that drag you down. Instead of an iterative process of discovery you're in a linear process of attempting to escape.
In other words, if the goal is "run from" instead of "run to," your incentive is to just get away not to go and do once you get there. You aren't drawn to what makes a big sandboxy play area fun - clearing, looting, fighting, exploring, meshing the knowledge gained at point A with that from point B to discover point C, you're just trying to get out.
You've basically inverted play in a way that makes most of the activities that drive this sort of game obstacles not rewards. Even treasure is an obstacle - that stuff will encumber you.
Wasted Work or Forced Participation
Second, unless you structure the dungeon or play area as a fairly linear or looping-back maze, much of it will not be used. Simply put, if the goal is to get out, PCs are incentivized to leave the area and bypass as much as they can. You'll need to force them to go through areas to ensure they'll actually make it there. Once they're physically past the beginning areas anything they bypassed is wasted unless you give plot or structural game reasons for them to go back. If you do this often enough, players get the impression that everything is required and it's a linear game in all but name.
It can be frustrating as a player, too. Nothing drives players more batty than a place they want to get into but can't. Nothing drives them to more bored frustration than a place they want to leave but can't. "We're stuck" is a terrible feeling, and it wears on you in a way that "I can't figure out how to get into that bit of exciting play area" actually quite enjoyable, especially when you figure it out.
Failure is the norm.
Every session that ends still in the dungeon is a failed session. It might be progress towards success, but you haven't succeeded yet. You're still stuck in the dungeon or the maze or whatever. Contrast that with a go-and-do dungeon - even the act of going there is a partial success as you've made another iterative attempt to get things done. Every little bit you do gets you more options, and your "to do" list grows with success. In an escape situation, every little bit gets you closer to success, but it's still technically failure as you haven't escaped. If the goal is "go into the dungeon and get stuff!" then any going, getting, and stuff is at least a small success. If the goal is "get out" not getting out is what happens the whole campaign until it ends.
It Must End
A game where you're escaping a place has a written end - it ends when you escape. If you are drawn back into it by unfinished business, you may as well have made the game about the unfinished business and put that in as a drawing point. A classic dungeon may have an end, a classic megadungeon may have one (chute to China, W*E*R*D*N*A on level 10, etc.) but might still draw you back in later on. It can be extended ("Hey, there was a secret entrance to five more levels this whole time!") if people want more. It's harder to do that in a positive way when the game is about escaping the dungeon. "Turns out, you escaped the dungeon only to find the outside . . . is another dungeon!" feels more like a gotcha. It's every horror movie that ends with a setup for a sequel, except you've told people that more adventure is more failure to escape like you're trying to do.
For those reasons, I prefer to have a game with a magnet than an anchor. I like a draw to go somewhere for some kind of gain rather than something you're forced into and have to escape.
I really try not to be negative on this blog. I'm hoping this comes across more constructive than critical. Using "escape the dungeon" as a sub-theme works. It's how I've launched a number of games, and it's the backbone of some very solid published adventures and adventure arcs. But makes it so easy to use in a short burst is what undermines it in the long term - it's ultimately something you need to end, not prolong.
All of this said, it's not really unfair to have a bounded area of play. It's also not unfair to give people a reason to be rooted down in a particular spot, either. I just find it is more effective and has less downsides if you give reasons to stay than to blatantly force you to stay somewhere.