Thoughts on campaign structure

A few things have been mulling around in my mind lately about how to structure a good campaign (good meaning: 'the kind of campaign I want to run'), and what sorts of limitations can be put on the game to achieve that.

My current Age of Chaos campaign was designed from the beginning with certain in- and out-of game limits which were attempts to define how it should work. Here are some of these principles:

1. A fixed group of characters is not required. This was the original impetus for the campaign, in fact. To be able to run a game with players who aren't all able to commit to coming to every session.
2. Each session should be its own self-contained adventure. This kind of follows from the first principle - if the group of characters can be different every session, then the normal idea of a campaign as a continuous story doesn't necessarily work out.
3. It's a sandbox - the players are the ones who drive the action, through their choices of where to go and what to do. The DM does not have some over-arching plot in mind which the PCs are implicitly expected to follow.
4. No moral imperative - connected with point 3 - the campaign won't take the form of saving the world or suchlike. It's more about exploration.
   

It's interesting, looking back, how the campaign has developed. The first point has become pretty moot, as the group of players actually is more or less the same every session - there's a core at least, and the occasional drop-in from another friend. This has also nullified the second point - the story can continue directly from where the last session left off. I guess I came up with those guiding principles from perhaps a more idealistic perspective (the idea, which really appeals to me, of having a large rotating pool of players), rather than a practical point of view (that actually we have a pretty static core of 4 players in our RPG group).

So far point 3 has gone well. I've not deliberately introduced any pre-planned story arcs (although hints of plots and subterfuges have come up during play, which is only natural), and the players have chosen their course of exploration. I do get the feeling they're a little shy of choosing their own path, which is perhaps due to a difference in experience, as they mostly grew up on D&D 3.5 or later, whereas I grew up on Basic/Expert D&D. I think the emphasis in those two eras of the game was pretty different. Since reading ars ludi's thoughts on his West Marches campaign I'm very much taken with the idea of the players deciding between games where they want to go next, giving the DM some time to prepare something which they might encounter there. This unfortunately also requires knowing how many people are going to come to each session - it's all very well planning a daring raid on a Goblin city, but when only two players come to the game it suddenly doesn't seem such a good idea.

Point 4 is a tricky one for me. The campaign currently seems to be veering in exactly the opposite direction - a band of witch-hunters out on a crusade against Chaos. It's not really what I had in mind, and I can't quite put my finger on what it is I don't like about it.

All in all it's very interesting to see that really what I'd prefer to run is a series of more modular games, where each session has a pre-defined emphasis ("we're going to explore the ruins of the old mill", "we're going to try to steal the Duke's diary from his library", "we're going to follow that treasure map we found", etc). A sort of modular sandbox. I want the players to be totally free in their course of actions, but I guess I want to know up-front what they intend, so I can prepare something engaging, and I want each session to be an adventure - with enticement, danger and reward (or pain)!

Ah I guess I'm just suffering from having had three slow-paced town based sessions in a row! Still, things are looking up - strange things await the PCs in those cellars...

One conclusion I can draw for now though is that I'd definitely like more players, to enable this kind of tag-team adventuring...