Some ideas make me angry -- not because they are bad, but because they are so elegant and so strong that I am angry I never thought of them myself. The design of Robin Laws' The Armitage Files for the Trail of Cthulhu RPG is one of those examples. Robin Laws provides a novel method for designing a campaign in a more improvisational way. In this, he provides an easily portable structure to facilitate a more open, player-directed style of RPG. It is the basis for the forthcoming Dracula Dossier and can be adapted to The Strange campaign design with little difficulty because it is essentially system agnostic.
Before I discuss the improvisational approach, I want to remind people of typical campaign design (from published campaigns, at least). There are two typical approaches. One is to have a series of keyed maps. Each location has a number and the number reflects a singular description within the campaign guide. There may be some opportunities for open-ended interaction between locations (say in town) or some rules for modifying the locations based on events in other, prior locations -- but the campaign is largely static.
None of this changes dramatically if you move from a map to a process diagram (as in, say, the Dark Spiral campaign). Your campaign may not go through all of the nodes in the process diagram -- creating some variety. However, the key elements are largely the same with minor modification. The climax will involve the same protagonists with the same goals and largely the same capabilities. The same holds for the core elements of the adventure.
This has been a problem from the beginning of RPG design, really. Of course, individuals developed their own sandbox approaches to home campaigns. Communicating such a design through a professional project has been difficult, though. I think Robin Laws has found a great way to do it -- and it can help you design your own sandbox campaigns.
In this improvisational design, you focus on designing elements rather than encounters. For each element, you create multiple potential uses. You then provide the players with a narrative that allows them to choose which elements they think is most important -- and let them be correct, whatever they choose. This likely sounds vague -- but I will flesh it out.
In The Armitage Files, Robin Laws develops the key elements for a Trail of Cthulhu campaign: people, organizations, and artifacts. For each person, there is a general description followed by a sinister, neutral, and helpful version. The GM can decide on the fly which version is appropriate at a given time (locking that choice in for the campaign, of course). The same is the case for organizations -- they can be helpful, neutral/red herring/uninvolved, or full antagonists. Artifacts can be hoaxes, of limited power, or greater artifacts with tremendous power. Each is left undefined until used in the campaign.
The Armitage Files campaign starts with the PCs getting a mysterious letter mentioning several potential leads and they can choose which one to investigate. The key to the improvisational campaign design is that the players are correct (largely) whatever they choose. If they choose to look at one suspicious person, the GM rewards that choice by making that person interesting (either helpful or a villain -- depending on the pace one wants for the adventure). You could, hypothetically, thwart player interest by making all of the leads uninteresting or red herrings until they go for your preferred element -- but that is contrary to the spirit of the improvisation design.
Allow me to provide an example of how this could work in The Strange.
At a briefing session at The Estate (or before the game session even -- to get things rolling quickly at the session itself), the players are given a dossier about a scientist whose research suggests that he may be involved with the September Project, some strange events centers on a crack house in the same city, and reports that an artifact may have been picked up by recursion miner. The players have to decide, in the context of the dossier, which lead is most important. Which they choose, they are correct (you can mix in some false leads later to taste). In The Armitage Files, these leads are in actual mock documents -- which is really cool. You may not want to invest that much in the leads.
Let's say they players choose the scientist as the starting point. You will have prepared a good, bad, and neutral version of the scientist. It is generally best to start the first element as good or bad -- so you choose one. If good, maybe he is aware of the Strange and actually trying to prevent some bigger problem -- launching into the night's session. If bad, he is developing dangerous tech and the players have to get in to stop him. You decide based on the context (where that session is in your campaign, etc.). You can link the first element to others - or not - as you deem necessary. It is probably best to avoid making things too complicated.
At the end of the session, the players get a new lead that points to several potential investigations (another person, organization, or artifact). This sets up their next choice and the next leg of the campaign.
You can't tell whether the campaign will be after several sessions. GMs invest in designing these elements and following where the players lead. In this mode, you do not create a plot per se'. Instead, the elements collectively design the plot as you go. The GM can define the plot to a limited degree by the sequence of lead sets (like the dossier) that are provided at a given time -- but the actual play will follow more from the choices of the players. Invest in several versions of the elements, not in the plot.
This does not require much adaptation for The Strange. A focus on people and organizations still works. Instead of artifacts, you may want to focus on items more generally or add locations as an element category (there is some of this in The Armitage Files, actually). For each, design several versions including dangerous, helpful, and innocuous versions. You will only choose one. If you are worried that this is wasting development time, you can always salvage an unused version attached to a different element later.
There is more to the system that I might expand on later (campaign spines, the generic adventure mad lib, etc.)-- depending on the interest level in this post. I think just this description may be inspiring though. I strongly recommend The Armitage Files. It really is inspiring -- infuriatingly so.