Briefly, I think it’s a great idea.
My Campaign and Scenario Design series talks about how to do exactly what he’s talking about.
The way I go about it, I can work at the scenario, campaign, or setting level with about the same ease, defining only as much as I need to right now. The templates I use capture what I need at a narrative level without digging too much into the mechanics of it, and they are really quite modular. When I devise a campaign, for example, I might map out various scenarios (‘adventures’/'modules’) that I think are likely to come up related to the campaign, and why the scenarios might happen and how the PCs find out and… so on.
Not all of them will be used, and during play I may realize others are possible I hadn’t thought of. I might need to be careful in how I integrate them, but I can expand on or change the direction of the campaign. Oddly, it’s usually easier to make a bigger change (a new scenario to a campaign) than it is to add a smaller piece (a new scene to a current scenario). This is mostly because I can usually make campaign-level changes offline, while scenario changes tend to happen real time, but also because they work at different levels of abstraction. I don’t need to make a scenario evocative and exciting (and rules-consistent) Right Now, I don’t even need details (like, say, names) for a while yet.
I like to use the GDQ series of modules as an example, even though I consider it kind of crude in execution. I might mentally map out several different scenarios (each of the giant strongholds, including some that weren’t present in the original series), providing links between them (and even hints and links to later — even the Hill Giant stronghold had links to the drow, though they were almost impossible to use at that point).
Doing it this way you don’t actually have to prepare detail until it looks like it’s going to be needed. I know there are five or six giant strongholds that could be raided, and I’ve identified the likely links (clues and travel paths) between them (and perhaps some of the benefits of pursuing them — special items or scenes I can imagine for inclusion, though I don’t push the non-critical elements yet, I just note the cool stuff that comes to me).
It doesn’t take long to map out a campaign this way. It can also lead to some wicked player agency because you need to think ahead about how things link up and can thus provide information about them (unlike the original series, which ramrodded the PCs through the giant stronghold — ‘go or we kill you’ instructions from your ‘noble’ employers if you balked, as I recall). You might get a hint that there is more information at this location, that that location is probably too hard to survive yet, this other one is where you will probably learn the most but you need to find out where it is, that there is some excellent tool/weapon/magic to be gained over here than can probably help…
By focusing your initial work on high-level (hierarchical, not power necessarily) entities and their relationships early you can get a flexible campaign with lots of player agency built-in. You don’t even necessarily lose much work if the players decide not to bite — the initial time investment doesn’t need to be very large, really, and any initial detail work you’ve done can probably be salvaged for use elsewhere.
What’s not to love?