No, this is not about convertibles, nor about drinking.
Once upon a time I studied computer science, where I first learned the concepts of
top-down and bottom-up design.
Top-down design focuses on the overall structure and the relationship of parts, and then adds in detail as the structure is resolved.
Bottom-up design focuses on the little pieces, leaving the relationship to be worked out later, on the presumption that once you get the details working, you can make them work together.
These two choices are available as design choices to dissertation writers. I think writers in the humanities are usually encouraged to do bottom-up dissertation design: first you write one chapter, and only once you've written that chapter do you then go on to work on the next chapter. In the sciences, I think that the structure of empirical studies forces the writer to use a little bit more of a top-down approach: the link between a literature review chapter and a methods chapter is clear: the literature review lays out the basic theories that you are working with, and then the methods discusses how those theories are operationalized. By contrast, a humanities dissertation that might, for example, have a series of chapters on different authors, and the structures and relationships of these chapters are far less clearly defined.
I'm a firm believer, in writing a dissertation, of a top-down approach. The more that an author can develop a sense of the overall project, the more easily that author can see what to talk about, and when. My experience with clients suggests that it's not a paucity of ideas that people struggle with, it's the organization of those ideas, and the making a coherent project. My experience suggests that writers with plenty of material often are thinking they need more material when what they need is the top-down view.
I know that I'm swimming upstream on this one. My experience indicates that it is almost universal for students in the humanities to try to write one chapter at a time, and for them to be encouraged to work that way. Finish the first chapter; then move on to the next. It's not how I like to think about these things, and I recognize that this is a matter of opinion.
My qualifying exam included a two-week writing period during which I had to answer five questions. When I was given the questions at the beginning of the period, I said something to my examination chair about making the answers chapters in one larger whole paper. She said I shouldn't try; that it would make the work harder. But I couldn't help it; it seemed natural to me to try to respond to each individual question and to try also to show how those questions formed part of a coherent whole. For me that sense of coherence provided guidance that helped make choices about where to include material, about how to use discussion of theoretical issues in an important fashion, etc. Maybe my preference for top-down writing is merely a reflection of a personal bias.
But I think it helps. I think that if we can see how the project as a whole is working, then we have an extra set of reasons that help us structure our discussion.
Recently a client was told "your chapter isn't complete; finish working on it before moving on." Obviously the advice of the dissertation chair is not to be taken lightly. But I really want to suggest to the client to keep on working on the next chapter (on which the client had started working after turning in the previous chapter draft). I want to suggest this because I think that working on the next chapter will give insight into the previous chapter, too. As the dissertation chair's comments were largely about making it easier to see and follow the points (i.e., the comments were on the writing, not the material), I can appreciate the top-downness of trying to make the chapter more coherent. It's hardly wasted effort to go back and work out the organizational and presentational issues that make writing hard to follow; a lot of good insight can come of that.
But I think it's six of one (working on the same chapter) vs. half a baker's dozen (e.g. 6.5; working on the next). Working on the organization of the same chapter will provide insight into what the structural issues will be, and how those will relate to the points that are to be made to build an argument. The thing is that working on the next chapter can provide the exact same insight, except in a larger context--in a context of writing a second chapter that works together with a first. In one way that's a slightly more complicated issue, but in another it is easier because of the additional perspectives available.
As for bottom-up?
I think there's a great place for bottom-up writing. I don't think it's as predictably time-efficient. It requires the synthesis to bubble up accidentally. There's a lot of value to be had by sitting down to simply write a little about your ideas, or to write a section of a chapter, or to write a chapter without worrying too much about how it will fit with the rest. But I think there's more value by having a better top-down vision.
Is it more complicated to do top-down, because then you have to accommodate more factors? Or do those extra factors help reduce the number of possibilities, thus simplifying the decisions to be made?
If you're getting stuck because you're having trouble coming up with ideas, then maybe bottom-up is better. If you're getting stuck because you have too much material and too many ideas and you don't know what to do with it all, then top-down might be better.