Not to be silly, but isn’t relying on the Finder, relying on one application to manage different kinds of data? Don’t worry, I’m with you on the concept of keeping programs tidy and focussed on their intent, but I’m not sure I follow you on this argument. An application should be able to address different kinds of data, if those kinds of data pose an intersection of purpose in alignment with the intention of the application (like the Finder). You even mentioned one yourself in a positive light. Preview can handle several different kinds of data (and in my opinion it fails at PDFs, but that is just me).

So the question is whether or not having research data together with your working draft a valid intersection? Some people couldn’t live without it. The ability to stop and start an audio stream while writing is very useful, or to have a PDF loaded in another split. For those that don’t need it, well, just don’t drop media into Scrivener then. :) It isn’t as if you are forced to do so. There is even a handy way to cross-reference URLs and local files to parts of the draft.

I think it does less managing than you think. There are no actual editing features for non-text media types. You can however open them in proper editors from within Scrivener, edit them and save them. In this sense it acts more like a hub (Finder) for data, than something which tries to be everything.

How is having a Scrivener project better than just having a project folder?

It actually isn’t all that different, and where they are different, there are valid pros and cons for each method. Scrivener projects are just bundles. The only “proprietary” data is binary optimised XML for speed. You can even disable that by turning on external versioning support, which makes the whole project plain text at the slight cost of saving and opening speed. That’s really it, XML, RTFs, and whatever other kinds of media you put into it.

And how good is it at legal/academic writing, which relies very heavily on footnotes that must be formatted just so?

It isn’t a formatting engine, as once again it doesn’t try to “be everything,” rather it is a drafting tool. Something you use before the final stage of making it look pretty or up to spec. Remember, this tool is focussed at writers, most of whom need never worry about fonts, line spacing, and so on. However it does export fully valid RTF footnotes, which can then be formatted using a proper word processor. That Pages cannot read valid RTF footnotes is not exactly Scrivener’s fault. File a bug report with Apple. Contrary to your experience, there are a lot of people who use Scrivener for academic, legal, and scientific use.

As for Word styles, no. Again it is an Apple limitation. Scrivener relies on Apple’s exporters for most things. These exporters haven’t been significantly updated since Tiger came out. The alternative would be coding an RTF engine from the ground up, trying to keep up with Microsoft’s constant fiddling with black-boxed formats, and so on. That isn’t really a practical use of time for a single developer.

Some people must have their styles though, from start to finish, if you need styles and simply cannot put up with Word’s auto-style to get 85% of the work done for you, then using a drafting tool in general is probably not the best move.