Besides Danny, I would use it! 
The point is not to use it as a front-end; that’s besides the point. Rather, many of us use DT as our note-taking database of our book-readings and research notes. We want to link those notes to their relevant bibliographical entries, so that full biblio info is only a click away, whether from DT to the bibliographic database (in this case, Boookends), or from the biblio database to the relevant folder of notes in DT.
Exporting the bibliographic info from Bookends, and importing it into DT, is possible now using “aliases”. It is manual, which nonetheless is fine; what is not fine is that once you try to reimport your biblio database again, all your aliases are gone, and you have to build them all over again. So, for example, if you add new entries into your biblio database (in Bookends for example), or modify a pre-existing bibliographic entry, you have to manually export just that/those entries and import them. Likewise if you annotate or otherwise modify the biblio entry in the DT sheets. There is no way to sync, or even to import without overwriting all one’s aliases and annotations.
But that’s only one benefit. The other major benefit?: applying DT’s amazing AI abilities on one’s bibliographical info. Many of my bibliographic entries have brief overviews of the book or article in the shape of an abstract. If I import these all with the rest of the bibliographic info in DT, DT can then “recommend” certain books and articles to me when I am looking at a note or another file in the database. That is powerful stuff: it reminds you of articles/books that are relevant to your present research!
This would also further enhance the current working relationship between Bookends and Mellel. One’s biblio entries and notes in DT, linked to the relevant entry in Bookends, which can then immediately place the relevant biblio information for the footnote to that note that one has copy and pasted into Mellel from one’s “notebook” in Devonthink. No one program has to do everything; that is bloat. For the programs to work together well would be most fruitful.
As to why Bookends, and not any other bib program, I offer that top-notch mac programmers in complementary types of software should work together whenever possible, support one another, and help develop the Mac platform as the most integrated OS. By taking advantage of underlying OS X technologies, and then working together somewhat, amazing collaborations could take place, all of which would (a) help the end-users like us, and (b) help each piece of software to find new users/buyers. When programs are closely integrated (as Bookends and Mellel currently are), it leads users of one program to often purchasing the other. This leads to further development, and the cycle continues, to the benefit of all.
