My issue with simpler citations like [13] is that it is hard to figure out whether two citations are to the same book. "Was [13] that reference to the horrible tome of complete mathematics, or was that [14]" "Was [15] the paper this is trying to generalize?". By creating a small amount of meaning, it becomes much easier to remember the actual reference.
My issue with longer citations, is that I generally don't care about the source of a statement. When I am trying to follow an argument, I will just take the statement as fact, and focus on the big picture. Only when the fact is surprising, or when I have gotten the big picture, will I care about the reference. By having full inline citations in the argument, the text of the argument gets very bloated. Making it much harder to get the high level point of the argument.
I can imagine this is a feature of very technical, close to mathematical proofs, kind of texts. For other fields, maybe this matter much less. Especially when 'truth' is a lot less objective, knowing the source of a statement becomes more important I guess.
That said, I do miss the fact that the form of footnotes on the printed page doesn't translate well to ebooks (except for PDFs on tablets) and HTML. I used to write IT industry research and our general style was fairly conversational. We tended to use footnotes a fair bit for things like:
- Bulletproofing. ("Yes, we know there some exceptions to the statement we just made like A, B, C but their edge cases that aren't really relevant to our central argument.)
- Short historical digressions that would drop the reader out of the flow.
- Short background that most readers probably know but just in case someone doesn't
And so forth. I do still feel like I've had a stylistic tool taken away given that I have to increasingly write knowing that I can't depend on having easy to scan footnotes at the bottom of the page and clicking through is way too obtrusive for the manner I usually use footnotes.
Either just name the damned work already in the text ("as Smith writes in Academic Observations (1972) ..."), or use a numbered note and cite the work fully or partially. Any decent bibliographic tool (LaTeX + BibTex) can handle long vs. short cites, and your reader will thank you. For any sufficient volume of references, footnotes are the only sensible tool.
There's also the silent notes form (unsure of the proper name) common in serious-but-nonacademic works --- no indictation in the main text but page-referenced notes with anchoring context. This solves one problem (interrupting the main text), but introduces another (failing to sufficiently notice additional references or comments).