Treating citation numbers as objects The following item appears in the $\LaTeX$ style-guide for the Journal of Integer Sequences:

To me this seems correct and reasonable, but it in most articles I read, authors tend toward the first example of "wrong" usage. Is this simply journal-specific, or is it abusive to use "In [1]..."?
 A: I don't think this kind of thing is "abusive." I'm certain I've done it in my own writing before, and I can think of several scenarios where it's better than the alternative:
(1) If a citation has no author, like a collaboratively edited industry standard for some software.
(2) If a citation has many authors and it's awkward to write "Smith, et. al. proved ... [1]"
(3) If a citation is an edited volume and it's unclear who contributed the thing you need to cite.
Generally, I think the guiding principle should be to do whatever will make your writing easier for the reader. It seems to me that there are many times when the reader just needs to know the result and which citation they should look it up in, rather than who was the first author on the paper where that result appears. Much more important to me than the question of whether I write "In [1], the authors prove..." instead of "So and so proved ... [1]" is that the citation contain a more precise reference, like Theorem 5.3.2 instead of just citing a 1000 page book and leaving the reader to go hunting for the result.
