[Mezzie]

Yes. We should record all of this, and turn them into easily browsable graphs/hypertext to easily assemble sets of papers to read/look into. At the very least things like 'background reading', 'further reading', 'supporting evidence' and 'addressed arguments' would be useful.

'We' meaning the librarians and archivists. You guys actually researching have more than enough to do.

[d110af5ccf]

Actually I think that's an intriguing idea for how to improve citations. Instead of a single <work> citation, have multiple <work, subset> citations that include a region of text as well as basic categorization of how the citation is being used in that instance.

I'm not sure if it would prove feasible in practice. It seems like it would aid the writing process in some cases by helping the author keep track of details. But in other cases maintaining all that metadata would become too much of a burden while writing, so it would get put off, and then it would all fall apart.

Very interesting to think about!

[Mezzie]

I was imagining a post-writing process akin to assigning a paper its DOI[0] or a book its cataloguing info. Citations as they are can be done by researchers without impacting the process because they're binary: Something either is cited or it isn't. It either contributed to the creation of the research or it didn't. This probably couldn't be done by the researchers, but you identified why: The citations are data but this would be metadata.

Definitely don't want to encourage papers to take even LONGER.