[dunham]

If the document were html, something like epubcfi might work better than regular expressions. This is what the epub people are recommending for a similar use case - sharing highlights and notes in epub books whose content may be updated.

For locating links/comments from people you don't know, I'd imagine a service like Google. Something that indexes the entire web and answers queries about which links overlay this document. I'd think they'd have to be ranked somehow. (Technically, Google already does this with "link:http://..." queries, but standard HTML links don't reference document fragments.)

[daveloyall]

Never heard of epubcfi but yes, if they came up with a way to highlight specific portions of a text, then yep, that would do it.

Regarding using google as a lookup service for something as fundamental as this.. No. It's important for the health of the Internet that we not do that. See https://tools.ietf.org/html/draft-nottingham-stakeholder-rig... . :)