A few particular examples: I follow physicist John Carlos Baez and mathematician Terry Tao who both have one or more academic blogs for various topics which they POSSE work to several social silos including Google+ and Twitter. While they get some high quality response to posts natively, some of their conversations are forked/fragmented to those other silos. It would be far more useful if they were using webementions (and Brid.gy) so that all of that conversation was being aggregated to their original posts. If they supported webmentions directly, I suspect that some of their collaborators would post their responses on their own sites and send them after publication as comments. (This also helps to protect primacy and the integrity of the original responses as the receiving site could moderate them out of existence, delete them outright, or even modify them!)
While it’s pretty common for researchers to self-publish (sometimes known as academic samizdat) their work on their own site and then cross-publish to a pre-print server (like arXiv.org), prior to publishing in a (preferrably) major journal. There’s really no reason they shouldn’t just use their own personal websites, or online research journals like yours, to publish their work and then use that to collect direct comments, responses, and replies to it. Except possibly where research requires hosting uber-massive data sets which may be bandwidth limiting (or highly expensive) at the moment, there’s no reason why researchers shouldn’t self-host (and thereby own) all of their work.
Instead of publishing to major journals, which are all generally moving to an online subscription/readership model anyway, they might publish to topic specific hubs (akin to pre-print servers or major publishers’ websites). This could be done in much the same way many Indieweb users publish articles/links to IndieWeb News: they publish the piece on their own site and then syndicate it to the hub by webmention using the hub’s endpoint. The hub becomes a central repository of the link to the original as well as making it easier to subscribe to updates via email, RSS, or other means for hundreds or even thousands of researchers in the given area. Additional functionality could be built into these to support popularity measures as well to help filter some of the content on a weekly or monthly basis, which is essentially what many publishers are doing now.
In the end, citation metrics could be measured directly on the author’s original page by the number of incoming webmetions they’ve received on it as others referencing them would be linking to them and therefore sending webmentions. (PLOS|One does something kind of like this by showing related tweets which mention particular papers now: here’s an example.)
Naturally there is some fragility in some of this and protective archive measures should be taken to preserve sites beyond the authors lives, but much of this could be done by institutional repositories like University libraries which do much of this type of work already.
I’ve been meaning to write up a much longer post about how to use some of these types of technologies to completely revamp academic publishing, perhaps I should finish doing that soon? Hopefully the above will give you a little bit of an idea of what could be done.