One of the many problems that’s emerging with webmention is it’s turned into a sort of Swiss army knife of notifications; the IndieWeb uses it not just to send responses to folks, but also for things like publishing to Bridgy Fed or syndicating content to content aggregators. It’s the basis of how notes work. It’s up to the recipient to try to disambiguate the meaning based on context and post-type discovery, and what things are can change over time, sometimes in unpredictable ways that fall apart.
Webmention-as-reaction is fine, but it’s very difficult to get good UX with webmention-as-comments. Reply chains are difficult to follow, especially if one of the links in the chain breaks. Webmentions that come in via brid.gy can make for a bad unstructured reading experience as well. There’s also still no real implementations around webmention for private posts; even with the progress towards private posting (with two theoretical ideas) there are no readers that support this stuff, much less webmention endpoints. Any ecosystem which supports this stuff is going to need a tighter coupling between the webmention endpoint and the authentication token broker. This can certainly be engineered for, but gosh are there a lot of pieces that need to come together to work together…
The more I think about this stuff, the more I feel like all the functionality needs to start being bundled together into one ball, and then that makes it start to sound a heck of a lot like the reasons I dislike ActivityPub. If I do ever get around to implementing Subl, will it end up having to implement the webmention endpoint too? I mean, it’s meant to subscribe to publications, including push events like WebSub and Ticket Auth, and what’s webmention but yet another type of published event?
I dunno, this all just feels so overwhelming at times.
Mostly I like webmention as a mechanism for getting notified about stuff that happens externally, and prefer native comments for the actual discussions on a post. It’s safer, it keeps all of the conversation in one place, it allows for easier moderation and abuse mitigation, and it doesn’t rely on a web1 of trust that ultimately is based on the benevolence of the Internet community. Which, as we all know, is tenuous at best.
Of course the other problem with webmention is the expectation that (like all things IndieWeb) everyone have their own website on their own hosting and their own domain. There are external glue services like commentpara.de that can help fill the gap but it feels like it’s being filled with spackle instead of something that can actually bear the load. And something about that also feels like it can become yet another abuse vector, too.
So, I dunno. I think I prefer to think of webmention as an “external” thing, a nice indicator of what others around the web feel and think about my stuff, but would rather work on further developing native comments for on-site commentary. I know, it’s very silo-type thinking of me, but it’s also what makes me feel comfortable.
As an aside, my current comment system certainly has its share of issues and jank (and I’m not a fan of its client-side rendering nature either) but its design is at least such that I can use it fairly reliably. Of course, now that Publ supports arbitrary entry attachments I have plenty of ideas about how to build a comment system that’s based on native Publ functionality. It would probably scale better, too. Heck, I could also implement my own webmention endpoint (or at least a webmention.io webhook) that does something similar, and turns incoming webmentions into native comments and maybe get the best of both worlds? I dunno.