This was never true. There was an XMPP-speaking endpoint into Facebook's proprietary chat system, but it wasn't a S2S XMPP implementation and never federated with anything. It was useful for using FBChat in Adium or Pidgin, but not for talking to GChat XMPP users.
[1] https://googletalk.blogspot.com/2006/01/xmpp-federation.html
Everyone has email, so you could imagine a social networking app that’s just a thin layer over your email, and every interaction is encoded as an email being sent under the hood. Want to share a picture with your friends? Send an email. Someone wants to comment on it? They just send an email. Etc.
The main purpose of the app would be to offer a nice, device responsive, consistent presentation. Additionally if this were an open, documented standard, an entire ecosystem of “email apps” could emerge.
(Of course as far as your actual email account goes you’d want to auto archive the emails + not get notifications for them, but that’s easily configurable)
For example, Facebook could create some kind of plugin API that allows you to interpose your filtering/ranking code between their content and their presentation.
For example, maybe they give you a list of N possible main page feed items each with its own ID. Your code then returns an ordered list of M <= N IDs of the things that should go into your feed. That would allow you to filter out the ones you don't want and have the most interesting stuff displayed first. Facebook could display the M items you've chosen along with ads interspersed.
Something like that could run in the browser or Facebook could even allow you to host your algorithm in a sandbox on their servers if that helps performance. (Which means you trust them to actually run it, but you have to trust them on some basic things if you're going to use their service at all.)
In other words, changing the acoustics of the echo chamber doesn't mean you need to be the one implementing a big chunk of the system. You just need a way to exert control over the part you want to customize.
The RSS feeds I subscribe to give me plenty of "presentation" or "branding". Logos, written descriptions [both short- and long-form], clear names of what I am subscribing to, URLs. Just the right amount for me, in fact; if I wanted to go to their website(s) for their particular buffet of blog posts, featured puff pieces on Page Five, twitter mentions, &c I can do that ... or not. I'm glad I don't have to if I don't want to, and all of these folks are more than able to drop into their RSS feed a "Please go here for our tour information with new stuff in our online shop" mention just as you are able to go straight to some website full of deep-thumping media flashing into your senses as you get to where you want to go instead of using RSS.
There seems to be a steady thread of this sentiment here on HN, yet over the years no one has quite cracked this nut. Solutions welcome!
There’s a lot of history around that split, and the motivation for HTML/CSS was about separating presentation from the content in many ways. For another example, once upon a time a lot of chat services ran over XMPP, and you could chat with a Facebook friend from your Google Hangouts account. Of course, both Google and Facebook stopped supporting it pretty quickly to focus on the “experience” of their own chat software.
The thing is that there is very little money to be made selling content, and a lot to be made controlling the presentation. So everyone focuses on the latter, and that’s why we live in a software world of walled gardens that work very hard to not let you see your own data.
There is some EU legislation proposal that may make things a bit better (social network interop), but given the outsized capital and power of internet companies i’m not holding my breath.