My take is that if we look at a longer view, right now messaging apps are fighting for marketshare by competing on features. This is a really useful fight to have - what do people want in messaging apps? GIF keyboards? Stickers? E2E encryption? Photo filters? Video messages?

A high churn rate is bad for federation - federation requires a spec, and specs are like unit tests. You don't want to write them in the experimentation phase before you know what you're building.

Over time the churn will slow down (this is happening already) and we'll probably see a standardish set of decent messaging features that everyone wants. (Good notifications, emoji, photos, low battery consumption, e2e encryption and obviously a gif keyboard.)

When we know what we want it shouldn't be too hard to write a spec and start drumming up interest in some decent opensource federated servers and clients. Signal might be a great place to start - but I think we should wait another year or two. Even signal isn't stable enough yet.

(And no, this hasn't happened yet with XMPP. XMPP has, for a variety of reasons completely failed. We need to let it die and start from scratch with something actually designed with mobile devices in mind)