I'm not as knowledgeable about federation and decentralized systems, but I understand one of the biggest drawbacks is keeping everyone running on the same version. Quoting from the article you linked

> to elaborate his thoughts on why he feels he “no longer believes that it is possible to build a competitive federated messenger at all.”

I think competitive is the key word here. Signal has moved fast in the last year, especially in the last few months. I can't imagine that happening on a federated system.

The other point is it isn't as trivial to join a federated system or create your own node. Signal? Download app and you're good to go. Matrix? Well who has a node? Where are we going? How do I set one up?

Personally I'll be happy with a bridge (like Matrix wants) when Signal is fully featured (I'd argue it is almost there. Just a few things like usernames, video scaling, channels, etc). But I couldn't imagine the mass adoption and moving at the speed they did (are) with a federated system. Even email didn't take off until Yahoo and Hotmail. Pretty much every federated system people are mentioning here didn't gain mass adoption until there was a large provider. Which has gone to such a point that rolling your own email now is difficult and you'll have a harder time avoiding all the spam. So even these examples have moved more towards centralization, if not at least a hybrid.

So what am I missing here? What's the big killer feature of federation that makes things better and proven?

Thanks for the heads up.

I just read both (Moxie and Matrix.org) and I will stay with Matrix response.

I understand Moxie points and the main one is about speed, as I understood. But, as Matthew Hodgson pointed out: the governance of decentralisation is hard, but not impossible.

So, in the end, Signal moves are driven by the market. And, IMO, sometimes it can be dangerous.