Youre setting the bar too high. RSS support in "microblogging" apps (including Facebook, ideally) gets you a simple federated social network. Discoverability could be handled by any number of third party aggregators, like Google. Publish your email to receive "DMs".

Something can be decentralized without bleeding edge p2p crypto tech.

The article criticizes Facebook because it "vacuums up information about our tastes, our shopping habits, our political beliefs, and just about anything else you might think of—is too powerful in the first place."

So what good does it do to push discoverability to the company that is currently vacuuming up all your searches and browsing history?

edit: typo

Mastodon, which is decentralised, looks like it's gaining traction, and I find it very easy to use.

Ok.

So what will happen that drives the adoption to scale? In other words, what is the Mastodon's decentralized equivalent of investors throwing tons of money into marketing a centralized Facebook competitor? (I'm wrongly assuming that centralized means proprietary here, but I already made that wrong assumption above so I'll stick to it for the moment...)

Probably nothing except more people using it, it doesn't have any (to my knowledge) unique features that Twitter and co. don't, so I don't think people are going to move there; rather I expect the same users of Twitter to go there and keep both their accounts; a few people I know have deleted their Twitter accounts to switch full time to Mastodon, with varying degrees of success.

So I'll echo what was said elsewhere: people aren't going to move and abandon their old social networks, at least in my opinion.

If we're going to build things without any likelihood of scaling them to replace centralized services, we'd be much better off spending the time on the UX side rather than the business logic side.

I'd much rather play around with some slick, futuristic-looking UI with stop-gap business logic behind it just to garner an idea of what would be possible if we had better decentralized platforms. How about a nice simple interface to choose N trusted friends to help me archive my posts? How about simulating a ransomware attack, then leveraging M/N friends to set up the account on a new device and having the system estimate how many posts the ransomware could access? Actually how about one that replaces passwords with M/N friends?

I'm probably being unnecessarily conservative in my ideas because I'm still stuck on our current technological limitations. The point is that if all the "alternatives" developers had been doing stuff like that instead of ineffectual Facebook/Twitter clones, there would probably be a lot more value in all that effort since the Snowden leaks.

I don't like it too much, as most people will just sign up for a public server and be be stuck there.

But, even if stuck there, they can subscribe to any user on any other GNUSocial compatible server.