The combined online persona of those accounts is only as strong as their combined security. aka: Why would services need to collude when they can get the job done by ineptitude?
I agree, a lot of services displayed a shocking amount of incompetence in that post. However,
a) The more proofs you have, the harder it becomes to force them. YC for example is one location, and is run (in my opinion) by very smart people where it would be hard to get a compromise.
b) My point is that this is an excellent alternative to not using anything in a way that is both friendly to people ("just make this post on [website]") and compatible with an older, better method of privacy (PGP) that people have been using for years.
It may not be as perfect as some of the more esoteric alternatives that people have suggested elsewhere in the thread (I'm not sure about this, can an incompetent phone company employee compromise some of the phone-based ones? I've come across a lot of incompetent phone company people), but much easier for the regular person to use.
This is a crazy story but I still think Keybase gives you a lot to defend here. You have to compromise all the accounts and change all the proves to actually be able to send valid messages to somebody else.
That is a tall order, even if you use the same email as a username everywhere. I use long random passwords and 2Fa on a number of the important accounts. I don't trust google and Facebook, but I trust them to have some interest in not letting accounts be compromised.
Also if somebody changes all the proves, they will all be new and a smart system should be able to detect this sort of stuff in the future.
https://medium.com/@N/how-i-lost-my-50-000-twitter-username-...