You beat me to it. One of those brilliant “obvious in hindsight” articles. Quote it all the time.
Not exactly a fan of bluesky's seemingly centralized state, weak network effects, and us just 'reliving' all of the development that we went thru with Twitter for over a decade as they 'build in the open'.... But there is some developments with the AT Protocol that have kept my mind open about the service. The way they are pushing towards domain-based identity ownership and also the fact that running your own PDS server is an option even tho no one is doing it yet. There aren't any other 'services' to migrate to as of yet, but if need be, there could be, and you'd be able to take your data with it. I suppose there are still the underlying federation problems 'at the whim of sysadmins' maybe? but who knows. Basing your main points on a Doctorow article from 2023 might not be sufficient here as so much has gone on with the masses. Again, still all being developed. The migrations happen out of perceived necessity for the most part when things got really bad and some momentum building, but I think there's also something to be said (as is often these days in the discussions on blogging) that social media should be treated as potentially ephemeral and you should approach it with a POSSE attitude, posting on your own site and syndicating / sharing on networks elsewhere.
> the fact that running your own PDS server is an option even tho no one is doing it yet
...and until considerable part of the network does, it can just stop being an option at any given time with very little consequence.
This seems true of any place where reputation matters. If it takes time and effort to build reputation then there will be a cost in leaving. Unfortunately, places where reputation doesn't matter tend to be absolute cesspools with little redeeming value.
Blogs on your own domain is the answer.
I don't think this is a real issue. Online services are temporary. Eventually they will evolve in a direction that makes them unsuitable for your purposes, if the developers don't abandon them first. But meanwhile, we may get a few years of value.
Maybe LSE should break the cycle by running their own relay and appview.