The spirit-of-the-nineties approach to recreating mastodon would probably look like one or two dominant native apps and something like a newsgroup server. Or maybe something like a really great RSS reader.
Questions about federation vs a walled controlled servive would be silly, because technologists are fighting it out to sell the best app. And of course a walled-in service is stupid because that would be as obsolete as Compuserve.
We’re all fighting it out on the web now. Well, except possibly on iOS and (grudgingly) Android. In some ways I think the web was how we geeks made survival on Linux possible, as we never had the mass to demand desktop apps.
Part of what I liked about the reimagining of email as social media was that it allowed for the return of the app. In the case of Mastodon, the protocols are presumably open and don’t preclude interaction via apps. So here the question of web or heavy client is not technical but one of point of view or preference.
This is all to say I kind of miss when we were chasing “the killer app” as opposed to competing for service share. When the focus was on the user’s desktop, any associated service was more or less seen as boring plumbing—-certainly nothing to have brand affinity for.
EDIT: Case in point, there’s another comment explaining that Mastodon doesn’t let you move instances and retain followers. In the 90s app-centric view of the world this is a non-issue, because following other people is the client’s job, for better or worse. Your computer crashes, you lose who you were following. If you want to publish who you’re following that’s you or your client’s job.
Giving up your social network graph to the service provider isn’t an issue in the heavy client model, because the 90s service isn’t going to pay to store that information on your behalf anyway. It wasn’t because the service viewed that personal information as “oily rags”; it was because they simply didn’t want to expend the resources to store it.