Why would we need to fork the software? It’s AGPL-licensed. Code changes and related discussions are public. And for now there’s no signs of the maintainers pushing their political opinions in the software itself, they just do that on their own instances (and it’s not like they’re hiding that, I mean their choice of TLD is a political statement in itself, do people think “ml” stands for “markup language” or something?).
Until they start messing with the code (which is something that is not going to go unnoticed), there’s absolutely no reason to fragment development efforts, that would just be counterproductive.
Well that’s our fault for letting information get congregated in a centralized service to be fair. Any information that is stored without redundancy on a single service should be considered already lost.
The Fediverse doesn’t fix this by the way, as far as I know. The data can be accessed from other instances, but as I understand it the data still lives on the instance. The day an instance does, poof, all the information it contains goes away.
But! It makes it easier to make information redundant, by having an instance that automatically archives information for example.
We had a problem, many people knew that we had a problem but we did nothing to fix it. We have the same issue on StackOverflow or even GitHub, by the way (although the latter is a bit mitigated by people having local copies of the repositories for example). It will come bite us in the arse one day.