* Cryptocurrencies who's sole reason to exist is to justify its own existence. "We need token X to perform operation Y. Operation Y is important to facilitate the trade of token X". Seriously, how many tokens exist for the only purpose of being "gas" for its own network?
* Social networks that only discuss itself and its own existence.
* Programming languages where the largest project written in that language is its own compiler.
Solutions in search of a problem, is what they are.
Federation is already successful, since email is federated (you can register on gmail.com and send an email to an account at yahoo.com, or even run your own email server, although the ecosystem has evolved/commercialized to make that more difficult than it strictly has to be).
But Lemmy, we'll see. Maybe a better-written competitor emerges, or maybe they get their act together, or maybe the world doesn't want a federated Reddit clone (although my personal bet is that it does and subreddits are sort of the centralized prototype of that). But there's no question that federation itself is useful in certain applications.
This is a phase that every single new platform/tech passes through. All of them.
(Yes, fediverse isn't new, but it's new to many right now...)
Apart from languages which never bother self hosting, this seems backwards. It could be considered a sign of failure of a language if it never attracts projects beyond its own compiler, but it isn't true that a project which has its own compiler as the largest project is doomed to failure. For example, this was once true of Rust and Go but no longer is.
As far as I've noticed those politics discussions boil down to different degrees of reverence for Deng Xiaoping, and if one is slightly to the right of Bernie Sanders you get banned.