Conversation

@noracodes @r14c @futurebird

To be honest, I'd like to see more totally-independent AP microblogging codebases, but I'm not sure if implementing one is genuinely an unavoidable hassle or what, because even the small ones are too big for one person to maintain?

Like, a tiny reference implementation with barely any features is the best thing for any protocol to have, because once you have that there's a pressure to keep the protocol simple enough that the reference implementation stays small, while at the same time, the small reference implementation can be easily forked or ported by anybody who doesn't like the behavior of a major implementation.

Mastodon is the opposite of a good candidate for this (and actually, this problem also affects SSB): it contains both frontend and backend code (whereas what you really want is just the server portion and let somebody else build a client -- frontends are always full of fiddly bullshit that needs to be tuned, and users should have control over their frontends anyhow), and it's full of features somebody requested that maybe somebody else doesn't want.

Often the cost of forking is just downstream of protocol design; there are a billion gopher clients because anybody can build a gopher client in an afternoon, and there are exactly two full-featured web browsers (firefox and various chrome skins) because a small group could not write a modern web browser in a lifetime.