I like the idea of GNU Social, but the Twitter-like frontend, Qvitter, is slow and feels clunky. And last time I checked, it still hadn't solved its "spooling problem": offline nodes won't receive updates and won't catch up when they're back online. A post you delete from your timeline might end up still visible on a node if it was offline when you deleted it.
I'm sort of tempted to try writing a UUCP- or NNTP-based GNU Social clone now.
I understand that Twitter is an impressive piece of engineering given the scale and instantaneousness at which it operates. On the other hand, it does kind of boggle my mind that microblogging should be as difficult as it is.
UUCP or NNTP are really interesting suggestions. They are protocols of the old net that seemed to scale very well and long ago solved many of the kinds of issues I'd expect to see with something like a federated Twitter clone. Maybe I'm overestimating the scale of newsgroups during their heyday?
Keep in mind that the Old Net, particularly in its heyday (pre-1993) was small. A few thousand nodes. A few million users perhaps. Usenet at the time was ~50-500k users according to a guestimate from Gene Spafford. Microsoft conducted measurements in the early 2000s suggesting a few millions of users IIRC. Marc A. Smith and others:
Wow, cascade of downvotes. We're talking about whether or not Twitter constitutes enough of a public benefit to consider whether it should be a non-profit. I'm genuinely interested to hear why the idea that it should also be an open platform is also controversial!
