Gotta say, the ability to rapidly deal with fuckwits on the Fediverse is supercool.

It occurs to me that possibly part of the reason security controls weren't added to protocols is because of the decentralised nature of so much of the early networks.

It was harder and slower to deliberately attack 100 separate BBS systems back then. And the sysops would take personal responsibility for their sites, and the users tended to know them.

Sysops were very trusted, and had a lot of power over their few people, but also, ironically, had less unaccountable power because they were less insulated from the people they looked after.

There was still hella drama, of course, but it was more contained, like when we lose instances here because of *stuff*, the rest of things carries on.

This drive for efficiency in lots of places, and centralisation is often more efficient, is a local optimisation that is not globally optimal, IMHO.
Partitioning is important for failure isolation, so the whole system isn't killed when one part gets sick.


None of this is really new, but I don't think the incentives for the global power consolidators are aligned with systemic health.

Sign in to participate in the conversation is an Australian instance of Mastodon run by @daedalus.