Post #2346347

2026-05-07 11:40 UTC

@uriel@x.keinpfusch.net @strypey@mastodon.nzoss.nz Yes, I can imagine. At the start of my career I worked in a large telecom project to bring phone number portability to the Netherlands, and indeed everything was meticulously specified up to the very last bits and bytes. The biggest problem in AP is not these differences of opinion, but goes to the very heart of the design leaving holes for "misconceptions" to creep in, to use the words of Rich Hickey, which are the most costly to deal with if not addressed. Great, recommended vid: https://www.youtube.com/watch?v=f84n5oFoZBc

Replies (1)

@uriel@keinpfusch.net 2026-05-07 11:44

@smallcircles@social.coop @strypey@mastodon.nzoss.nz The biggest problem in AP is not these differences of opinion, Yes, and this is precisely why I am always a bit worried about sounding offensive in these discussions. I am not really criticizing the people involved, or claiming that one implementation is “wrong” and another is “right”. What I am trying to say is that situations like this ideally should not happen at all if the protocol description were precise enough — for example, in the same spirit that a .proto definition in gRPC describes interfaces, semantics, and expectations with very little ambiguity. Because if the goal is to create a federated protocol that can be adopted broadly, across very different implementations and even very different use cases, then interoperability issues should ideally emerge from explicit extensions or deliberate incompatibilities, not from uncertainty around interpretation itself. So to me it feels less like a technical criticism of ActivityPub specifically, and more like a difference in protocol design mindset.

Open ##2835036