https://about.iftas.org/2026/10/15/2026-social-web-operations-report/
This year’s report reflects responses from over 400 ActivityPub communities hosting about a third of all known ActivityPub accounts, as well as responses from ATProto operators and legacy social platforms. 58 distinct platforms participating.
While operator domain expertise is higher than ever, the overall network continues to grow while volunteer capacity is stretched thin and confronting rapidly-escalating agentic spam, state-sponsored influence operations, and fast-moving global Internet regulations.
Direct PDF link: https://about.iftas.org/wp-content/uploads/2026/10/2026-IFTAS-Social-Web-Operations-and-Needs-Assessment-Report.pdf
Reading the capacity findings, the gap I’d flag is structural rather than tooling-related: ActivityPub has no standard way to carry the signals moderation actually runs on. Block exists as an actor-level activity, but there’s no vocabulary for instance-level domain blocks, allowlists or spam signatures, so each implementation ships its own representation. Mastodon’s CSV domain-block import/export is the closest thing to an interchange format, and every team re-curates the same lists by hand. With 400-plus communities hosting about a third of known ActivityPub accounts, that duplication is a fixed cost borne hardest by the small instances the report describes as stretched thinnest.
So the question I’d put to the group: is anyone working on a vendor-neutral, signed and revocable format for distributing block and allow signals that an instance could subscribe to and audit? And if so, how would it handle a false-positive appeal when the publisher and the affected server are different parties? That second part is where I’d expect most proposals to stall, but it’s also where the load reduction would actually come from.
It sounds like FIRES does some of what you want.