FediForumTM

Event Listing

Project Map

@fediforum@mastodon.social
@fediforum.org
FediForum April 2026:

Governing Together: Toward Infrastructure for Community-Run Social Media + Sneak Peek Demo of Evidence-based “Nutritional Labels” Tool for Community Governance

/2026-04/session/8-e/

Topics:

Convener: Mayel (@mayel@bonfire.cafe)

Participants who chose to record their names here:

Paper online: https://dl.acm.org/doi/10.1145/3772318.3790855

Paper TL;DR:

  • Looks at inter-community friction that comes up due to different rules, norms. They did a workshop with various community organizers to understand more about the organizing process. There was a recorded session and followup interviews w/ the participants of the workshop.
  • Some findings:
    • Governance decisions are not always visible to people in the community;
    • There are trust bubbles - people have different people they trust when they make decisions.
    • Some tools could be created to help with this.

Notes

  • Question: Have you come across server admins that were using democratic voting tools (Polis (https://pol.is/home), etc) to decide their policies?

    • We talked w/ Rudy from BlackSky and Mike Masnick about Polis, used for decision making w/ BlackSky

    • Noticed that most Mastodon servers are an admin and then members - not a lot of opportunity or usage of democratic tools for decision-making.

    • Other tools:

    • Social.coop uses loomio for democratic decision making

    • Hachyderm via Nivenly does do voting on topics, but does not (iirc) have a dedicated too like polis or loomio.

  • Question: trust bubbles: what causes them to form? how do admins get together?

    • via affinity, often organicly, admins borrow from each other, example: HCI dot social “what are like minded communities, then copy and pasted”

    • Trust bubbles emerge from shared blocklists also. This is one aspect.

    • Are trust bubbles like Archipelagos?

      • Not necessarily. Archipelagos are built from opt-in federation. There weren’t people from archipelagos in the research paper.

    • Trust bubbles are also dicsussion groups for admins.

    • Admins may be part of multiple of these bubbles

  • Nathan Schneider’s work at University of Colorado Boulder is the “exit to community” framework and platform cooperativism.

  • Jaz: There’s conversation about expanding the nodeinfo file, and creating some standards about what goes into this. Nodeinfo is a machine-readable form of information about a server that can be ingested into moderation tooling. Adding some of the “nutrition information” from crowcards to this could be very useful.

    • Unnamed: Yes, we are doing this! Working to create a JSON description of the information from crowcards.

    • In the ActivityPub Trust & Safety taskforce, we have this issue for doing standardisation of federation of rules: https://github.com/swicg/activitypub-trust-and-safety/issues/2

      • We also have the issue for having a moderation group actor and providing a relationship between actors and their moderators. This enables sending reports to the right place and discovering data about moderation, and even tackling the ability for moderators to communicate between each other (e.g., notes on reports can be federated.

  • Laurens: In the decentralized space, its smometimes hard to make decisions - coordinating between people and getting them onboard is harder than the technical issues sometimes.

    • Can we use Crowcards to help facilitate these conversations? Crowcards fields can encourage conversations about policies, governance, etc.

  • Mayel: A part of the goal for Crowcards is to record the answers that people provides, and share them w/ people as examples for community organizers. “Fork and adapt” the example governnace structures, rather than having t come up with this form scratch.

  • Were there interviews or resarch w/ end-users? How are they impacted by governance? Do they think about it?

    • There is more work to be done here. They’ve found that end-users don’t have a lot of awareness, especially outside “geek” communities.

  • Ivan shows Bonfire has implemented some of the crowcards functionality in groups (rules on the side).

  • One issue is that ActivityPub only has Actors. Servers are not actors, so there’s work in the ActivityPub T&S taskforce towards creating distributed moderation actors

    • This is also relevant for Bonfire’s group moderation.

  • Paul: I’m part of an association. In our org, a lot of moderation time goes into defederation decisions, there’s a lot of discussions about this.

    • Unnamed: Moderators report that a lot of time goes into moderation. There doesn’t seem to be an efficient way to solve this.

    • A quote from the interviews: “It wold be so cool if you could see the moderation history of the party that we are here to judge” This is not always desirable, though.

More Resources:

Selected posts from chat

Jaz:
Strongly recommend you follow Laurens’ newsletter
Unnamed:
Blacksky have modified Polis to now have login with their authorization server so only their community can vote
Unnamed:
It was super interesting to read the voting and discussions around AI usage on polis recently
Mayel:
we’re also excited about integration with tools like PolicyKit to automatically apply democratic decisions (eg. a server block)
Unnamed:
Another new aligned project from Louis Barclay: https://modgov.pages.dev/
Unnamed:
more about the island netwokrs https://writer.oliphant.social/oliphant/islands-an-opt-in-federated-network
Unnamed:
This was another resource that came up in a similar discussion on Tuesday: https://metagov.org/projects
Unnamed:
Just thinking through those categories seems so valuable!
Anca:
Interesting comparison & crossover w/ “Stefan’s tool” for bot detection and new account moderation
Jaz:
I just want to point out https://about.iftas.org/trust-safety-services/iftas-community-library/ has example rules for all defined harms :(As in defined by DTSP)
we’d also like more example COCs for https://about.iftas.org/library/example-codes-of-conduct/
Zentropi can do that labelling based on those policies btw
Unnamed:
So can Coop from Roost
Jaz:
And here’s the beginnings of a conversation around server reputation/nodeinfo: https://stefanbohacek.online/@stefan/116476612796168994
Mayel:
yeah it’s also about protect moderators mental health and burnout
Laurens:
building a new social protocol where we communicate via updating nodeinfo
Paul:
https://fediverse.foundation/en/instanzen/
Jaz:
https://about.iftas.org/library/why-should-i-have-a-federation-policy/
Jaz:
shared signals inside trust bubbles greatly diminishes individual moderator conflict resolution issues
Mayel:
because you can then do “friend of a friend” with communities