|
Top-level
 There is a paper written proposing FIRES, which was written with the financial support of @nivenly. I'd reached out and received extensive peer review from various active members of the fediverse who specialise either in technical, trust & safety, or social areas. Following their peer-review, I decided that I needed to rework the proposal to outline things clearer, which is why I've not yet published it. 7 comments
Another idea it puts forwards is the use of structured data for labeling decisions, for instance, maybe you trust a providers' recommendations related to racism but not transphobia, or maybe you want the CSAM and DNI data, but not to block bridges. This is all possible if we've more structured data. Finally, FIRES makes zero opinions on how data enters the system: whether it's through a collective voting on taking specific action, or an algorithmic sourcing of data, this is up to upstream software to decide, FIRES will just keep a record of those decisions over time. Within scope of funding may be building some of these tools, that's still to be decided. I look forwards to working with NLNet to help bring distributed moderation data to the Fediverse. Just to be absolutely clear: I'm not entering the business of curating a blocklist, I'm writing the software to help distribute moderation recommendations and advisories.  @polotek for this month, I'm mostly needing financial backing (since I've been unwell & my billable hours are down significantly) The reference implementation of the FIRES server will be written in Node.js though, so once I start building it out perhaps there's room to bring on paid collaborators? (I need NLNet's approval for this, afaik) https://support.thisismissem.social (one-time at bottom, addresses only used for invoices) |
Gregory's Server
FIRES outlines a few different things, one of those being a different model of federation management, not centered on binary decisions of "blocking or not", but instead a more firewall based approach: allow / filter / reject.
FIRES is also not focused exclusively on domain based federation management, but looks at issues that will become more pertinent as the Fediverse grows, for instance Hashtag or Link management.