fedi dev meta 

i believe in decentralization because i am a decentralized systems nerd. it is literally my profession. i recognize the tech has a long way to go with regards to fostering safe communities, but i'm hopeful because a lot of people are taking it seriously. the folks with SSB and Cabal in particular have begun the long journey of architecting for effective moderation. Mastodon, in this regard... has stagnated. tooling is only a part of the solution, of course, but i'm hopeful because there are people taking it seriously.

decentralization is not enough, but it is a vital foundation.

Follow

re: fedi dev meta, moderation 

my goal in moderation tech is *information privacy* and *distributed immunity*.

information privacy so that when you block someone, they can't access you anymore. they can't see you, the server rejects their interactions, etc. proactively, a server could prevent strangers from seeing you without some pre-existing point of contact, like a mutual follow or a thread interaction.

distributed immunity means when your friends take moderation action, you can see it and act on it -- like , but outside of the timeline. this ensures that known vectors of abuse become known, and that discussion of vectors retains discretion. in particular i'm evaluating a web of trust architecture that allows moderation actions to propagate, so that you can automatically and manually build up allow- and block-lists, and retain documentation for each action.

it's important that this effort produce standards and not only tools or applications. decentralization means we aren't hostage to corporate opinions, or even to particular SDKs, but can distribute and iterate on best practices and robust implementations as determined by our communities. SSB, Cabal, ActivityPub -- all of them can benefit from a moderation overlay. we don't need to wait on BDFLs to get it to protect and support ourselves and each other.

re: fedi dev meta, moderation 

tooling is only part of the solution but as a toolmaker i feel it's important i contribute my abilities to the effort. other parts include: community organizing, and a commitment by individuals to unlearn their own weaponization, ex: antiblackness. anticapitalists must hold dear the destruction of whiteness, for it is nothing but a weapon upon which capitalism depends.

i'm proud that my corner of fedi (and isn't that the magic of this place-of-places?) has rebuffed a number of fascist encroachments, and organized considerable resources for mutual aid, but we have also failed to spread immunity for known vectors and to effectively de-escalate and resolve interpersonal conflicts, which have instead had a tendency to escalate. this isn't strictly a software problem, but in digital spaces that are literally made possible by software, better software can help us do better.

Show thread

re: fedi dev meta, moderation 

@garbados I know some of us in the community are tossing ideas around what a more privacy-aware and moderator-friendly architecture would look like in ActivityPub.

There's some demos floating around but it'll take time to build community consensus and get a real solution in place.

re: fedi dev meta, moderation 

@cj yep, that's how it is with software. i'm just glad folks are taking it seriously, y'know? it's hard work and i'm glad we're doing it.

re: fedi dev meta, moderation 

@garbados Do you know maymay's work, i. e. the various Predator Alert Tools?

re: fedi dev meta, moderation 

@adrianheine i've heard of it but i'm reading up more now. good stuff, thanks for the recc :)

re: fedi dev meta, moderation 

@garbados @mhoye If you two are familiar with one another's work, then I apologize for cluttering your timelines. But if not, I think there might be some useful collaboration to be done in this field.

re: fedi dev meta, moderation 

@dubious_dragon @garbados Thanks for the introduction, K.

G, your point about distributed immunity is really important. Zero-interaction shared inoculation is one reason we went with Matrix: auto-updating shared blocklists. I talk a bit about it here, about how it offers federated systems the possibility of sharing verbs, not just nouns, and what that means when we're talking about federating user agency and safety. matrix.org/open-tech-will-save

re: fedi dev meta, moderation 

@mhoye @dubious_dragon this is great. i'm looking forward to reading this :D

re: fedi dev meta, moderation 

@garbados This is some pretty good food for thought. It seems this day doesn´t end completely without interesting input.

re: fedi dev meta, moderation 

@garbados note about block lists: please please please consider not propagating blocks. instead, propagate reports.

re: fedi dev meta, moderation 

@trwnh could you say more about the importance of the distinction? why not share both?

re: fedi dev meta, moderation 

@garbados blocks are personal, hold no accountability, and their propagation means that people at the root have wildly outsized influence. see: twitter + randi harper -> wil wheaton.

reports on the other hand have a reason attached, are only created when there is a reason to, and can simply be forwarded to a wider audience than they already are.

re: fedi dev meta, moderation 

@trwnh i think this is an architectural question. when i receive a hypothetical propagated block, for example, i see "[user] blocked [user] for [report]" so you get the context, and your system may choose based on how you the user configured it to accept that action optimistically, or to defer it for user review. in both cases, the user may examine the context and act appropriately.

in a web-of-trust system, that level of configurability might look like this: for sufficiently trusted users, optimistically accept their actions and propagate them. for followed users that you haven't marked as sufficiently trusted, hold the action for review and do not propagate. as an example, a "sufficiently trusted user" might be an alt, whereby a user blocks someone on one instance and their account on another instance also performs that block.

re: fedi dev meta, moderation 

@garbados @trwnh

i have a stupid anecdote on this:

i once got blocked for simply /not knowing a fact/ that somebody was far right,
which as far as i could later find had only really been discussed on reddit, and anybody very easily could have missed

with automatically shared blocklists people like me would get blocked across the world for nothing & maybe never get unblocked

and I worry a lot about what the state of discourse would become at that point

re: fedi dev meta, moderation 

@Valenoern @trwnh i want to make the important distinction between “propagated moderation actions” and “automated blocklists”. automated blocklists do what you’re afraid of: shut people out, silently, and mismanagement can lead to disaster very easily. in the web-of-trust scenario, your client only carries out a propagated action IF you have set the source as sufficiently trusted, which IMO should only be if the source is one of your alts. in a general case case, actions would have to be manually reviewed and approved. it’s like a distributed reporting system, where the report and the action form part of a broader context that you can further examine.

re: fedi dev meta, moderation 

@garbados @Valenoern ooh, one point i was reminded of just now -- automated blocklists only came about because of the total failure of twitter's own moderation system. twitter failed its users. it's important for users to be able to trust their admins to protect them. that's why i think this system should hook into the existing mod tools, and the simplest extension is to allow forwarding Flag activies to other instances.

re: fedi dev meta, moderation 

@trwnh @Valenoern i'll have to keep that in mind. i haven't really looked into extending activitypub yet; i've been thinking more about a system-agnostic standard for that web of trust thing (and... a million other projects, of course)

Sign in to participate in the conversation
Toot.Cat

The social network of the future: No ads, no corporate surveillance, ethical design, and decentralization! Own your data with Mastodon!