So, one thing I said at #FediMTL yesterday is that Fediverse software should ship with the IFTAS DNI list as the default, minimum blocklist.
-
I’m gonna scope-creep this by adding: an easy API for publishing and subscribing to multiple blocklists.
Like, why are we still trading CSV files when we could just publish this info over ActivityPub?
Let’s make a tool for building/managing blocklists as an AP actor for anyone to subscribe to.
Add in a “polyfill” to subscribe and load this data straight into big apps like Mastodon, and we have instant moderation as a service.
I can start on Monday.
> @benpate@mastodon.social said:
>
> Like, why are we still trading CSV files when we could just publish this info over ActivityPub?Oh, I'm all about using ActivityPub for things but even I think serving a CSV file is perfectly fine
That said, if IFTAS were to also publish the revised lists as a Collection over ActivityPub, I would have my instances subscribe.
-
So, one thing I said at #FediMTL yesterday is that Fediverse software should ship with the IFTAS DNI list as the default, minimum blocklist. Ideally, the list should periodically auto-update.
If you *want* to see gore, racist harassment, or transphobia, you can consciously remove those sites from your blocklist. If you want to subscribe to a different blocklist, you should get to do that, too.
Fediverse admins shouldn't have to learn how to find the DNI list and install it the hard way.
Or at least, on installation, give the option for the administrators to put the banlist as strict or slacking as they wish to. I, for one, don't bother using banlists as my instance is single-tenant and I can just unfollow users I don't want to hear about. Other instances might want to preemptively block the worst actors - but having an informed choice of what means to do so on behalf of the instance's users is also important. -
> @benpate@mastodon.social said:
>
> Like, why are we still trading CSV files when we could just publish this info over ActivityPub?Oh, I'm all about using ActivityPub for things but even I think serving a CSV file is perfectly fine
That said, if IFTAS were to also publish the revised lists as a Collection over ActivityPub, I would have my instances subscribe.
Yeah, at the end of the day, CSV just a different data format. And we already have lots of tooling around it.
But with CSV, I think we’re missing out of real time push. And it’s ironic that a community built around a real time push protocol has to fall back to CSV
But I think there’s a lot of interesting things we could do with “outsourced moderation” and id love to contribute to our next-gen solution.
-
a default makes sense because this place won't grow in a decentralized way if newbies have to either figure out which already big servers protect them from assholes, or end up in a place where they are harassed and never come back. my concern is that it also won't grow if certain views aren't allowed and that determination can change.
imho the middle ground is say whatever you want like on your own blog, but no harassment. maybe you'll be ignored but you'll be widely federated.
-
I’m gonna scope-creep this by adding: an easy API for publishing and subscribing to multiple blocklists.
Like, why are we still trading CSV files when we could just publish this info over ActivityPub?
Let’s make a tool for building/managing blocklists as an AP actor for anyone to subscribe to.
Add in a “polyfill” to subscribe and load this data straight into big apps like Mastodon, and we have instant moderation as a service.
I can start on Monday.
@benpate You probably want to look at FediMod FIRES by @thisismissem before you start coding: https://fires.fedimod.org
I have it on my radar to try out, just to take a look, even though as a single-user server owner I don't feel a pressing need from a moderation perspective. If it weren't for time pressure by other projects I'd be looking into it in much more depth.
-
@benpate You probably want to look at FediMod FIRES by @thisismissem before you start coding: https://fires.fedimod.org
I have it on my radar to try out, just to take a look, even though as a single-user server owner I don't feel a pressing need from a moderation perspective. If it weren't for time pressure by other projects I'd be looking into it in much more depth.
Thanks. I'm checking this out now!
@julian@fietkau.social @thisismissem @evan @julian@activitypub.space
-
So, one thing I said at #FediMTL yesterday is that Fediverse software should ship with the IFTAS DNI list as the default, minimum blocklist. Ideally, the list should periodically auto-update.
If you *want* to see gore, racist harassment, or transphobia, you can consciously remove those sites from your blocklist. If you want to subscribe to a different blocklist, you should get to do that, too.
Fediverse admins shouldn't have to learn how to find the DNI list and install it the hard way.
@evan WordPress and BridgyFed offer it as a default option, would love to see more platforms educating their admins as to the basic why-you-might-want-a-denylist and offering recommended options.
-
IFTAS is Federated Trust And Safety. It is a non-profit dedicated to improving user safety in the Fediverse and supporting the work of Fediverse moderators. The "I" doesn't seem to stand for anything!
IFTAS
IFTAS is a non-profit initiative advancing independent, sovereign social networks, and supporting the people who uphold trust, safety, and inclusion on these decentralised platforms. We provide guidance to organisations and institutions engaging with the open social web, and we support volunteer moderators, service providers, and communities operating across the diverse and evolving landscape of federated…
IFTAS (about.iftas.org)
DNI is the "do not interact" blocklist maintained by IFTAS. It's a "worst of the worst" list that prevents interaction with, among other things, dedicated instances for racist, transphobic and homophobic trolls.
IFTAS Do Not Interact (DNI) Denylist
About this denylist The Do Not Interact List is an IFTAS-curated list of domains that are highly recommended for defederation. Each domain is labelled using a shared vocabulary label. Each domain is investigated by human review for governance, common content, network and service activities, hosting location and more. Inclusion on this list means we believe: Federating…
IFTAS (about.iftas.org)
@evan /me whispers "independent"
-
@evan WordPress and BridgyFed offer it as a default option, would love to see more platforms educating their admins as to the basic why-you-might-want-a-denylist and offering recommended options.
@evan I also believe platforms that offer a "flagship" server should make /their/ denylist available at install, after all, they know a lot more about what they're seeing on the platform than anyone else. "We block these servers, you might want to, too'
-
@evan I also believe platforms that offer a "flagship" server should make /their/ denylist available at install, after all, they know a lot more about what they're seeing on the platform than anyone else. "We block these servers, you might want to, too'
-
I’m gonna scope-creep this by adding: an easy API for publishing and subscribing to multiple blocklists.
Like, why are we still trading CSV files when we could just publish this info over ActivityPub?
Let’s make a tool for building/managing blocklists as an AP actor for anyone to subscribe to.
Add in a “polyfill” to subscribe and load this data straight into big apps like Mastodon, and we have instant moderation as a service.
I can start on Monday.
@benpate @evan let's talk about labels, I believe any nextgen ingestion should use the shared vocabulary to take in blocks by label https://about.iftas.org/library/shared-vocabulary-labels/
Maybe you want to block CSAM, but don't care about copyright infringement... (This way we could run one list instead of five and you choose the harms you want to block, as everything is labelled)
-
@evan I also believe platforms that offer a "flagship" server should make /their/ denylist available at install, after all, they know a lot more about what they're seeing on the platform than anyone else. "We block these servers, you might want to, too'
-
@renchap
oooh, I'm very interested in this -
I disagree. The only servers that should be blocked are those that fail to address their users who engage in unwanted tagging. That's harassment/abuse. It drives away users. I could support a list of servers somehow so verified.
But to go further than this is a kind of censorship of content allowed elsewhere on the web. Stuff we basically never stumble on while browsing. Without algos here, it will never be forced on us.
I can easily see posts critical of Israel called antisemitic.
@wjmaggos @evan I disagree. The only servers that should be blocked are those that the operator of the service exposing themselves to the responsibilities and liabilities of hosting user generated content wish to block. Everyone gets to make their decisions, for themselves or for their community. Not me. Not you. Them.
Federation is a privilege, not a right.
-
R relay@relay.infosec.exchange shared this topic
-
@wjmaggos @evan I disagree. The only servers that should be blocked are those that the operator of the service exposing themselves to the responsibilities and liabilities of hosting user generated content wish to block. Everyone gets to make their decisions, for themselves or for their community. Not me. Not you. Them.
Federation is a privilege, not a right.
I imagine a 90s world with a diversity of browsers that all made their own choices re what sites to render and many fewer people leaving AOL.
EDIT: let me add that this would quickly lead to sites and a few dominant browsers that weren't so strict and we'd lose some decentralization that might have been if people had tried harder to balance expressing all opinions and maybe warning about upsetting stuff instead of requiring a different browser.
IMHO
-
@benpate @evan let's talk about labels, I believe any nextgen ingestion should use the shared vocabulary to take in blocks by label https://about.iftas.org/library/shared-vocabulary-labels/
Maybe you want to block CSAM, but don't care about copyright infringement... (This way we could run one list instead of five and you choose the harms you want to block, as everything is labelled)
Yes. This is a fantastic list, and would be a good starting point for anyone publishing their own DNI list.
How do you feel about each publisher defining their own terms? I’m conflicted.
It might be nice to have something truly decentralized, like hashtags, where anything goes.
But there’s probably more utility in some set of standardized codes. It would allow admins to check off the topics they want to follow.
Or maybe there’s a middle ground that could achieve both goals?
-
I imagine a 90s world with a diversity of browsers that all made their own choices re what sites to render and many fewer people leaving AOL.
EDIT: let me add that this would quickly lead to sites and a few dominant browsers that weren't so strict and we'd lose some decentralization that might have been if people had tried harder to balance expressing all opinions and maybe warning about upsetting stuff instead of requiring a different browser.
IMHO
@wjmaggos @evan I was there for the 90s. We are not in that. We are in an interconnected world of 40,000 services sending copies of their content to be stored and hosted by their peers. Some of this content is illegal in most countries. Some of it has caused severe harm to people I've worked with as they melt down. Some of it is appropriate for a given community, some of it is not. Some of it is regulated by some of the countries some of these services are located in. Welcome to the 2020s.
-
@evan /me whispers "independent"
@iftas That's what I thought!!!
-
Yes. This is a fantastic list, and would be a good starting point for anyone publishing their own DNI list.
How do you feel about each publisher defining their own terms? I’m conflicted.
It might be nice to have something truly decentralized, like hashtags, where anything goes.
But there’s probably more utility in some set of standardized codes. It would allow admins to check off the topics they want to follow.
Or maybe there’s a middle ground that could achieve both goals?
-
So, you're saying we should use a standardized list? I'm fine with that if we're confident that all of the (realistic) reasons are covered.
So long as "pissed off the mod this one time" isn't a standardized code
