As the Fediverse grows, the amount and frequency of harmful spam also increases. The feudalistic structure of the network creates notable moderation overhead as every instance admin has to ban offending remote accounts and remove their content themselves, if the originating instance does not act quick enough or such delete actions are not federated properly.
An example (and my motivation for this post) are reports from the last days about accounts which post CSAM content. This is especially problematic, as media content gets duplicated to every receiving instance.
The following idea came to my mind: A software which sits on top of Activity Pub and creates a kind of court. Instances of this software are independent courts. All users can elect admins / mods of their home instance as judge. Everyone can post a case via normal Fedi posts to the court and the elected judges give a verdict via majority vote (votes are also normal Fedi comments with special formatted content). Once a case has a verdict, all instances should immediately apply the decided action against the offending content / account. This can be done via a bot which is set as an instance moderator and follows the published verdicts of a court instance the local admin trusts.
This way swift action can be applied through all instances which follow the court at once. By having periodic elections it is kept democratic.
And the task of the court is to defend universal rules all following instances agree on. Not to handle politics between instances.
What are you thoughts on this matter? Or do other approaches to solve this issue already exist?
I’m not sure how a process modelled off of a tribunal is supposed to act as a fix to mods or admins not acting quickly enough. Trials and tribunals are notoriously slow, because they are called when correctness is more important than speed.
By design, these are not vehicles for “swift actions”.
If a certain website is known to have a problem with their moderation, then that’s what server block lists are for. Communal server block lists already exist. Having a standardized format and the ability to subscribe to one would be nice, but as of right now the amount of effort to update a server’s block list with a new version is relatively simple.
The thing is, as the fediverse grows, and especially as the group-based part of the fediverse grows, we have to accept network splits as a natural part of that process. It is going to get patchy here. It’s a natural part of social network development. What we need to get ahead of is showing people that that is OK, and that frictionless access to the whole of the social web isn’t necessarily a good thing.
I recently learned about discourses trust levels:
https://blog.discourse.org/2018/06/understanding-discourse-trust-levels/
They’re pretty smart.
Overall, my utopian vision for the fediverse is local servers per interest, with members who willingly participate because they value the community, including paying for server costs. These local communities then moderate according to their rules and servers block each other based on preferences.
In that utopian model, spam is not an issue, because the members are locally known, you don’t even really get in without some kind of invite or knowing someone.
I would be against some overarching court that has the potential to make decisions I disagree with. The freedom of the internet is precisely in not being governed by something like that. Including communities around “wrong opinions” and “disgusting content”.
Everyone can post a case via normal Fedi posts to the court and the elected judges give a verdict via majority vote. Once a case has a verdict, all instances should immediately apply the decided action against the offending content / account.
My immediate thought reading this was, how does that work across jurisdictions? Is the idea that every jurisdiction runs its own instance of this software? Something illegal in one jurisdiction might not be in another.
There’s also the question of representation. Does every (lemmy/mbin/piefed) instance get the same representation? If there’s proportional representation, that just promotes centralization of the fediverse. If there’s not, you can just set up new instances to manipulate the vote.
And the task of the court is to defend universal rules all following instances agree on.
Sure, but are there checks and balances which ensure that?
I administer my own instance of Lemmy and Mastodon. The CSAM issue was something I learned about real quick, unfortunately. I was able to implement community maintained blocklists. It’s not perfect, by any means, but I’ve not seen any more harmful content.
The idea of a centralized judicial system does not seem compatible with the spirit of federation. The safety we expect from centralized platforms is something we gave up for our freedom. With that, we do have to be diligent and deal with the bad actors. It’s the worst trade off in the whole deal.
It’s also impossible to enforce, I think. The Rage Against the Machine “f you, I won’t do what you tell me” spirit lives in all small server admins (by virtue of rebelling in the first place).
I’m not sure about centralization but delegation would probably make more sense. So I could delegate my own instances moderation to a instance I trust. This way we still can have seperate ways of moderating so the Tankies can moderate their stuff, the Nazis their stuff, Liberals theirs, etc. pp.
This is especially problematic as media […] gets duplicated to every receiving instance.
I’m not sure this is actually the case. The posts certainly do get duplicated to every instance, of course, but the media contained in them are always hosted by the originating instance, as far as I can tell. So once it is deleted from there, no one can access it anymore.
I think it’s an interesting idea though. Since you mentioned votes by members of the individual instances, I wonder how you would handle the different sizes of instances. I’m the only user of my instance, for example. I probably wouldn’t be able to be judge then, right?
From what I’d seen in the announcements of lemm.ee when it was alive (RIP), one concern by the administrator was if the instance should keep local copies of files for as long as possible, as a timed cache, or link to the original instance. So taking it as an example, the technology for local copies does seem to exist on Lemmy at least.
The posts certainly do get duplicated to every instance, of course, but the media contained in them are always hosted by the originating instance, as far as I can tell.
Some software creates thumbnails on the server storage.
I wonder how you would handle the different sizes of instances.
That is a problem to discuss after it is decided if the Fediverse even wants such a service ;)
But I’m aware and one possibility would be to limit the number of electable judges from one instance. The upper limit must be in proportion to their user-count to be somewhat representing. But I think instances with only a handful of users won’t ever get seats without the jury being way too big. Or they are so popular that they get enough votes from other instance users.
I like the idea. Better yet: Don’t put it on top, but make it part of the platform itself. I think moderation is an integral part of a platform and should be baked in.
And another idea: Some other platforms have moderation be some kind of feed. We could expose moderation decisions as some kind of feed people can subscribe to. That way I could say: I trust team XY to do the moderation for me. And someone else could subscribe to a different “court” or team of moderators who are more or less strict. I kinda like that idea as well. But it’s also guaranteed to be (ab)used for the usual political, online ideology/identity war. Not sure if that’s a bad thing or a good thing.
If you combine both concepts, you’d get a solution which isn’t “central”. And it offers customizability to the users, and/or individual instance admins.
Don’t put it on top, but make it part of the platform itself.
Then we would need to convince every platform dev to implement it with a predefined standard and wait for the release. With the independent server + bot approach it gives more flexibility and every server admin can add the service without the platform to support it.
We could expose moderation decisions as some kind of feed people can subscribe to.
But then it would depend too much on the views of one instance. Every instance has their own set of rules, which is their right and is fine. But not all instances would like to apply the mod decisions based on rules they might not even have. With a common 3th-party jury mod actions are limited on only a set of rules all instances agree on, while they can apply more fine-grained rules themselves and not interfering with other servers.
But it’s also guaranteed to be (ab)used for the usual political, online ideology/identity war.
This is also a big problem with following mod actions of a server which also publishes posts. There will be a bias and if that server has beef with someone else, it will lead to injustice.
Also another thing about “feed people can subscribe to”: this proposed system is not for curated blocklists between users or such. It is for actions which have to apply to whole instances like account purges to ensure illegal media is not distributed.
I think 1) and 2) are kind of what makes the Fediverse fun. We’re not just an accumulation of sometimes compatible servers. We do make an effort to span one big, interoperable network and put in the effort to standardise and find common solutions. And we also try to offer means to individualise the experience on different instances.
It is for actions which have to apply to whole instances like account purges to ensure illegal media is not distributed.
I’m afraid we also have additional issues with different jurisdictions. I mean I’m a server operator based in Germany. I have different rules for hate speech than for example someone based in the USA. How I can and can’t phrase certain things. The UK has some different/strict internet laws. And someone from Japan for example has a whole different set of rules for some kinds of posts… It’d be difficult to come up with a central solution which isn’t some lowest common denominator. Which probably will be overblocking in this case?
It’d be difficult to come up with a central solution which isn’t some lowest common denominator. Which probably will be overblocking in this case?
It should be lowest common denominator with underblocking; so the intersection of rules, not the union. If one needs more strict blocking, nobody stops them. The court is not supposed to replace instance local moderation, just to augment it to allow more swift central decisions on certain topics.
Also the implementation can allow following more than one court. So for example once for international internet law (really lowest common denominator) and one for German internet law.
Guess we’re going for different goals, then.
For me in the role of an instance admin, I’d like some solution which properly cuts down on work that ends up on my plate. I’d rather have something so I can pool resources with other admins. I feel technically, it’s not all that different from your suggestion, so I’d just go the extra mile and maybe that’d be something to properly rely on. I mean ultimately one of my main responsibilities is to enforce law and the rest of moderation comes second to that. But I’d also use whatever people have to offer to at least make it a bit easier.
And nobody wants to see all the spam, nor do I want to get nasty pictures in a DM. All that stuff should disappear more swiftly.
Interesting idea. Not an instance owner, and not familiar with the finer details of federation, so no idea if it could be done. My initial concerns is that a majority vote won’t work, if you have to wait for the majority of instance mods to vote it’s going to be too slow. Maybe some degree of instance level config helps. For example, if i had my own instance, i would follow ban/delete recommendations from the likes of .world, .dbzer0, but maybe not from .ml without a human reviewing it first. Also need to ensure the ‘court’ does not get flooded.
Why not .ml?
They have a… different… take on politics.
Just as food for thought, there has also been an idea in the opposite direction - decentralize voting even more
It does sound like a decent idea, just not sure how it will work with the different ideologies of the Lemmy instances.
Things that are banworthy on one instance are not banworthy on another.I think it sounds good only until put in practice.
If 99% is useful but 1% goes against the instance’s philosophy, they have to accept this 1%. But if someone involved in picking what is put in this overarching moderation starts pushing this 1% to 2%, 4%, ever more through the Overton Window, it would deteriorate more and more the instances that had to accept the original 1%.
And even without the possibility of intentional sabotage, one rule being applied for every individual instance would still mean instances would need to adapt for the lowest common denominator, meaning a slow end of individuality. And if a small instance is no different than a big one, those bigger ones start canibalizing the userbase of the smaller ones.
What I expect to happen, and that already does to some extent, is to clusters of instances with similar interests to form, and by extension being prone to accept aligning moderation and philosophies with their direct allies.
But that’s still mostly a bilateral move, not an one rule to govern them all, so dependence or independence is still tied to individual interests.
My idea is to limit the scope of the rulings to be limited to very basic laws (like preventing illegal material). Everything outside this scope is still the sole matter of the individual instances. It is like the International Court of Justice, only to judge based on common rules, while not interfering with “local politics” (while it can be used to get consent of all participating instances for other procedures).
When it is visible that judges use their position to apply rules of their home instance to everyone else, they should be voted out.
Illegal in one country may not be in another. Like in Brazil, to my knowledge there is not rules against console mods and game dumping, but in Japan console mods are illegal and in the USA dumping Nintendo cartridges. Furthermore, in the Brazilian constitution, one of the opening statements is that “you shall not do or cease to do anything if not in virtue of law”.
So in this example, if 3 instances are tied by such an external organization, one for each country, which will be breaking their local laws to allow or not discussion on modding and dumping?
And on judges being voted out, Brazil also works as an example: Ill-intended judges just need to sit quietly enough for long enough until they have a majority, and then they can rule as they please.
This is a way for white people to abuse power dynamics and harm marginalized people.
Moderation is a skill and requires experience, more than most moderators have. It’s ideally a paid role that people have to compensate them for the difficult work they do.
The mentality that its free or non-skilled work is due to tech companies being terrible role models for morality, worker equity, and justice. You cannot imitate these systems to succeed, they are based in bottomless pits of money that they use to squash human rights violations.
Judges in court systems are ideally well trained and experienced in the law. Its a paid gig and prone to being terribly abused.
When some instance that wants mutual aid posts to disappear gets a “judge” to do this would you tell the disabled person on your instance that they cannot ask their community for help? Where would you tell them to go when the people who make the most impact are cut off? Would they be banned for that? What choices to people harmed by “judges” have when they are targeted unfairly?
The amount of overhead managing moderation on one instance can be difficult, but sifting through decisions made by a bunch of anonymous “judges” sounds like an utterly depraved torture chamber. In the end the choice would be whether the instance agrees with the (unqualified, unvetted, unpaid) “judge’s decision” which ultimately brings this back to where we are now.
