TROM Social | Rokosun @ TROM Social

Types of moderation issues

A third type is unfair moderation, when different accounts are treated differently by moderation. For example, lemmy.ml 's admins have recently been accused of couple of times to moderate differently people with different political opinions

What content should get blocked?

I think the answer here should be simpler : an instance has rules, and should ban (local or remote) content that does not obey them. In the absence of refined moderation tools, an instance could block any instance that does not have the same rules.

I have been thinking about a concept of tiered rulesets. For example an instance could allow the (non CW'd) diffusion of naked nipples, but could require the user to tick a specific box when writing a post that contains them so that it does not get sent to federated instances who forbid them.

Note that an instance's rule do not only depend on users' preferences and social norms

@Rokosun

in reply to Liwott

Rokosun

in reply to Liwott • 3 years ago from Fedilab •

A third type is unfair moderation, when different accounts are treated differently by moderation. For example, lemmy.ml 's admins have recently been accused of couple of times to moderate differently people with different political opinions

A third type is unfair moderation, when different accounts are treated differently by moderation. For example, lemmy.ml 's admins have recently been accused of couple of times to moderate differently people with different political opinions

When users don't agree with the moderation decisions made by their moderators, that's when these users accuse it as "Unfair moderation". For example, if I was a moderator and I started banning all religious content, then some users will be very happy with it and thank me for removing all the "religious propaganda" and other harmful ideas, but then other users will be pissed at me for doing the same and accuse me of "Unfair moderation". This is the same thing that happened in my example of that big instance blocking the feminist one, some users were happy but others were pissed. Actually, the entire point of this post was to make people understand how subjective these things are, different people want to block different types of stuff than others. Now here the only reason this problem happened is because those Lemmy instance admins were blocking content on behalf of all users, including ones who didn't want to block out people with different political opinions. If users could choose for themselves what content they wanna block then this problem wouldn't exist.

I think the answer here should be simpler : an instance has rules, and should ban (local or remote) content that does not obey them. In the absence of refined moderation tools, an instance could block any instance that does not have the same rules.

What difference would writing down these rules make? If that big instance changed their rules to mention that feminist instances would get blocked before actually blocking them, then the impact is the same. All that the rules does is that it adds some transparency maybe. In the previous post where we discussed some of the technical ways to improve moderation, there I discussed about transparency and suggested to notify users when an instance is getting blocked and also show them the reason for its blocking.

I have been thinking about a concept of tiered rulesets. For example an instance could allow the (non CW'd) diffusion of naked nipples, but could require the user to tick a specific box when writing a post that contains them so that it does not get sent to federated instances who forbid them.

I feel like this would complicate things, and its bad UX when a user has to go through a checklist to see which all applies to the post they're writing. Note that nudity isn't the only thing people get offended for, so the list would be very long. Just to be clear, to me its not a big issue if an instance admin want to hide some type of content from their local/federated timeline, its their instance so it sounds fair to me. The issue is when they outright block users, so if my admin blocks ToplessTopics then I as a user won't be able to follow that account, this looks more like censorship to me because I explicitly chose to follow a person that's not even on your instance and then you're stopping me from doing so.

Note that an instance's rule do not only depend on users' preferences and social norms, but also on law and moral.

I just wanted to point out that law and moral is dictated by the social norms of the time. For example, it was illegal for woman to vote, gay marriage was and is still illegal in many countries. Also this is a post I saw last day which talks about how slavery and other oppressive systems were legal at the time - mastodon.social/@197winstonsmi…

For exemple, an admin may not want to contribute to any sharing of child porn. For the latter case, hard block remains necessary.

If I'm an instance admin and a user on my instance accessed illegal content from other instances that I have nothing to do with, then am I contributing to sharing that content? I'm not a lawyer or anything, so I don't know how it'd look from a legal perspective. When you visit content from other instances then your server is simply acting as a proxy for viewing that content right? Proxies have existed long before fedi, so it'd be interesting to see how they're handled legally. And I don't know if your instance will cache the remote content that's being fetched, then that might complicate things. But then again, web crawlers have the same issue right? So if Google's web crawler fetched and cached a website that's sharing illegal content, then is Google liable for saving and sharing of illegal content?

Also, I kinda feel like maybe this whole issue could be solved from a technical perspective, if caching is the problem then disable caching, if proxing is the issue then it'd be nice if users could just directly fetch the remote content instead of going through their instance. There have been some issues in the fediverse that causes remote instances to show incorrect number of likes/reshares on a post, and sometimes even some of the posts won't be federated properly, so the same profile might appear different when viewing from different instances. So the proxy/caching system has other usability issues with it too, it would be nice if they could fix these.

When I see the symmetry of the social structure's description (the conclusions seem similar for sexists and feminists), I fail to understand what was the point of doing that "feminism in the past" detour.

I did that to make people understand how subjective these things are, which was the whole point of that post, I didn't wanna go into the technical stuff that much and just explain the social impacts of moderation. If I was born in a sexist society then I would've lived my entire life being a sexist piece of shit without even realizing why or how that was harmful. Unless maybe if I got exposed to feminist ideas and it changed my mind, then I'll be one of the few people at the time that supported feminism. If you look at the history of human civilization then you'll see that sexism is only one of many things we've been wrong about, our sense of "right and wrong" have changed a lot during the years. I think @Tio has even written a book about morality and ethics, you can read that from tromsite.com/books
I haven't read that particular book, but I know it talks about the same things. And this applies even today, there might be some ideas in our society that we'll realize later that it was harmful. This idea might make people uneasy, but its just the truth. There might be some things we do today that future civilizations might look at with disgust and anger/fear, just like how we look at ideas like slavery/sexism today. So today I have my own sense of morality, my own method of distinguishing between right and wrong, but I do acknowledge the fact that I'm probably wrong in many of the things I believe today. If I was made a moderator for say a 100 people then that would make me very uncomfortable, because I don't want to choose for these 100 people what they should see or not see, that would be like me imposing my own sense of morality on others. As an instance admin I'll have the power to kick people out of my own instance, and maybe also hide content from the local/federated timeline of my instance, but choosing which accounts my users should be able to follow from other servers, that's too much power I think.

And the conclusions were similar for sexists and feminists because I wanted to showcase how all 4 of these groups think they are right from their own sense of morality - Type 1 Feminist users, Type 2 Feminist users, Type 1 Sexist users, Type 2 Sexist users. And there's just simply no way for us to know which one of them are the "correct" one we should listen to, what usually ends up happening is that moderators will go along with what they think is right and impose that choice on all their users. In a sexist society its fair to say that pretty much all feminist instances would be blocked out of fedi, because when we look at the overall network then social norms have a big impact on it.

@Tio @Liwott

Tio likes this.

in reply to Rokosun

Tio

in reply to Rokosun • 3 years ago •

Nitter, Invidous and the like, cannot be blamed for fetching remote content from twittter or Youtube. So if someone posts child porn, or god forbid GUNS on twitter or youtube, and others access it via our nitter/invidious instances, it is not even legal in our stupid world of stupid laws, to accuse me. It is simply proxying and the content from youtube/twitter is not even on my server. Perhaps federated networks should do that if possible. It is already doing something similar in the sense that I, as admin, can choose to remove remote cached content daily or immediately. You can still connect to others, but the content needs to be fetched every time you "poke" at them or something like that...

This is the saner way of doing it. Not the butcher way please! Cut and slice! But the librarian way. Sit down and come up with a solution that is future-proof and does not impede people's ability to connect with others.

Rokosun likes this.

in reply to Rokosun

Liwott

in reply to Rokosun • 3 years ago • •

(mind if I don't reply all at once? it may become quite long otherwise, and some subtopics make sense to be split)

different people want to block different types of stuff than others.

What I meant with unfair moderation is when the same content is treated differently according to who wrote it. For example, see this comment on a Lemmy drama. The first user seemed to benefit from a laxist treatment because they shared the admins' political opinion, hence the accusations of unfair moderation. (for the record, that first person ended up getting a 5-days ban) Please, anyone reading this, let us not turn this about a discussion of whether or not there was a double standard, I only meant to say that this kind of double standards may exist, and not every dispute on moderation concerns the two types mentioned in the OP

Rokosun likes this.

in reply to Liwott

Rokosun

in reply to Liwott • 3 years ago from Fedilab •

What I meant with unfair moderation is when the same content is treated differently according to who wrote it.

Okay I understand what you mean now. So your point was that moderators could treat people differently based on their political believes, religion, race, etc. This might happen sometimes, as moderators are humans too, and well, humans can be biased. For example, an admin might be more reluctant to block an instance if they have close friends on it compared to other instances where they don't know anyone (so they block it without much thought). I think this could be seen as one more reason why users choosing for themselves is better than letting admins handle everything.

@Liwott

Liwott likes this.

in reply to Rokosun

Liwott

in reply to Rokosun • 3 years ago • •

law and moral is dictated by the social norms of the time

They are often related to them, but not always dictated by them.

In the case of law, it is independent from the people's social standard if the people are not participating to the decision, like in a dictatorship.
Even under a democratic regime, it may not change as fast as the social standards.

As for ethics and morality, deontology is pretty much dictated by the social norms, but I do not think it is that clear for other forms of ethics. One may be utilitarian and act in a way that minimises suffering,
independently of what is socially acceptable.

By adding other layers of subjectivity, this goes towards confirming your point that those things are subjective.
Note that this is not a point that is specific to federation, and also concerns the instance's own terms of service.
Why is why...

What difference would writing down these

law and moral is dictated by the social norms of the time

They are often related to them, but not always dictated by them.

What difference would writing down these rules make?

I was referring to the instance's terms of service rather than a new set of rules that would be written for the occasion.
So the admin would moderate remote content in the same way as the local one.

But the admins of one instance are not enough to moderate the whole fediverse.
This is why they rely on other admins to do their job.
And if that job is not compatible with their own rules, and they don't have the manpower needed to do it in their stead, then they may chose to defederate.

When you visit content from other instances then your server is simply acting as a proxy for viewing that content right?

There is also the fact that your instance may have made you discover that content, through a federated timeline, a reshare. See this related subthread.

I'm not a lawyer or anything, so I don't know how it'd look from a legal perspective.

Oh, me neither !
But independently of the legal status, we can discuss the morality of it.
I took the extreme example of child porn because I think most people would not want to facilitate its diffusion.
So if another instance is laxist in banning that content, federating with them may contribute to distribute it.
So I may want to block that instance to provide an obstacle to that diffusion.
In that case, what I want is really hard blocking.

I did that to make people understand how subjective these things are

To conclude this part, I agree that there is subjectivity in choosing the rules, but people who set up an instance have the right to subjectively chose those rules, as they don't have any legal or moral obligation to provide an inconditional service to anyone interested.

@Rokosun

Rokosun likes this.

in reply to Liwott

Rokosun

in reply to Liwott • 3 years ago from Fedilab •

They are often related to them, but not always dictated by them.
In the case of law, it is independent from the people's social standard if the people are not participating to the decision, like in a dictatorship.
Even under a democratic regime, it may not change as fast as the social standards.

Yeah, in a dictatorship the dictators will force their laws upon everyone even if people don't agree with them. And yeah, in a democratic regime usually the social standards change first before it starts affecting the laws, lots of people had to fight for their rights in the past, and there still are many people who do that. In a way, this is what ToplessTopics is trying to do, if non-sexual nudity becomes normalized only then we could expect the laws to change, currently you could get arrested for not wearing clothes in public.

As for ethics and morality, deontology is pretty much dictated by the social norms, but I do

They are often related to them, but not always dictated by them.
In the case of law, it is independent from the people's social standard if the people are not participating to the decision, like in a dictatorship.
Even under a democratic regime, it may not change as fast as the social standards.

As for ethics and morality, deontology is pretty much dictated by the social norms, but I do not think it is that clear for other forms of ethics. One may be utilitarian and act in a way that minimises suffering, independently of what is socially acceptable.

I know that some people might have a different sense of morality than others, but in general its fair to say that most of us are influenced by the society we grew up in. So if we look at the society in general instead of looking at individuals, then social norms have a big impact on it.

But the admins of one instance are not enough to moderate the whole fediverse.
This is why they rely on other admins to do their job.
And if that job is not compatible with their own rules, and they don't have the manpower needed to do it in their stead, then they may chose to defederate.

I think its unreasonable for us to expect an instance admin to moderate the whole fediverse, they might be able to moderate their own instance because they have full control over it, but other instances are out of their scope, and content shared on other instances are not the responsibility of this admin. This will become more clear later.

There is also the fact that your instance may have made you discover that content, through a federated timeline, a reshare. See this related subthread.

Admins could hide content from the federated timeline so that they're not discoverable, this way only the people who choose to explicitly follow someone from that instance will see their content. And if you find the content through a boost/reshare then that'll only happen if you follow someone who boosts such content. And if you did follow someone like that then you would unfollow that person when you see their boosts. IMO its worse if you're not able to see their boosts, if someone you follow is boosting illegal content from other instances then you should know that, right? If you don't see these boosts then you'll keep following that person thinking that everything is fine. Also, what if that person just shares a link to any illegal content instead of boosting fediverse posts? This is a problem that exists for the entire internet, and there's no easy way to fix this.

But independently of the legal status, we can discuss the morality of it.
I took the extreme example of child porn because I think most people would not want to facilitate its diffusion.
So if another instance is laxist in banning that content, federating with them may contribute to distribute it.
So I may want to block that instance to provide an obstacle to that diffusion.
In that case, what I want is really hard blocking.

I agree CSAM (child sexual abuse material) is an extreme example, its fair to say that almost everyone would agree to blocking such content, so how do we deal with it? So remember when @Tio said before that fediverse is kinda like a browser? When bad content is hidden from the local/federated timelines then it really is like a browser because users will have to go and manually find these content or else they won't reach it, or maybe they find it because someone they follow linked to it which is similar to you finding a website on the internet through links from other websites you follow. So this is where fediverse is a bit different from other centralized platforms, if you find CSAM in Facebook for example then you can just report that content and it gets removed from the platform itself, but on fedi if you do the same then 2 things can happen - if the content is on your own instance then your admin could remove it just like Facebook did, but if its from other instance then its up to the admin of that instance. If the other admin chooses to remove that content then its fine, but if they don't then it gets a bit tricky. So the only thing your admin can do at this point is to defederate from that instance, but note that the content is still there, your admin don't have the power to remove it, so anyone can still link to that content and it'll still be there, this is not the case with Facebook because once they remove it then its gone. So this is very similar to your browser blocking you from viewing bad content that's still there on the internet, but I'm pretty sure that people will get pissed if Mozilla firefox or Google chrome just forced this upon their users. If you use an ad-blocker like uBlock Origin then it'll actually prevent you from viewing malware sites and such, but this is okay because you have full control over which blocklists to enable and such, these are not forced upon you. Now I use tblock.codeberg.page for blocking ads, malware, etc system wide. For example here is a blocklist you can use - github.com/StevenBlack/hosts
You can see how they have different categories like adware, malware, fake news, gambling, porn, etc and users are free to choose whatever blocklist they want. Having choices like this would account for the subjective views of users, here they get to choose what content they wanna block.

So now when fediverse admins block content then its just like a local block, like how I can locally apply blocklists to block malware and such on the internet, except that its forced upon all users without even giving them a choice. And the bad content is still there on the other server and anyone can view it using a web browser, its not even removed like in Facebook. The reason I suggested the idea of blocklists before is to make this whole thing more efficient, transparent and easy to manage. If users could modify which all blocklists they wanna enable like in uBlock Origin, then it'd also solve the issues related to bad moderation, they can choose for themselves what type of content to block, how strict the blocking should be, etc.

To conclude this part, I agree that there is subjectivity in choosing the rules, but people who set up an instance have the right to subjectively chose those rules, as they don't have any legal or moral obligation to provide an inconditional service to anyone interested.

People who set up the instance have the right to subjectively choose the rules for their own instance, the issue is when they try to control what happens with other servers which are basically like separate websites on the internet out of their control. I feel like most of the issues could be solved by simply hiding the bad content from the local/federated timeline of that instance. And I've heard that the official mastodon app don't even have the local/federated timelines on it, they use the newly added explore page for finding users and posts. And now to the other issue you mentioned earlier - its not just you finding bad actors, but bad actors can find you and attack you. For solving this issue blocklists could be very useful, a well maintained blocklist is much better than relying on a single admin to do all the moderation for you, and you could apply multiple blocklists too. And another solution is to make your posts private by default so that bad actors can't find you, choosing who can see your posts is much more powerful than relying on moderators/blocklits to protect you.

@Tio @Liwott

Tio likes this.

in reply to Rokosun

Liwott

in reply to Rokosun • 3 years ago • •

Also, what if that person just shares a link to any illegal content instead of boosting fediverse posts?

A link to illegal content may be forbidden by the instance's rules. Or required to be appropriately CW'd. Anyway when we say "bad content" in the context of moderation, it is not wrt a kind of universal moral standard, it is wrt the instance's rules. If the rules happen to not be clear, this is not a moderation tool issue.

I think its unreasonable for us to expect an instance admin to moderate the whole fediverse, they might be able to moderate their own instance because they have full control over it

When I register on a website, be it a fediverse instance, with some rules, I expect the content that I can find on that website to obey the rules in question.

And if you find the content through a boost/reshare then that'll only happen if you follow someone who boosts such conte

Also, what if that person just shares a link to any illegal content instead of boosting fediverse posts?

I think its unreasonable for us to expect an instance admin to moderate the whole fediverse, they might be able to moderate their own instance because they have full control over it

When I register on a website, be it a fediverse instance, with some rules, I expect the content that I can find on that website to obey the rules in question.

And if you find the content through a boost/reshare then that'll only happen if you follow someone who boosts such content.

Or if someone from this instance replies to such content and you see it on the local timeline. (At least soft-)Blocking is useful in that case.
Even about people you follow. If you register on an instance with some rules and start following people there, you expect their content to correspond to the instance's rules.

IMO its worse if you're not able to see their boosts, if someone you follow is boosting illegal content from other instances then you should know that, right?

Good point. Maybe rather than (or in addition to) (hard of soft) blocklists, it would be appropriate to have CW-lists. The sae way Friendica allows you to CW text containing a given string, the admin could enable automatic CW for posts coming from certain instances.

so anyone can still link to that content and it'll still be there, this is not the case with Facebook because once they remove it then its gone.

For that matter, one can also post links on Facebook, and this goes back to the question of whether links to bad content is itself considered bad content by the rules. I think on Facebook it is, right?

A linked content is anyway less accessible that content directly on the website, as users don't necessarily click on every link they see. So one does limit the diffusion by modding out the remote post.

So this is very similar to your browser blocking you from viewing bad content that's still there on the internet

A browser in a piece of software that runs locally. You do not use Mozilla's or Google's computing power to load the bad content that you see via the browser. The admin may not want their servers to work towards facilitating to diffusion of bad content.

the issue is when they try to control what happens with other servers which are basically like separate websites on the internet out of their control.

Federating with some content makes it accesible from your website, so it is really about moderating the content shown to your website users.

@Rokosun

in reply to Liwott

Rokosun

in reply to Liwott • 3 years ago from Fedilab •

A link to illegal content may be forbidden by the instance's rules.

I don't think you understood the point I was trying to make. I wasn't talking about instance rules here, this was in response to you saying that users could get exposed to bad content via boosts from people they follow, and your point was that if the admin had blocked the instance where these posts are originating from then these boosts won't show up. So I was trying to point out that you can't fix this issue by defederating, maybe it'd hide the boosts but people could still post links instead of boosts, so there just isn't an easy solution for this. Whether linking to bad content is forbidden or not will depend on the instance rules of that particular user, but in your example the same user was boosting bad content instead of linking to it, so I assumed that if their instance allowed boosting then it'd allow linking too.

When I register on a website,

A link to illegal content may be forbidden by the instance's rules.

When I register on a website, be it a fediverse instance, with some rules, I expect the content that I can find on that website to obey the rules in question.

Yes, you can expect the content you can find on that website to obey the same rules. But the beauty of the fediverse is that you're also able to interact with other separate websites where the rules are different from your own, and as long as you properly understand how this federation thing works then it won't be confusing for you, I know some people new to fediverse might find this concept a bit hard to grasp but given enough time they'll get it.

I think its reasonable for you to expect the content you see on the local/global timeline of your instance to respect the same rules, and also everything the users of that instance posts there. But when you get into other instances, then its really like a separate website that you're browsing, the interface and everything might look the same but the content is hosted on an entirely different website with different set of rules.

Or if someone from this instance replies to such content and you see it on the local timeline. (At least soft-)Blocking is useful in that case.

Oh I forgot that Friendica also shows you the post if someone you follow replied to it, this is not how other platforms work and even in Friendica you could change this behavior in the settings. And yes, people may find soft blocking useful, if you think about it then soft blocking is kinda like subscribing to a blocklist made by your instance admin. Soft blocking is generally better than the default way of blocking we use now, but blocklists would be a much better solution, not only because it gives users more choice and freedom, but also it makes the job of moderators much easier because they can just collaborate on making a really good blocklist.

Even about people you follow. If you register on an instance with some rules and start following people there, you expect their content to correspond to the instance's rules.

If you only follow people from your own instance then yes you can expect their content to correspond to the instance's rules. But when you follow someone from a different instance then they'd be following the rules of their own instance not yours. Again, remember that these instances are separate websites independent from one another, just because you followed someone through your instance doesn't mean its the same website, you could be interacting with people from other instances which are wildly different from your own, sometimes they can even be entirely different platforms.

In general though, if you're finding users by browsing through the local/global timeline which is curated by the admin to hide content which are against their rules, then you probably won't be exposed to any content which differ from their rules. For example, if nudity is against the rules of an instance then you probably won't find users like ToplessTopics through their timeline. Your only way of finding ToplessTopics would be if someone else told you about it and you decide to follow them. Maybe a person you follow boosts posts from ToplessTopics and talks about them very often, then you could get exposed to it that way. This is similar to how we form relationships in real life, if you have a friend who's friends with someone else you don't know, then both of you could meet and find each other.

Good point. Maybe rather than (or in addition to) (hard of soft) blocklists, it would be appropriate to have CW-lists. The sae way Friendica allows you to CW text containing a given string, the admin could enable automatic CW for posts coming from certain instances.

This feature is called content filters in Friendica. And they also have an Advanced content filter which is way more powerful, it has like an entire syntax and everything for making rules based on that, so for example if you want to filter posts containing more than 3 images then you could write a filter rule that does that. So yeah, that's very useful. You might actually be able to use the advanced content filter to write a rule that filters out all posts from an instance. I don't know if admins should be allowed to set the default filter rules for their users, but maybe it'd be helpful in some cases. As long as the user is able to change these defaults for themselves then its not a big issue if there are some predefined filter rules. In my previous post I explained that its good to have different levels of blocking, so if an admin could just write up a simple filter rule for their users instead of outright blocking an instance then that's much better.

For that matter, one can also post links on Facebook, and this goes back to the question of whether links to bad content is itself considered bad content by the rules. I think on Facebook it is, right?

On Facebook I think they'll remove your post if you link to bad content. But on fedi this is complicated because there are multiple instances, so if the content is from your own instance then your admin will be able to remove it and its gone, just like how Facebook does it, but if the content is from other instances then your admin don't have the power to remove that content, maybe they can block it for users on this specific instance but the content will still be there.

A linked content is anyway less accessible that content directly on the website, as users don't necessarily click on every link they see. So one does limit the diffusion by modding out the remote post.

That makes sense, it might reduce the impact a little bit, but its not very effective. People also have things like link previews enabled by default, so there is not that much of a difference between a boost and a link now, and on Friendica they also have this quote-share thing right?

The point I was trying to make here was that in Facebook when they block/ban a content then they're banning it from their own servers they have full control over, that's why they're able to remove them. But on fedi when admins block content from other instances that content is not stored on their servers and they're not able to remove them.

A browser in a piece of software that runs locally. You do not use Mozilla's or Google's computing power to load the bad content that you see via the browser. The admin may not want their servers to work towards facilitating to diffusion of bad content.

Considering the way fediverse works currently, maybe a better analogy is to use a browser with a proxy. So when you're using a proxy/VPN to view bad content then you're using the computing power of their servers to load the bad content. But if these proxies or VPNs starts banning content they don't like then users will get upset, in fact many people actually make use of proxies/VPNs/Tor to bypass internet censorship. I know that many people living in countries like China and Russia face a lot of censorship and they have to use tools like Tor to get past this censorship. And when you ask the internet providers of these countries why they're censoring the internet then they give you a similar answer - "Our company don't want our infrastructure to work towards facilitating the diffusion of bad content". This is why I focused mainly on this idea of "bad content" and how subjective this notion is. There are countries today that ban LGBTQ+ websites because according to them that's "bad content". Now even if we assume that 90% of the people on that country agree to blocking these websites, this will never justify banning it for everyone including that 10% who did not want these websites to get blocked.

Federating with some content makes it accesible from your website, so it is really about moderating the content shown to your website users.

You can moderate the content that's on your website, but if you're trying to control what the users can access from other websites then that's closer to censorship. I'll explain the difference between these. Let's take the example of China, if I wanted to go to China and they don't wanna allow me inside their country then that's not seen as censorship, its their country so they can choose who they wanna let in. In a similar way an instance admin can control who can join their instance and what they can post there. So China not letting people in is not the issue, but when they do let people in then they control what part of the internet people can access from inside their country, this is what people call censorship. And I find this similar to when an admin tries to control what their users can access from other instance. If they're moderating their own instance then it makes sense to me, but if they're trying to block their users from accessing content from other instances then that can lead to censorship. Imagine the same thing happening with other federated services like email, you're sending an email to your friend who's on a different server and then Gmail blocks you by saying you can't send emails to this server. By doing this Gmail is breaking interoperability between email servers and this is a threat to decentralized services.

@Liwott

Liwott likes this.

Unknown parent

Rokosun

Unknown parent • 3 years ago from Fedilab •

@bellerome
To be clear, I never support things like name calling or racial slurs. If you read my above post you'll understand that my main point was about the subjectiveness of what we consider as offensive content or not and how this notion changes over time. I'm also not sure what you mean by "Democratic Government Ideology". Its kinda funny when you see almost every government on earth call themselves a "democracy", even the most oppressive governments call themselves that 🤦‍♂️
So this word "democracy" became a simple label and it has lost its true meaning, like the many other ideas which ended up the same. Communism for example was a good idea in its origins, but then later people used the same idea to oppress others and kill people, countries like China and Russia still call themselves a communist country despite having nothing to do with the original idea of communism.
So I'll never say things like "Any talk in opposition to <whatever ideology> should be blocked" even if I agree with the ideology itself, because ideologies li... show more

@Liwott

Rokosun

3 years ago •

Rokosun
3 years ago •

How to fix moderation on the fediverse

So there has been some talks recently about how fediverse handles moderation and some of its issues, and to see how we can improve moderation from a technical perspective. Now I have an idea that I want to share with you all, because so far I haven't seen anyone else talk about it, and I think this could solve a lot of issues with our current method of moderation. But before going into that, I want to explain the current situation and why this is important, so that you understand what issues I'm trying to solve.

Current situation

When it comes to moderation, Fediverse follows a more mainstream approach, where admins/moderators of an instance block users and other instances on behalf of you. And as a user you also have the tools to block/mute users or instances.

The issue

Current situation

The issue

Compared to all of the other centralized platforms, there is something different here on fedi, and that is the ability to block entire instances, which will, on an instant, block a huge chunk of users that are on it. This can be a very powerful feature for moderators, but if not used carefully this can lead to problems. For example, if someone decides to block an entire instance just because of a few bad actors on it, then they're also blocking all of the other users on that instance who haven't done anything wrong to get blocked.

If you're a fediverse user and your instance admin blocked another instance from federating, that means you can't interact with anyone on that instance anymore, but you as a user is never notified of any of this so you won't know when it happens. Just think about that for a moment, this is equal to cutting the cord with that server, stopping the federation entirely. And if you had any friends on that instance, then you will lose contact with them.

Federation doesn't solve all problems

So what people usually say is that in the fediverse you can move instances, so if you don't agree with the moderators of one instance, you can simply move to a different one. While it is great that we have this option in the fediverse, it is not as easy as it sounds. And one project that I want to note here is Hubzilla because from what I've read so far they do it much better, they have this thing called nomadic identities which are independent from the server you use, and so you can very easily move from server to server and bring all your data with you. And furthermore, on Hubzilla you can even clone your account to different servers, so this will be the equivalent of being on several different instances at the same time and they all mirror each other, how cool is that?!

Centralization makes it worse

Email is another federated service just like our fediverse. And if you look at email the problem there is that most people are on a few servers like Gmail, Outlook, etc. And so this gives the ones in control of these servers more power over other servers on the network. And if you look at the fediverse now there is a similar problem arising here, most users on the entire fediverse are on a couple of popular instances, so if your instance got banned by these big instances then you're pretty much out of fedi. Centralization & censorship is a threat to federated networks, so keep that in mind.

And some people also say you can start your own instance, but you can't expect the average social media user to go through the process of setting up a server and maintain it, if you have the money and technical skills to do that then you can, but this is not a viable solution for the majority of people. Peer to peer apps are much better in this regard because there everyone is automatically made a node in the network, so they don't have to do anything for it. But on fedi this is not the case, you have to rely on servers.

Solutions

There is no simple solution to this problem, but I'll list some of the key points here that I think are important:

Transparency regarding moderation actions

So I've mentioned above how users are not notified when their admin blocks an instance, this is the first and foremost problem we should try to solve. All of the moderation decisions should be made transparent to the users of that instance, and the admin/moderator should also explain the reason why they chose to block these. So I've been informed that they recently found a solution for this on Friendica, so that's great 😀. What they did is that they implemented a notification system that notifies all users of an instance when their blocklist gets updated, and it'll show a link to that blocklist where they can check to see all the instances that are currently blocked and the reason for their blocking. This is how it should be done, and I want other fediverse platforms to do the same.
One thing that I want to add here is that it'd be nice if users are informed about it when someone they follow (or that instance) gets blocked. As a user I want to know when this happens, or else I will be in the dark about it, not knowing that I can't see/interact with that person anymore.

So the more transparent you make moderation, the better it'll be.

Different levels of blocking

This is an important one I think, moderators/admins should have the power to "hide" a user from the public instead of outright blocking them. There will be some edge cases when it comes to blocking, so the moderator may not want to outright block it, but just simply hide it from the public timelines or something. This feature might already be on some fediverse platforms, I remember opening a profile from my mastodon account and it blurred the profile and showed me a little warning saying "This profile is hidden because of sensitive content" or something like that, and I had to click on a button to agree to seeing that. I really liked this little thing, because the profile wasn't blocked or anything, but my moderators could show a warning instead.

So blocking shouldn't be an all-or-nothing strategy, there should be different levels of blocking, because this is better for moderators and users alike 🙂

My idea

So this is that big idea I had in my mind, let me explain. I mentioned in the beginning how fediverse allows its users to block/mute other users and even instances. So currently when you first join fedi your personal blocklist will be empty, even though your instance admin might have blocked some users/instances on your behalf. But what if we actually made it so that the instance blocklist syncs with the user blocklist? 😀 So in that case when you join fedi your blocklist will be filled with your default instance blocks, and every time the moderators block a new server it'll get added to the blocklist of all users. This kind of mechanism would allow moderators to create a safe space for everyone, but at the same time if someone wants to unblock any of these they can do that on the user level. So if I found out that I have some friends on an instance that is blocked for me, I can just unblock that specific instance so that I can still talk to my friends, and this is a user level action so I can only unblock it for me, not for other users on my instance.

So lets call this method of blocking as "soft blocking" because as a moderator you're simply just updating the default blocklist for your users, and if any of them really want to then they can unblock it for themselves. This should be the default way of blocking for fediverse moderators because I think it satisfies the needs of both moderators who want to create a safe space for their users, and also alleviate the concerns of people who are worried about censorship and losing contact with their friends.

Exceptions

There might be some exceptions where instance admins might want to apply a "hard block" so that users can't unblock it even if they want to. I'm specifically thinking of blocking illegal content here, because in that case there might be some laws that make it tricky, so if you're worried of legal issues then having the option of a hard block could be useful. One thing to note is, for the users they will still see these hard blocks on their personal blocklist, but there won't be that unblock button that allows them to unblock these.

P.S. Let me know what you think of this idea? Is it practical or doable? Is there anything else I missed?

Addendum

So far we've been talking only about the technical side of things, to see how we can improve moderation by giving more power and choice to the users. But looking at some of the replies I got on this post I felt like maybe a lot of people don't understand why we're doing it in the first place, why should we give more power to the users over admins? So I wrote a separate post, to answer that question, but also to make people understand how much of a complex topic moderation actually is, it isn't just simply "blocking bad content".
The social impacts of moderation & censorship - social.trom.tf/display/dbc8dc4…

The social impacts of moderation & censorship

This post is related to these previous discussions we've had about moderation: https://social.trom.tf/display/dbc8dc44-4262-df68-a596-611155190199 https://so...

^{social.trom.tf}

Tio likes this.

reshared this

in reply to Rokosun

Tio

in reply to Rokosun • 3 years ago •

I agree with what you wrote. I do however think that the Fediverse is already difficult to grasp for most people and if they are now asked to manage a blocklist that is updated by the admin, can be super tricky. Also, what does soft blocking mean for the instance that is blocking? Display the blocked content only for the users that accept to see it? How to do that is it possible?

For me the most sane and simple approach is that of blocklists like adblockers. Say in your settings have lists called: Nazi, porn, bots, scams, etc.. And you can toggle them and that's all. And the fight and work goes in the backstage with us the admins and some vocal users trying to keep these lists sane.

The best would be to simply let the users do whatever they want in regards to blocking, and you as the instance admin can remove users from your own instance and keep it "clean". Stop trying to moderate other instances! This is a bad practice.

In any regards I strongly feel that we need to talk about these things now. I do not see any good approach currently, except for how Friendica

Rokosun likes this.

in reply to Tio

Rokosun

in reply to Tio • 3 years ago •

Display the blocked content only for the users that accept to see it? How to do that is it possible?

I don't know much about the technical side of this, but on platforms like mastodon they already allow users to block other instances, so this isn't really a new thing. If all users on an instance except a few blocked an instance, then only that few who has unblocked it will see that instance.

For me the most sane and simple approach is that of blocklists like adblockers.

Display the blocked content only for the users that accept to see it? How to do that is it possible?

For me the most sane and simple approach is that of blocklists like adblockers.

I have talked about this idea before and I also think this is the sane way, but now to me it seems like it'd be better suited for a peer to peer network where there are no moderators or instance admins, because there you would actually need such lists as there is no one else moderating for you. The important difference between these two is that in a p2p system all users are equal, but on a federated network like this users are relying on instances that are run by other people, so from a technical perspective that gives instance admins more control over the network. Currently all of the moderation is done by instance admins, so if we implement an ad-block like approach then what will these admins do? What if an admin wants to block all Nazis from their instance? what could they do then? You could say this should be left to the users, but I don't think people running these servers would be comfortable with Nazi content federating with their instance, and also there can be legal issues. So the truth is that instance admins want that control, they don't want to leave everything to the users because they feel like they should have a say in how their server works, because they're the ones maintaining it, paying for it, etc.

And my idea of syncing instance blocklist with the user blocklist, that is actually an extension of this ad-block idea. But here instead of users subscribing to a blocklist maintained by someone else, their instance admin maintains it. This gives instance admins some say over what happens on their instance, as opposed to leaving everything to the user which is more suitable for p2p systems.

@Tio

in reply to Rokosun

Tio

in reply to Rokosun • 3 years ago •

I have an instance, right?! I control who registers (who is) on this instance....isn't that enough? Why should I control what others see from other instances via mine? It is like Firefox blocking certain websites. These platforms should be regarded as browsers for the users. Yes temporary data is stored from other locations on my server, but it is temporary and that's an important point. You can complain about the users that are registered on my platform, since I am the one allowing that, but you have no right to complain about remote content that is temporarily stored on my server. And if these man-made rules called "laws" do not see this difference, they should adapt.

As for your idea I get more of it now, but I'd very super uncomfortable to have an admin forcefully adding a list of blocked entities in my settings, even if I can undo that. Maybe ask them if they want to apply that blocked list...

Rokosun likes this.

in reply to Tio

Liwott

in reply to Tio • 3 years ago • •

These platforms should be regarded as browsers for the users.

I disagree. The browser does not serve you any content. An instance's federated timeline puts you in contact with the content from another instance, even when you did not ask for it. Conversely, if your instance federates with one containing undesired users, it shows them your content without them knowing you in the first place. Imagine what happens if a jewish person registers then publishes on an instance that federates with an instance full of nazi trolls.

Note that I am not saying anything against the shared blocklists solution here, I actually find it very interesting.

in reply to Liwott

Tio

in reply to Liwott • 3 years ago •

An instance's federated timeline puts you in contact with the content from another instance, even when you did not ask for it.

Then we should focus on that aspect. Admins should be able to remove that or restrict that. The visibility of it.

Liwott likes this.

in reply to Tio

Liwott

in reply to Tio • 3 years ago • •

True for that way, but in the other way (not wanting to be exposed to e.g. bigots), it is really about controlling how posts are transmitted to other instances, not just visibility on your own.

in reply to Liwott

Tio

in reply to Liwott • 3 years ago •

If you do not want to be exposed to "bigots" then you can simply not search for them. You can easily control your news feed to only see posts from your chosen friends. I think that's default in friendica for example. Like a search engine, you won't see content unless you search for it.

By the way when you comment it posts twice.

in reply to Tio

Liwott

in reply to Tio • 3 years ago • •

By the way when you comment it posts twice.

Oh yes I reacted wrongly to not being able to edit comments in this thread. Won't happen for the future ones, sorry !

@Tio

Tio likes this.

in reply to Tio

Liwott

in reply to Tio • 3 years ago • •

If you do not want to be exposed to "bigots" then you can simply not search for them.

No, I was referring to my previous point :

Conversely, if your instance federates with one containing undesired users, it shows them your content without them knowing you in the first place. Imagine what happens if a jewish person registers then publishes on an instance that federates with an instance full of nazi trolls.

So it is about bigots randomly seing your content and then deciding to target you.

@Tio

Rokosun likes this.

in reply to Liwott

Tio

in reply to Liwott • 3 years ago •

Ah ok you mean if others see your content and then decide to troll you. But for one you can make your posts unlisted (only available to your followers not public - friendica has that). Second you can setup your account to not receive any interaction from anyone outside your circle of friends. These are tools already available. So they cannot reach you even if they wanted to.

Rokosun likes this.

in reply to Tio

Liwott

in reply to Tio • 3 years ago • •

@Tio
True for that way, but in the other way (not wanting to be exposed to e.g. bigots), it is really about controlling how posts are transmitted to other instances, not just visibility on your own

@Tio

in reply to Tio

Liwott

in reply to Tio • 3 years ago • •

@Tio @Rokosun

These platforms should be regarded as browsers for the users.

Note that I am not saying anything against the shared blocklists solution here, I actually find it very interesting.

@Tio @Rokosun

Unknown parent

Tio

Unknown parent • 3 years ago •

Good point actually.

Rokosun likes this.

in reply to Tio

Rokosun

in reply to Tio • 3 years ago •

@Tio @Émilie Fecteau

You are absolutely right Em, and there is the irony in that, I'll explain. So your point is that admins will abuse their power and do hard blocks more often than necessary. But.... If admins actually wanna do that, if they really are that power hungry, then how do you think instance admins will react if we didn't even give them an option to hard block? This is like taking power away from them and they're not gonna accept it, I don't even think Fediverse developers would be in favor of that decision if it upsets many instance admins.

So the truth is that you can't actually give all the power to the users as long as that user is relying on someone else. Self reliance doesn't exist on the fediverse, they're only possible on p2p systems, this is actually the reason why I talked a lot about p2p platforms before even when our di

@Tio

Dr. Percy reshared this.

Unknown parent

Rokosun

Unknown parent • 3 years ago from Fedilab •

@Émilie Fecteau
Em, what you're referring to here is the ability to hide a user from public timelines so that only people who directly interact with them can see their posts. This might already be possible, IDK about that.

What I refer to as soft blocking is different, its an idea I came up with where the admin is able to set the user's blocklist, and the user is able to override these changes if they so wish. And I don't think any fediverse platforms currently support this.
@sтυx💙💛

@stux⚡

Dr. Percy reshared this.

Unknown parent

Tio

Unknown parent • 3 years ago •

The long term consequences of this mindset are not pretty. You will decide for hundreds/thousands of users at once. Keep that in mind.

And let me give you an example: I know someone who tries to normalize nudity so she appears topless in many of her videos/photos. Most humans are afraid of the female nipple so she finds it hard to have a social presence. Now the only instances that allow nudity also allow all sort of crap: nazi, gore, etc.. She now makes an account there. You then see some of these crazy people interacting with your instance. I am a user on your instance and friends with this girl. You block that instance, and you have removed my contact with this person. Not good.

There are lots of saner ways to deal with this that won't ruin the fediverse.

Rokosun likes this.

Unknown parent

Liwott

Unknown parent • 3 years ago • •

RE: How to fix moderation on the fediverse (Critical)

@Émilie Fecteau @Rokosun

It is not actually possible to impose any constraint to instance admins, as they could just use another software (or fork it) that does not have the constraint. Hence why, as you said in a later comment, one needs to find a solution to the reasons behing the existence of instance blocking.

@Rokosun

Dr. Percy reshared this.

Unknown parent

Rokosun

Unknown parent • 3 years ago from Fedilab •

@Émilie Fecteau
Not just transparent, but if any particular user wants to unblock an instance for themselves, then they could do that. This prevents them from loosing contact with friends in case if that instance got blocked. Remember when you unblocked Aaron's instance so that you could talk to him? You could do that because the blocklist you applied was a user level block, that's the advantage of it.

Unknown parent

Rokosun

Unknown parent • 3 years ago from Fedilab •

In a mood

@Émilie Fecteau
I think maybe you might've misunderstood what I was trying to say. How this soft blocking works is that you as an admin can block an instance for all your users, just like before. The only difference is that users who really do want to can unblock it, but for only themselves. So if an admin soft blocks an instance then that instance will be blocked for almost all of the users, except for a few ones that decide to unblock it.

So this would still allow admins to moderate and create a safe space for the users, but at the same time users who choose to can still communicate their friends. So I think this would satisfy both their needs, the admin and user is happy because they all get what they want.

Unknown parent

Rokosun

Unknown parent • 3 years ago from Fedilab •

@Émilie Fecteau
Which problem does it not solve?

If I'm an admin, and someone's instance isn't moderating itself well, I should be able to block it. If I don't block it, I will see more trolls interacting with me and everyone else who is on my instance.

This problem is solved because you can block the instance and then trolls can't interact with you or anyone on your instance. Maybe if a few users manually unblock the trolls then they'll be able to interact with those specific users, but that was their choice to unblock, so they should be fine with it I guess...... 🤷
In most cases users aren't gonna unblock trolls, they'll unblock things only when there is a reason to do so, such as loosing communication with one of their friends maybe.

I don't really want anyone on my instance interacting with people on an instance I've determined is poorly moderated. That sends the message that

@Émilie Fecteau
Which problem does it not solve?

If I'm an admin, and someone's instance isn't moderating itself well, I should be able to block it. If I don't block it, I will see more trolls interacting with me and everyone else who is on my instance.

I don't really want anyone on my instance interacting with people on an instance I've determined is poorly moderated. That sends the message that what the other admin is doing is okay, which it isn't.

Again, admins are able to moderate here, we're not removing any of the moderation tools. When it comes to soft blocking, only a small number of users who manually unblock these instances can still interact with them. So except these few users who explicitly choose to see these, others will be safe from these bad instances.

And if you think an instance posts illegal content and want to do a hard block, then that option is still there for the admins. So there is really no downside for this.

If the poorly moderated instances are blocked by all of the well moderated instances, that tells their admins that they need to do better, or else no one will federate with them.

This sounds good in theory, but in practice these kind of things usually lead to censorship. When it comes to federated networks, servers with the most number of users will have more power over other smaller servers. Think of email here which is a federated network like fedi, I've heard multiple people complain that their self hosted email servers got blocked by Gmail and other big servers, and so now they can't communicate with anyone there. Mind you, Gmail is banning people left and right, without putting much thought into it, so you don't really have to be a bad actor to get blocked. A similar issue is present in the fediverse because most users are on a couple of popular instances, and if you got banned by these big servers then you're pretty much out of fedi.

Dr. Percy reshared this.

in reply to Rokosun

Bill Statler

in reply to Rokosun • 3 years ago • •

Hello @roko@social.trom.tf -- I like your idea because it adds another option for moderation, and extra options are (usually) a good thing. I'm not a programmer, so I have no idea if it's feasible to implement it.

I don't think there's one single correct way to handle moderation. Some users will want complete control over who they block, and others will want to outsource part or all of this work to an admin/moderator. This is a matter of personal preference. And some admins will want to protect their community of users, and others would rather just block the illegal and legally-risky stuff. Your suggestion provides another option for everyone.

Given that your idea doesn't exist in software at present, I wonder if there's an easy way to share suggested blocklists, for either admins or individual users to use. Since everyone has their own idea of who to block, a single centralized "Fediblock" list might be doomed. But maybe just a Fediverse group open to admins, but readable by all, where someone could post their list of who they're blocking and why.

@Rokosun

like this

in reply to Bill Statler

Rokosun

in reply to Bill Statler • 3 years ago •

@Bill Statler
I'm glad you liked my idea, on this post I mainly focus on the concept of "soft blocking" that I came up with, but we've talked before about using blocklists similar to ones that are used by ad-blockers. You can read about the previous discussion here - social.trom.tf/display/dbc8dc4…

Now the idea is to use a git repo for maintaining such blocklists so that anyone can make contributions to it. And its not just simply sharing your list, but also categorizing it into different types of content, so if someone wants to block nudity then they can just subscribe to that list and it'll get blocked, if someone don't mind seeing nudity then they can just simply not subscribe to it. There can be a list for bots, spammers, nazis, etc......
And just like on ad-blockers, this doesn't have to be a centralized blocklist, there can be many lists like this that users can subscrib

@Bill Statler

Bill Statler likes this.

Unknown parent

Rokosun

Unknown parent • 3 years ago •

@Émilie Fecteau
I felt like I should make a separate post about the social impacts of moderation, and I've addressed your point there - social.trom.tf/display/dbc8dc4…

Dr. Percy reshared this.

Unknown parent

Rokosun

Unknown parent • 3 years ago •

We really shouldn't be trying to turn fedi into a peer-to-peer network.

We're not trying to make fedi into a peer-to-peer network, in fact I don't think its technically possible.

Having admins, moderators, and instances with formally written policies has its own set of strengths that I think you're glossing over to make your case (confirmation bias)

Okay, then explain some of the advantages that you see with moderators deciding for the users as opposed to users deciding for themselves. You're not giving me any information about this.

Social pressure is the only way to keep bad ideas at bay

One of the main points of that post was to understand how difficult it is to recognize bad ideas from others. People usually just go with the social norms and don't realize what they've been doing was harmful until much later, ideas like slavery, sexism, etc. are some examples of how we've been wrong about what is good and bad, and I'm sure it'll happen again.

Unknown parent

Rokosun

Unknown parent • 3 years ago •

What are you saying that I should be acknowledging?

Unknown parent

Rokosun

Unknown parent • 3 years ago •

I'm intentionally not giving you any information. It's not going to help.

I'm sorry, but I don't know how to understand someone's point of view without them explaining that to me. I don't even know what you want me to understand or acknowledge, lol.

We are having an ideological argument right now. People always disagree on these.

If people disagree so what? This doesn't mean you shouldn't be explaining your argument. If you don't tell me then I'm not magically gonna "get it" and come back saying you were right.

Unknown parent

Rokosun

Unknown parent • 3 years ago •

@Émilie Fecteau
Oh, that makes more sense. Now I at least understand why you're not telling me anything, even though I don't think I'm gonna get your point.

in reply to Rokosun

Rokosun

in reply to Rokosun • 3 years ago •

I actually think if I did engage with you, you would just dig in more to what you already believe. 😅

BTW, if you really do have a good point and explain it well then this won't happen. Also its not like its a competition where we're trying to make the other person believe us, its more like we're explaining our point to others, that's why I made a separate post about it, because I wasn't really trying to change your mind, but I just wanted to explain about the social impacts of moderation so that others can read my post and share their thoughts about it.

Unknown parent

Rokosun

Unknown parent • 3 years ago •

I just sent you a message about it, basically if you look at this like an argument then its not gonna work. I don't like having arguments to make the other person change their mind. Especially when I write about something online, that is for explaining some things and having a discussion about it. Did you see the conversation me and Tio had below that post? Did you see the replies of others below this post? That is how you have a proper discussion.

in reply to Rokosun

Rokosun

in reply to Rokosun • 3 years ago from Fedilab •

@Émilie Fecteau
And you also made some good points earlier, and that's what initially gave me the idea to write that post.

Unknown parent

Liwott

Unknown parent • 3 years ago • •

If my friend is on a poorly moderated instance, I want to get them off of that instance, not make an exception to my block just for them.

If you want to pressure your friends into leaving (what you define as) bad instances, you still can decide to not unblock them, except you have to take responsibility for that.

there is no pressure on admins to actually moderate their instances

There is still actually the pressure related to discoverability. There is also the question of what happens to boosts.