So, here's a problem I have with Mastodon: let's say I make a post and someone replies with a racist, sexist, homophobic or transphobic comment. I can block the commenter, but that only hides it for me. Other people who come to my page will see the comment, and believe that I tacitly condone that behaviour. I'd like to be able to delete the reply from my replies list entirely. Or at least hide replies from blocked accounts. And, yes, I know that wouldn't delete it from the originating server.
I should also say, as always, I really respect the work that the Mastodon team has put into this platform. Mastodon breathed life into the ActivityPub spec, and made a working social web with a lot of loosely-defined wording. We would not be where we are today with the fediverse without the Mastodon team bridging that gap.
@evan it’s the fundamental problem with the idea that a reply to your post is not a reply on your post, it is literally someone else’s post that is linked to yours. Which causes so many problems.
I guess unlinking them should be made possible in some capacity—you can’t delete someone else’s post, but you can make it not-findable under yours.
Yes, certainly. Something you use only when a very deliberate choice. There may be some additional aspects as for whom the block notification would be visible.
@evan While deleting racist replies is a valid incentive, I don't think that OP should be the one to do it. We already have server moderators precisely for cases like these.
In fact, the mere existence of censorship tools available to OP will inevitably lead to bad people posting bad (but not rule-violating) stuff, without anyone being able to correct them. Think of people sharing BILD content and posting populist statements to influence public opinion – no one could call out their nonsense.
@evan you can post a public reply to such comment, state that you completely despise the contents of it and that you’ll block and report the person.
This way anyone reading through your page will see that you don’t condone that behavior.
@chris@evan It's not the same thing. It's pretty well understood that the comments of other users are their thoughts, and not representative of what the OP believes. That's not the case with Twitter ads, which is explicitly a sponsored promotion. We also know that companies have some say as to which content their ads appear with. We're talking about apples and oranges.
@slcw@evan I disagree. If a commenter comes along and starts getting into an argument that includes hateful speech or something else, whether its directed at me, or even another commenter on the reply thread, and I leave that up. Then that starts to taint how I look, especially if I don't have the time or resources to interject.
@chris@evan But there's no mechanism for removing a specific comment from the thread. Additionally, I haven't seen any evidence or empirical data that suggests people are erroneously associating a user's comments with the OP. I think it's important to establish the existence of a problem before assuming there is, in fact, a problem that needs to be addressed.
@evan The challenge with this is that it can also be used by bad actors. Let us assume that we roll out the feature you have requested.
I am a bad actor spreading disinformation. You reply with a link to the corrected information. I delete your reply therefore ensuring that nobody who follows me can have their bubble pierced.
Better yet, I can reply to you and then Block some relevant replies. Leaving only enough to make you look like a bad actor.
@evan yeah, imho jwz had the right idea: replies to a post should be a timeline/collection, curated by the author, hosted with the the post...
"inReplyTo" already allows thread-chaining for clients that want to assemble a thread view, AP already has excellent timeline/collection support, all it needs is a field on the post (like "replyCollection"?) to indicate the id of this collection.
this could be a short FEP. i'd just want to try implementing it in my server first to make sure it'd work.
@evan You’re probably already aware of the weaponisation of replies that are only visible to the recipient, the sender, and the sender’s network, as an abuse vector. It sounds to me like related issues, the delinking getting propagated through the network.
@evan Sorry, I shouldn’t have said private. I had undersood that if someone does a reply that’s restricted to their followers, this is used for harassment because their like-minded followers see it but no-one else. Your description of a de-linking seems to me to be a way to reduce the impact of that too.
So, our only recourse at this point is to block the person, delete our post, and then repost with a warning (not a link) about the person we blocked's prior behavior?
@KatKimbriel I think that's it. Another option is reporting the account and hoping it gets deleted. Lastly, you can reply to the post saying they're blocked or whatever, and then block them. It at least lets people know you're trying.
@evan i think the normal response would be to report it, and the server it originates from should ban them, and if they don't, this server should get limited federation, for being badly moderated, but yeah, it's not an immediate response, and it'll certainly frequently fail.
But on the other hand, the ability to delete any response you don't like might be a bit much.
@tshirtman why? I can do that on blog comments. It wouldn't delete it from the replier's server or anywhere else; just from the page that shows my original post.
@evan to me it seems normal to be able to expect contradiction on social media, as long as it stays civil, it feels like hiding/removing content is a moderator's job.
@evan Yeah, that's been a consistent ask for a very long time, as I understand it. I think fep-7888 (and fep-5624, revised to apply to the context field as described in 7888) gives the most robust and reliable mechanism to federate that kind of reply moderation.
Somehow you make it sound like a post you make belongs to you (as in property) and you should be able to control who can respond to it.
Let's act as adults and realize that once you make a public statement you no longer "own" this statement, it is there in public domain, people can agree/disagree, adopt, or reject it, or even modify it and restate the original. It is what we want, isn't it?
Every object has a replies collection. What we need is:
When showing the replies to an object, use the replies collection -- not a search for inReplyTo or ostatus:conversation, which is what Mastodon does right now.
Let the user add and remove objects from the replies collection.
(Optional) Notify addressees when a new object is added to the replies collection (Approve or Add activity)
@funbaker "unpack". Meaning, "explain in further detail". I posted a problem I have, and you said "There's no problem with each person be responsible for this on their own tho." Like, I think there is a problem, which I described, so I don't know why you think there isn't.
@evan Blocking that Account on your End is fixing this Problem. For you.
But you can't fix this Problem for others, because they may think differently about it. No matter if right or wrong.
And you may not change them either by removing Posts.
But I also can be interpreted as being overpragmatic and strictly solution-orientated in how to make the world a better place, and I also may have a more resilient view on things other people post, so there may be a twist to it.
@evan I think “disassociation” might be an interesting approach. Let the comment stand in isolation, unlinked from your post as a response. If not the first time this has been posited, I’d welcome thoughts as to why it wouldn’t work.
@nyquildotorg I’m assuming that replies have some linkage to a “parent” post where stand-alone posts don’t. If true, just “unlink” from the parent being replied to, and it stands alone. Consideration should be a combination of architecture capabilities (my assumptions may be wrong) as well as spirit/morality. Others may weigh in on why, even if technically possible, it shouldn’t be implemented in such a way.
@kevin_hardiman oh, I see what you mean. This already naturally happens when deleting a post; all replies to it are "disconnected" and this is something people actually do. I thought you were pitching that as a solution.
The challenge with what you're proposing is the post is "owned" by someone on another instance, and the local instance has no power to disconnect it. The "connection" is (currently) handled when creating the reply, as a part of the metadata attached to it.
Coming up with a mechanism to supply that power is what Evan and jwz were saying is desperately needed. So I think we're pretty much on the same page here
@evan It's one of these classic, fundamental problems in decentralised networks ;) Who "owns" the list of replies? I would say the author of the starting thread gets to maintain the list of accepted replies and this list propagates across the fediverse. Others will say that at max a flag should be added with a qualifier and that other instances should be free to decide if/how they show/hide flagged replies. But who gets to add flags? Just the thread "owner" or any reader? Etc.
@evan I'd favour the first approach. When you load a post from me (be it an original post or a reply I posted to someone else's post) you should receive a list of replies I accept. And I should be allowed to define what that list contains. But it's a complex issue.
I will admit, for threads that I think could attract argumentative people, I make the last post something slightly incendiary so I can do exactly what you describe without reducing reach for the thread more broadly.
@evan could you say more about that? My only point of reference is what’s implemented in masto. What’s in the protocol that could be implemented differently?
@evan update: I’ve gleaned the basics from your replies to others - thanks. Also boy do a lot of people want to explain AP to you. How do you stay so calm?
@fraying The protocol-as-implemented is different, sometimes a lot, from protocol-as-designed. I depend a lot on people telling me how things really work.
@fraying well, actually, not entirely. We have a part that says that there's a collection of replies, and we have a part that says you can edit any collection that belongs to you. It doesn't say explicitly that you can edit that particular collection.
I don't know if this is just Mastodon's bizarre way of organizing comment threads, or what, but I'm really not seeing anything about better ActivityPub implementations in this thread, and keyword search for "Friendica" isn't turning anything up.
I saw some stuff about Cohost and Post, but I don't know if those are AP?
Any chance you could link to the relevant portions of the thread?
Sorry for being unclear. I meant that doing ctrl-F on your thread wasn't turning up anything about Friendica. Was wondering if you could provide links to any toots about AP implementations with better author control over replies.
@jwildeboer it's not a fundamental problem; we've always made the OP responsible for the list of replies. That's why there's a replies collection in every ActivityPub object.
@evan interesting. So if a server allowed the activity's actor to remove items from the replies collection that would be considered appropriate in terms of compliance with ActivityPub? @jwildeboer
@evan 100% agree. If we all pick up the dog poop off of our front lawn, the entire neighborhood benefits.
Bonus points for a system that summarizes the top posts that have been booted by users to the admins so that they can consider booting the users/instances from the admin level. This way the community is helping the admins do their job.
@fraying@evan
The problem is that we can't have a feature that "removes poop", because it would mean that poop is automatically and unquestionably identified.
What we are discussing here is merely a way to hide or orphan replies that one does not like.
If this feature was introduced, a fascist may hide the replies of antifascists to his posts, a misinformation actor may hide replies that debunk his claims etc.
Reply and report (and block, eventually) is preferable and already in place.
@Eh__tweet@evan nobody was talking about an automated process. Everything we’ve described has humans making decisions. And everyone here already knows about report and block. The tools are insufficient.
@fraying@evan
I did not mean an automated process, my point is that what is bad for me and you is not bad for someone else. So we can't have a feature to "hide only the bad replies", because the system can't decide what's good or bad. We just can have a feature to hide replies one doesn't like. And this feature would also be available for fascists, racists, homophobes, propaganda agents etc. It would strengthen their communication.
if Mastodon treated all replies like reddit does (giving us votes but not boosts or favorites), we could down vote this stuff into oblivion.
I really want to see Mastodon and lemmy or kbin merge. hashtags function like subreddits. eventually all content should be a POST on the fediverse. all this combined functionality would make us better than any closed platform could be. the #fediverse should be the attention layer for the open web.
@evan A simple solution would be to reply to the offensive post saying "I do not condone this and will now block your account" or similar. Then block. Presumably your reply will remain visible on others' timelines.
It wouldn't have occurred to me to blame you for someone else's behavior. But if you really want to make it clear that homophobic reply guy doesn't speak for you, you could reply back. You could mention that you're both reporting the reply to the moderators, and blocking the person.
Add comment