Mastodon's Mastodon'ts.

There are a few fundamentally broken things about how Mastodon posts work that are terrible vectors for abuse, as well as being bad for basic usability. Maybe they are fixable, I don't know. To be clear: I am a fan of Mastodon....
https://jwz.org/b/ykC_

Mastodon's Mastodon'ts

There are a few fundamentally broken things about how Mastodon posts work that are terrible vectors for abuse, as well as being bad for basic usability. Maybe they are fixable, I don't know. To be clear: I am a fan of Mastodon. I have been enjoying my time there much more than I ever enjoyed Twitter or Facebook or Instagram. And I am 100% in the "I won't touch anything Jack Dorsey has ...

@jwz "I own the replies to my post" is very important. A lot of grief on the Internet happened because people didn't know they could (and should) delete jerks from their own posts (regardless of platform), rather than give up on comments ever being useful.

@isagalaev @jwz
Holy shit it is so nice to hear people say this aloud. I've been quietly over here in my corner thinking about launching my own entire rant to this point wondering if I was the only person who got this.

I find Mastodon's complete failure to grasp/implement this principle both maddening and incredibly ironic. This has been, until quite recently, the number one problem with Twitter (and similar systems like G+), and what caused it to be such an utter shit show. Mastodon's failure to allow users to own their comments and provide them with moderator tools reconstitutes exactly the toxic dynamics of Twitter. Mastodon has tried to compensate for this with a model that has the people running instances functioning as moderators, which is a "solution" of such terrible long-term consequences as to be indistinguishable from active sabotage of the Fediverse.

@siderea @isagalaev @jwz Yeah. While i personally think that "deleting other people's posts" might be a bridge too far, at the VERY least you should be able to SEVER THE REPLY-TO RELATIONSHIP of any post you don't want in your thread. People can say shitty things if they want, but you shouldn't have to automatically boost the reach of that just because it's a reply to something you said.

@adrienne @siderea @jwz "i personally think that "deleting other people's posts" might be a bridge too far — this is exactly the problem I was talking about. Too many people adopted this… reverential approach to something someone else wrote, which just shouldn't exist.

As jwz mentioned in the article a person coming to *your* comments effectively uses your platform, of which you should stay in control. If they want to be in control they should write their own post, mentioning yours. (cont…)

@isagalaev @siderea @jwz Yes, i agree, but someone replying to your post kind of IS "writing their own post, mentioning yours". It still lives on their server! It's just that "mentioning" creates a relationship (reply-to). Being able to sever this relationship at will would allow people to maintain their own content which they did, after all, post on their own fucking server -- but removes it from association with your content.

(FYI, i'm on the mod team of a fairly popular Fediverse server which is extremely full of queer & marginalized people; i'm not stupid, i'm not inexperienced, and i am certainly not "reverential" about shit other people write.)

@isagalaev @siderea @jwz And i do, in fact, ACTIVELY RESENT the implication that i didn't read and thoughtfully consider JWZ's post (which i have in fact just posted to the Discord server for our community, noting "these are great ideas!")
@adrienne P.S. I didn't imply you haven't read the article!
@adrienne I wasn't trying to attack you personally with "reverential". I was describing a position I refer to, which your words reminded me of. Sorry if that was unclear!
@isagalaev Thanks! Sorry, i'm a little on edge myself!

@adrienne

> but someone replying to your post kind of IS "writing their own post, mentioning yours".

That's the problem: it is, but it shouldn't be.

This isn't just a technological problem, it's a cultural one.

The Blogosphere has a very different culture than Twitter does. It has very clear boundaries that Twitter and many other "social media" platforms don't share. On the Blogosphere, whether or not you have jurisdiction over something you write depends on where you write it. If you didn't want to give somebody else authority to delete it you shouldn't have written it as a comment on their blog. You have your own blog where you have the authority and nobody else can delete what you write.

Mastodon is implemented like Twitter, where there is no distinction between different people's spaces, such that it makes sense to conceptualize a comment a user leaves on another's post as "theirs". After all, all of the comments each of us writes appear on our comments tab.

@isagalaev @jwz

@adrienne

I think this is a terrible mistake. I believe, strongly, "good fences make good neighbors".

One of the beautiful consequences of this stark doctrine, where everyone has their own space and if you go into someone else's space to leave a comment there you do so knowing they have complete authority to delete that comment, is that it puts huge social pressure on people to mind their manners. If you want your comment to remain up - or even appear in the first place, given premoderation - you need to make it acceptable to the person you're addressing it to.

The fact that no such pressure exists on Twitter explains an awful lot of why Twitter is like Twitter is.

@isagalaev @jwz

@siderea @isagalaev @jwz I mean, we're not going to agree here, i don't think -- which doesn't surprise me, since while you probably have no idea who i am, i know who you are, and we've been in disagreement about a great many things over the course of the last 20+ years of the internet. But i do have a lot of moderation experience, myself, and i feel qualified to disagree here.

@siderea @isagalaev @jwz Yes but from their perspective they are, in fact, writing it in their own space -- it lives on their server! I agree with you that the expectations and affordances here are cultural as well as technical, and, again, I AGREE WITH JWZ ON THE NEED TO BE ABLE TO KEEP PEOPLE FROM SHITTING UP YOUR CONVERSATIONS.

But frankly, i absolutely do not want people being able to send a request that deletes my own shit from my own profile on my own server! And i think there are solid ways to approach the issue of "make this more like a blog-space where the original poster has priority" that do not have that side-effect!

@adrienne @siderea @isagalaev Under my proposed design, their perspective that they are writing in their own space is simply wrong. They are not. Posting a reply is asking the author to consider adding their reply to the metadata of the post. If that's not what you want, make your own top level post.
@adrienne @siderea @isagalaev Likewise, characterizing deleting a comment as "deleting a post from someone else's server" is not how this works. If I make a post, and my server sends a copy to yours; then I edit the metadata to correct a typo; then my server sends an updated copy to your server, which updates it; what gives me the right to do that? Because that's how the software works. Because that's how database replication works. Because that's what everybody wants.
@jwz @adrienne @siderea @isagalaev I love this framing. Pair this with the following assertions and, baby, you got a stew going:
1. Public-key cryptographic verification ensures parent poster cannot modify content of replies / impersonate other users, only remove.
2. Deleting a reply should generally notify the target of moderation, giving their client an opportunity to handle as desired (e.g. deleting copy from their timeline; reposting as a top-level quote-toot)
@siderea @adrienne @isagalaev @jwz It can be made into a technological one. I've been working out how you'd do a real decentralized (no coinbro shit) version of the fediverse, and "replies don't show up unless they're accompanied by proof of authorization to reply" is an ingredient. Authorization would normally be by assigned proxy authority (something like "well moderated instance or known friend automatically gets authorized but revokable).
@dalias @siderea @isagalaev @jwz Well, i mean, i also do in fact want a way for posters to explicitly sever a reply-to relationship. I want affordances for that specifically. Like "this reply in particular, i don't care if it comes from a trusted server, i do not want it attached to my post, so boom, it is not attached anymore." That's still a technical problem, but a different one!
@dalias @siderea @isagalaev @jwz The frustrating thing about this entire conversation is that i AGREE with the points JWZ is making about conversations and replies! I think they're really important!
@adrienne @siderea @isagalaev @jwz I can't see them. Not sure if I'm blocked or what..?
@dalias the article's over here at his blog too: https://jwz.org/b/ykC_
Mastodon's Mastodon'ts

There are a few fundamentally broken things about how Mastodon posts work that are terrible vectors for abuse, as well as being bad for basic usability. Maybe they are fixable, I don't know. To be clear: I am a fan of Mastodon. I have been enjoying my time there much more than I ever enjoyed Twitter or Facebook or Instagram. And I am 100% in the "I won't touch anything Jack Dorsey has ...

@adrienne @siderea @isagalaev @jwz Yes. That'd be a revocation whereby, if an instance were still trying to show the proof of authorization to reply after having been presented with proof of revocation (both notarized to sequence them), you'd have cryptographic proof it's a malicious instance and it would automatically get blocked.
One way to look at it is that whoever creates a thread should have control over who can participate and on what terms. With that framing jwz's model is saying "you can only reply to this thread if you give me permission to delete your reply if I want". So it's a complement to other useful limitations on who can reply -- that also aren't implemented here and really should be. No question that Mastodon's current behavior is broken, whether you look at in terms of the Twitter model of "each tweet exists on its own but people can control who can reply and which replies are shown" or the forum/blog/Facebook model of "replies only exist within the context of a thread, and whoever created the thread can delete them".

Then again, when OP make their right to delete replies a condition for replying, under what circumstances will the increased visibility of posting as a reply be seen as outweighing the risk that the original poster will get annoyed at you and delete your reply as well as blocking you, or that your reply will be collateral damage if they decided to burn it all down? It'll be interesting to see what happens the first time a guy with a big following throws a hissy fit and deletes all replies to a post that's gotten a lot of traction -- or worse, deletes all the replies to all his posts before ceremoniously flouncing off.

Of course, a lot depends on the implementation, but it wouldn't surprise me if a lot of people will choose to make new top-level posts instead of replying (as happens on Facebook and forums, although blogs resstrict who can make top-level posts). So my guess is that if things go that route we'll see more quotes and pseudo-quotes (screenshotting or cut-and-pasting and including a link). Which isn't an argument one way or another, just something to consider.

@dalias @adrienne @siderea @isagalaev @jwz

@jdp23

Point of order, "although blogs resstrict who can make top-level posts": no, blogs do not restrict who can make top-level posts. Anyone can make a top level post. *On their own blog.* Which anyone can go out and get, for free, on a variety of different platforms.

Some platforms, most notably LiveJournal, enforced all users having their own blogs (journals). Unless a poster allowed anonymous commenting, you couldn't comment at all on somebody else's posts except by signing up with LJ, which came with a side of your own blog to use or not as you saw fit.

Not to pick on you in particular, but it's been really fascinating how this entire discussion has really exposed a kind of digital provincialism, where people who are only really familiar with some specific platforms, perhaps Twitter, really have no idea what the dynamics, affordances and social norms are in other parts of the internet.

@jwz @isagalaev
@dalias

@siderea @jdp23 @jwz @isagalaev @dalias do you envision it that if i comment on your post and you delete it, i still have a instance of my comment as part of my own time line? so my followers could see it, but yours wouldn’t

Or is the only instance of my comment part of your timeline so deleting it would erase all copies of the comment?

I don't have a strong opinion, just curious how folks are thinking about what their ideal scenario would be for this hypothetical, but good, change in behavior

@rdp

> do you envision it that if i comment on your post and you delete it, i still have a instance of my comment as part of my own time line? so my followers could see it, but yours wouldn’t

Where do you think your replies appear, right now?

Have you been perhaps confused by how when you make a reply to someone else's comment it appears in your timeline *to you*, when you're logged in? (I have no idea why Mastodon does this; it's misleading.)

For instance, this comment of yours, that I'm responding to right now, does not appear in your timeline *to anyone except you*, and is not pushed to your followers.

People who go to your profile page and click on the "With Replies" tab there can see it, regardless of whether or not they follow you.

Your followers do not know that you are participating in this conversation at all, unless you boost it to them.

Your replies to comments already do not appear in your timeline as far as anyone else can see.

@jdp23 @jwz @isagalaev @dalias

@rdp

Now, if you know all that, and you're just really really concerned that any comment you make has a permanent record under your "With Replies" tab, and that *you* see it in your timeline, I have no problem with that. Might be a pain to implement if we go with the metadata model that @jwz proposes, but I don't have an objection to that.

But what we're proposing is that if you leave a comment in reply on someone else's OP and they find your comment objectionable or undesirable for any reason whatsoever, they get to remove it from that context. Which is the only context it has so is pretty much coterminous with deletion, the same way if an editor removes a passage from a book that they then publish but saves a draft of the removed material on their local machine we generally refer to that as a deleted passage.

@jdp23 @isagalaev @dalias

@siderea @rdp @jwz @jdp23 @isagalaev That's not the only context. Arguably the main context it's seen in is the home feeds of all your followers. Would it disappear from there too? Would clicking on it there cease to expand with the context of the post it was made in reply to? This is a primary vector of unwanted contact and motivation for desire to restrict replies.
@siderea @rdp @jwz @jdp23 @isagalaev My view is that nobody should ever be able to delete your writing, but removing it as a reply should unlink it in both directions, and that blocking should perform this kind of unlinking on all past interactions.
Other potential contexts include the local feed on your instance, hashtag searches, federated feeds on instances that federate with yours, and search engines (for people who haven't opted out) (well actually also for people who have opted out but that's another topic).

How far do you see the two-way blocking going? For example if my reply includes a link to the post I'm replying to, does blocking mean that link is no longer traversable to people who get to my reply from some other context? Of course there's certainly a lot of value just in straightforward unlinking even if it doesn't deal with these more complex situations, so I'm not pushing back on the idea; the current "solution" here is horrible and needs to be improved. But Other potential contexts include the local feed on your instance, hashtag searches, federated feeds on instances that federate with yours, and search engines (for people who haven't opted out) (well actually also for people who have opted out but that's another topic).

How far do you see the two-way blocking going? For example if my reply includes a link to the post I'm replying to, does blocking mean that link is no longer traversable to people who get to my reply from some other context?

Of course there's certainly a lot of value just in straightforward unlinking even if it doesn't deal with these more complex situations, so I'm not pushing back on the idea; the current "solution" here is horrible and needs to be improved. But like I said, the details of the implementation matter a lot, so it's important to do this kind of analysis of any proposed solutions.

@dalias @siderea @rdp @jwz @isagalaev

@dalias

> Arguably the main context it's seen in is the home feeds of all your followers.

No it's not. Do you think that all of your comments are being pushed to your followers? That is not presently how Mastodon works.

This comment of yours I am replying to right now, your followers have not seen it, not unless you've boosted it to them. Your followers have no idea this conversation is happening.

@rdp @jwz @jdp23 @isagalaev

@siderea @rdp @jwz @jdp23 @isagalaev They don't see it, but they see my reply (this post). And if they expand this post, they're instantly able to see and interact with your post I'm replying to.

I'm saying that "unlink reply" should probably be two-way and not provide a low-friction path for my followers to interact with you through my reply if you considered my reply unwelcome.

@dalias

> They don't see it, but they see my reply (this post).

Nope. Not unless you boost it.

Don't take my word on this, go check.

@rdp @jwz @jdp23 @isagalaev

@isagalaev @rdp @jwz @jdp23 @siderea @dalias to add just a bit more pedantry: I saw this conversation in my feed, but that’s because I follow both of you.
@chrisamaphone @isagalaev @rdp @jwz @jdp23 @siderea I just checked from another account and @siderea is right, at least on that Mastodon instance. I'm quite surprised and this will likely change my usage patterns. But I would assume this behavior is specific to the follower's AP software (Mastodon) and some others might choose not to hide replies..?

@dalias

Over at my place, we're having a hell of a conversation trying to figure out under what circumstances - because apparently there are some - a reply to someone one follows will show up in one's home feed, and, I got to tell you, we're not getting a lot of traction on the problem.

At this point I need to invoke my two favorite backup explanatory hypotheses:

1) There's an RNG or equiv involved.

2) The protocol is so complicated/flexible, we should not assume the different implementations will do the same thing, simply because it is too hard for anyone, including developers, to understand correctly what it's supposed to do.

@chrisamaphone @isagalaev @rdp @jwz @jdp23

#2 is certainly the case. It's complex in ways that are, at least for me, very unintuitive; and compatibility testing is very ad hoc. Also there are non-obvious interactions. For example, turning on authorized fetch disables inbox forwarding, does that play into it? Who knows, I wouldn't be surprised either way.

@siderea @dalias @chrisamaphone @isagalaev @rdp @jwz
@chrisamaphone @isagalaev @rdp @jwz @jdp23 @siderea @dalias and I saw this, because I follow three of the seven of you...

@siderea @jwz @jdp23 @isagalaev @dalias makes sense.

yes, i was thinking about people still being able to go to your account and see the reply that had been deleted by the original poster.

seemed like a good middle ground of the OP being able to moderate what happens under their post, and being able to post as “i" please.

the main goal is for the OP to have better control of what their followers see and experience.

It's true, there are a lot of different variations that I didn't discuss in my post. LJ/Dreamwidth are certainly a potential model for fediverse interactions. Similarly, Indieweb blogs' POSSE approach means that deleting a blog post doesn't actually delete the comment. And the Friendica/Hubzilla/Streams family of fediverse platforms is closer to the reply-in-context-of-thread model but has its own quirks. But my post was long enough without bringing all those up, and I don't particularly think they change my key points, although of course reasonable minds may differ.

And yeah, in discussions like this, people tend to project their own experiences and knowledge as some kind of global truths (like the person who thought we should be using Usenet as a model). Worse, so do the people actually doing the system design and implementation. So it's always fascinating to see how it plays out.

@siderea

@dalias

Oof. I appreciate this line of thinking, but please make sure that you understand how flexibly the blogosphere handles what are basically ACLs because I think that diversity of options is important. One of the things that's kind of awful about the Twitter model that Mastodon has inherited is it's one size fits all.

Off the top of my head, users might choose to default allow comments from:
1) manual list of authorized commenters
2) users on own instance
3) users on known good instances
4) users they follow
5) users that follow them
6) users that are followed by someone they follow (FOAF)
7) users that follow someone they follow (fellow fans - I don't know I've ever seen this implemented)
8) users on any instance
9) anons? (If that is even possible)

@adrienne @isagalaev @jwz

@siderea @adrienne @isagalaev @jwz Yes, that's exactly what I mean. Ultimately the proof would be thru a program signed by you to evaluate authorization, which could use any of those criteria. Proof of authorization is an input (involving other signatures & notarizations) that your program accepts.

@dalias @siderea @adrienne @isagalaev @jwz I think from a preventing abuse standpoint that makes a lot of sense. On the other hand, I've seen plenty of threads on Twitter where the reply space is a way to organize righteous opposition to oppressive proposals or hateful statements by authority figures.

I know it's not the same but it makes me think about the original Mastodon lack of expansion of linked toots. It was intended to discourage abusive behavior but marginalized people pointed out it actually prevented necessary criticism.