Bluesky has reply-gating (you can set who can reply to a post, like people you follow or a given list or no one) and is now testing out post-publication reply locking.
I just want to yell for a second about how humane and consent-forward these features are, especially after seeing some people here losing their minds when someone asked for gating recently because they felt (alas, not a paraphrase) entitled to always be able to respond.
[ETA sorry gotta mute thread to save my brain ✌️]
Believing you are morally entitled to reply directly to someone because they speak in public is…definitely a position, I guess, but I think being able to specify how you want to be interacted with is deeply humane. (And being able to do it after a thread starts to go sour is so important.)
Someone will ask why blocking doesn’t make this unnecessary and the short answer is that prevention is better than picking off unwanted interactions one by one, and having to process each one as you do it.
Someone else will say that if you don’t want everyone to be able to say whatever they want in your thread whenever they want to say it, you shouldn’t post in public because that’s the bargain.
But like…I don’t want the old bargain. The old bargain is kind of shit! We have a million ways to contain and shelter even “public” conversations offline, let’s have that in the good online places, too.
Anyway the big fedi research project I’ve been on for six months is nearly ready for publication and there’s so much I’m excited about in it.
Having Mastodon be so central to the fediverse by the numbers means we could really use attention to these safety/humane conversation/good experience features here. I hope that happens.
@kissane @alter_kaker That is exactly what I was thinking about when thinking about reading and writing access on this subject earlier. Wasn’t sure if it was both too mainstream and too old to cite, though…
And having seen LJ snark communities in action…yes, it can absolutely be abused. Bur so can group chats; to stop a group of people from being dicks ever you have to get rid of the people, first.
The other danger of drawing from LJ is pulling in roleplaying accounts.
@kissane @alter_kaker yes I miss that granularity of post lists.
I miss Livejournal. It was great being able to decide who could reply or even see my posts.
I liked the group accounts as well.
@kissane Hypothetically, I can see some narrow issues. Using lists of people to block before saying something derogatory or hateful so they can't respond for example.
Thing is, this is one of those things we can't predict how it will work out until it's deployed. Maybe it will be wonderful, maybe it will be awful in some way we wouldn't have predicted. Hopefully it turns out good, having more tools to moderate your web experience is a welcome thing.
Thank you for telling us about this development!
@kissane looking forward to it.
The feature I really miss from old LJ days is pre-moderation, that is, the ability to approve comments (and thus, possibly, reply to them).
@kawazoe @cek You’re raising good questions, but not showing a fundamental incompatibility, just questions. We can answer those questions and give it a shot.
Will the answers be wrong? Maybe, and we can change them if they are.
Will adding this be janky given the various servers that would need to support it? Of course, but even likes/favorites are janky right now, not to mention more fringe features like emoji responses, quote-boosts, etc.
One of the strengths of federation is the ability to experiment. I try not to discourage experimentation simply because I, personally, can’t see things working out. Others can see further than me. That includes, for example, people who have spent six months collaborating on a research project, and who were already extremely knowledgeable about the fediverse and social media more broadly before engaging in that research project.
@kawazoe
The feature will work in the same way as "followers only" posting mode. I haven't read the proto, so I'm afraid I'm just guessing, but this shouldn't be that hard to implement.
You mentioned the notion of trust, but it's the cornerstone of every messaging system, you can't live without trusting someone or something, unless it's an anonymized board (aka cesspool).
Implementation-wise, you would have:
* sending user (which is the owner of a post) signing the post
* receiving server (or user) checking the sig
All posts would belong to users posting+signing them.
* In case there's a premoderate flag set in the post, replying user would sign, then encrypt with user's pubkey and send via DM (aka private mentions?).
* Receiving user would have a moderation queue. He decrypts the msg, then decides to "approve" it by pushing ("releasing"?) the reply comment/post. The reply post belongs to the user who posted/signed it.
@cek You could always defederate instances that don't support the feature, but then you just proved my point. It isn't possible to have a single network that supports both modes of operations. You'd end up with two completely separate federations at this point.
You simply can't have this feature on a bymodal federated network, because no amount of sig will protect you against a single instance not supporting the feature.
@kawazoe I understand the concern but I don't understand why it's the concern. You'd have the same underlying network but with extra features on those instances that support them.
On legacy instances you will have full reply chaos which will be filtered out on "modern" instances according to the [original] post flags. In case a legacy instance decides to forward a spoofed post, it will be verified against the sender, and if the sender instance is "modern", dropped on modern instance (no sig, thus spoofing). Legacy instance receiving modern message will just display a text with extra commands/sig.
"Modern" moderated (approved) replies will bear a token or a sig from the original post they're replying to.
Seems doable.
@cek you can hope that B will play nice and check with A for permissions to do this, but you cannot trust it will. You can't force it. Thus, C either have to trust B's content, or check with A for every post it receives. Congratulations; you just created a centralized system. If A goes down, C goes down with it.
This is a very difficult problem to solve in CS, and most people in the field believe it fundamentally cannot be solved. It's the same thing that breaks e-voting.
@cek Because anyone can make an instance or client that can ignore part of the spec, even if you include a signature, there is no guarantee it will be checked, and you can't just outsight stop sending data to those clients because they might be from previous version, or maybe newcomers in the field. Like HTML, you don't want to assume everyone supports JS or even CSS.
I'm not saying "I think it can't be done". I know it can't be. It's a fundamental limitation in software like databases.
@cek If this was possible to do, then email wouldn't need GPG. All databases could be ACID and scale infinitely. Blockchains would have taken over all monetary transactions. And every country would use and trust e-voting machines.
The problem is that the second you have more than one party in a system, you have to chose to trust either the communication channel OR the destination. The only way to have both is to run a centralized system, typically on a single machine.
@cek even bots like you suggested would need to sweep all instances on the network since you can't trust an instance to honor deletion requests from the federated feed. It would be borderline impossible to even be aware of all of those instances. Even if you were, you can't trust them to not just block your bot, which they are very likely to do because of the additional load it would cause on their server.
It's like asking a hacker to delete your SSN from their DB and hope they comply.
@powerllama @kissane Bluesky’s blocking features enable garbage behavior as well.
People will post incorrect degrading stuff about you, then block you, removing your ability to respond.
I don’t like other people having the power to delete my posts.
It’s an incredibly toxic place:
https://bsky.app/profile/marshray.bsky.social/post/3kv5hhqxdlz2c
@kissane @marshray @powerllama I was about to reply along that line, I think the apocablock (love that term) has its points and issues.
As Erin notes it’s more visible there because BS has more “virality” to it, which leads to a lot more “the avalanche has already started, It is too late for the pebbles to vote.” moments.
Apocablock doesn’t “delete” your posts. It removes you from my “living room”.
@Wraithe @kissane @powerllama This is an example of person re-posting me to their followers.
Then, after a pile-on of insults against me, she blocks me.
Whose “living room” is that?
@marshray @kissane @powerllama In this case it would be hers.
She retweeted a post of yours, you replied, she blocked. Conversation ends.
Your post isn’t deleted, it just breaks the thread. I’m looking at it right now using a tool.
There’s a good thread here explaining the motivations behind the apocablock: https://bsky.app/profile/legalminimum.bsky.social/post/3kunmefkdth2u
TL/DR: it’s to prevent “brigading” (a tactic used heavily by the alt-right on Twitter for anyone who doesn’t know)
@Wraithe Sure, I’ve heard that justification before.
Here the “brigading” by toxic jackholes with dumb insults was solicited by the same person who employed blocking to prevent reply.
@oblomov @marshray @powerllama This is why I actually really like the US court ruling requiring gov officials to keep replies open and not block people. Public official accounts should absolutely work differently, IMO.
I kinda feel like the cultural moment for reply-yelling to be functional for like…brands and celebrities and execs and all that is almost a decade in the rear-view now, but I agree that it absolutely worked until about 2015.
Erin Kissane
Marc Trius
Jay
Esther Payne 
Andromeda Yelton
Just Boby
Jesse
Raven Onthill
Robots can be gay deal with it
cek
kawazoe
jmelesky
Jenniferplusplus
Austin
Marsh Ray
Wraithe
Oblomov
lippyduck
Zack
Sally Strange
BenAveling
prex
Willa