What's that you say? Bluesky crossed a few hundred thousand users and immediately started having moderation issues? You don't say! Who could have possibly predicted? etc.
@seldo It is absolutely mind-blowing that they didn't include a simple block feature even prior to beta... it's a minimum viable product feature for a social app.
Jack Dorsey may not be CEO but it was his baby and he's on the committee... he should know from Twitter experience AT LEAST, if not 99999999 other forums, app chat sections and social media apps how important block is.
Not sure why so many decent folks keep saying "c'mon, give 'em the benefit of the doubt"
@seldo but Laurie, it’s decentralized*! So they have unique problems nobody else has considered, like how blocking should work when there are multiple servers involved! Totally unsolved problem across decentralized platforms. Yep, no way to implement it. Just can’t realistically be done.
In fairness to them, they're trying to build in public, and didn't expect this much attention yet.
They opened their invite-only private beta, where they hoped to control a tiny user list, before they'd even implemented blocks.
Sign-ups are protected by sign-up keys, but unfortunately, they didn't rate-limit their sign-up API, and the keys were quite short, so someone wrote a script to brute-force guess sign-up keys. So some nazis got in. 🤷🏿♂️
@mekkaokereke@marypcbuk@seldo “We’ll get to that after the first few hundred thousand invites” says a lot about what your priorities will be later when you have to do it right; specifically it says “it’ll be an afterthought that we only do as much as needed to avoid bad press”.
@luis_in_brief@mekkaokereke@seldo that doesn't feel fair; they were working on block and had already said it was planned for probably next week and folks who were invited or got in through the waitlist had a chance to make the choice of joining something missing a LOT of features. their mistake here is a badly protected signup system: that's what I'd view as casting doubt on how prepared they are to do this thing
@marypcbuk@mekkaokereke@seldo And “this is a beta” is fine, I’ve certainly launched incomplete apps before! But also, as a user, what features are in your beta tells me what features are an afterthought and which you’ll prioritize and refine later.
@luis_in_brief@mekkaokereke@seldo I think it's fair to build the features you have the architecture established to build; more of a question whether those are enough features to do a closed beta with. I'd personally give them a harder time over the failure to think about how signups could be brute forced
@mekkaokereke@seldo I normally deeply respect your takes on these things, but:
they drove hype through the usual "invite+iOS-only" exclusivity thing; it's specifically designed to drum up press and inspire FOMO.
when rolling out literally any public service one of the very first questions you have to ask is WW4D - "What would 4chan do?" You don't release unless you have good counters to the obvious answers.
@mekkaokereke@seldo the fact that Bluesky spectacularly failed the WW4D test before it even launched does not bode well for their future ability to deal with bad actors and protect marginalized populations; they clearly don't have the right people in the room for it.
@tess@mekkaokereke@seldo From their public statements, they seem to believe the right way to handle abuse and extremism is to let people mute those accounts. If nobody sees it, there’s no problem, right?
They just hadn't built blocking yet, but started their invite only beta before blocking and moderation were finished.
Mute is easier to implement. Blocking should be finished next week. Moderation controls are still in design. Their moderation model is very interesting, but it's not finished yet.
Looking at what little is out there, the whole "thresholds based on reputation" bit seems generally bad and easily gamed to spread or throttle particular speech (as it was on Twitter).
I mean interesting as in their design allows users and admins to choose and compose their moderators. As in, i can be on a server called "Cat city🐱" the admin uses a moderation service called "No dogs!" which flags all posts that mention dogs for moderation. As a user, I can also use a moderation service called "general common sense moderation" and "help staying sober"
If I decide that I don't like the "no dogs" moderation rules, I can move to another server.
@mekkaokereke@Green_Footballs@seldo does "No dogs!" prevent people who talk about dogs from seeing my posts? Because if it doesn't, then I (me; personally; Dana Fried) am going to have a problem being on the platform
@mekkaokereke@Green_Footballs@seldo to add to that, does "no dogs" protect me from people talking obliquely about canines? Does it block posts from my friends talking about how they were attacked by a dog?
This is the fundamental problem with automated filtering and individual blocks. My Twitter experience only became tolerable after:
They instituted strong global moderation policies
I had access to mass-blocking and chain-blocking technology
@mekkaokereke@Green_Footballs@seldo I spent a lot of my time in my last few years on Twitter hardening my account against hate movements; if I didn't do it actively the service gradually became unusable.
The problem was only partially preventing those people from interacting with me; a lot of it was also preventing them from seeing me or my associates.
Mastodon works for me because those people aren't allowed on here and/or aren't generally allowed to federate with my server.
I agree very strongly that no one wants to be on the social media version of big brother where you are broadcast to everyone but you don't see what they are saying about you, that is kind of worse than nothing at all.
@mekkaokereke@Green_Footballs@tess@seldo I have a hard time taking any social media company in 2023 who didn't start with how to handle the bad actors as a fundamental design choice. "We haven't built blocks yet" is another way of saying "cishet white men first, everybody else is an add-on later".
I never used Twitter and I only started using Mastodon out of curiousity, and have stayed because I like the company. What does Bluesky offer that would attract users, given the not-so-great history of Twitter abuse?
A much better, more inclusive experience (so far) for Black users. But as Dana points out, this is a better experience through curation of initial invitees, not through moderation of bad actors. The very first nazi into BlueSky wreaked havoc for days before they could be stopped. There's not a good justification for launching to real users without blocking and moderation.
@mekkaokereke@bkahn@tess@seldo Honest question:
Is this AT protocol thingy basically just the same thing as ActivityPub?
What’s their reasoning behind not just going with ActivityPub but reinventing it?
To me this feels like they thought:
„This Mastodon/ActivityPub might actually be a good Idea. Let’s try it again but this time with money in mind.“
@mekkaokereke@seldo This feels like it's unnecessarily generous to them? They've built social networks before and done this kind of rollout before. This was a very problem that they chose to ignore, that shouldn't give them a license to ignore their responsibility for the outcome.
@runewake2@seldo Probably? But "they" haven't built social networks before. The BlueSky team is 5 people, lead by Jay Graber. They wasted 5 years trying to make crypto happen, then (hopefully) realized that crypto sucks, is not going to happen, and ruins lives, but that Merkle Search Trees are kinda cool, and didn't the web used to be more fun way back when. So they made BlueSky.
She has a suspect investor / board member, but Jack is not building this. All praise/criticism should go to Jay.
@mekkaokereke@runewake2@seldo I think it’s fair to say that anyone who is just now figuring out that crypto is a bad idea is absolutely not qualified to run a social media network or design its protocol, and nobody should trust them for another decade or so. The sorts of decisions that we see here are not overtly malicious but are fully consistent with and emblematic of a sort of juvenile utopian libertarian worldview that’s wreaked plenty of havoc already.
@glyph@mekkaokereke@runewake2@seldo couldn’t have said it better myself. The design philosophy they have has all of the wrong priorities. I really hope they don’t “win” the current social media wars.
Their motto seems to be "move fast and break things (that were built by or belong to other people, and it's those folks who'll have to pay to repair or replace what you damaged because they can't afford to sue you)".
@runewake2 @glyph I can't find it anywhere, but I stumbled across a screencap of Jay Graber tweeting something along the lines of "Oops, my bad. LOL" in response to certain issues raised by users or potential users. Man, I wish I favorited that toot, so I could at least share it with you.
@mekkaokereke@seldo sorry but I have to disagree. If you’re building a social network in 2023 and you’re not starting with solid moderation capabilities your development philosophy is really broken. As soon as you get real people using things they’re going to try to find the boundaries of what’s acceptable. Especially when there’s internet kudos to be had for shitposting and being “edgy”. Test it amongst friends you trust, not internet randos.
@CatherineFlick@mekkaokereke@seldo the fact that you could brute-force guess sign-up keys doesn't bode well for their security skills either. A startup in stealth mode could get away with that, not the much hyped Son of Twitter.
@CatherineFlick@mekkaokereke@seldo I think there's things to be said for learning the dynamics of how harassment will work on your network so you can build appropriate tools, but otoh not having blocks at all to begin with is a bit of a mess.
Blocks in federated systems are an interesting one, though. Do you federate blocks to other servers? There are advantages and disadvantages to each! Funnily enough there are different decisions on this just in ActivityPub space, Mastodon chose to federate blocks and the Pleroma family didn't.
(I tend to be in camp "You should not federate blocks by default, because you cannot assume other servers are cooperative and will not use knowledge of who's blocked them for harassment)
@erincandescent@mekkaokereke@seldo yeah, you could at least have a go, it doesn’t take much to wargame a few scenarios you need to protect the network/users in. If you get it wrong that’s a different thing, but nothing at all just screams a low priority for user safety in general.
@erincandescent@mekkaokereke@seldo yeah it seems a shame to not at least look at what AP has done and learn from it. Kicking the “it’s a hard problem” can down the road only makes it harder, not easier.
@CatherineFlick@mekkaokereke@seldo there are things we got wrong because we didn't understand how harassment dynamics would play out in a distributed system. If you'd watched ActivityPub closely over the last 5 years, you'd see them.
I have ideas for how to fix them, and the fixes aren't particularly complicated; but it seems like they're gonna repeat our mistakes
@erincandescent@CatherineFlick@mekkaokereke@seldo I'm a bit newer around here and haven't had much opportunity to observe the challenges. Would you be willing to share (maybe you already have?) some of the things you think were mistakes?
@CatherineFlick@mekkaokereke@seldo Yeah, we wrote a tiny social network for JoCo Cruise that’s only used on a boat once a year and even with a tight community of no more than about 2000 liberal-leaning nerds we’ve had to implement moderation. It’s pretty much a requirement as soon as you think more than a double-digit number of humans are gonna be on it.
When you ask on BlueSky why blocking isn't implemented, you get referred to an airy, wandering theoretical post about how there are Interesting Technical Questions To Be Answered. Server level or client? Public or not? What about servers which don't respect blocking?
Completely bonkers that they think "interesting philosophical questions, amirite?" is a valid response.
@mekkaokereke@seldo there’s a lot of pearl clutching going on on Mastodon right now ¯_(ツ)_/¯
I think the main mistake BlueSky made was not setting expectations on sign up, e.g. a simple “This is an alpha version and some features aren’t implement yet: …. If you need one of these features please wait before joining.” message.
Many Black Fedi users, including me, have received comments on their posts including graphic images and videos of real humans being murdered, along with racist slurs. Most white Mastodon users don't even see these threats, because of how comments and replies work on Mastodon.
When Black users complain about this, they're told to CW the discussion of their own abuse, but the abuse vectors don't get closed. And it's been years. 🤷🏿♂️
Add comment