Aaron Rodericks @aaron.bsky.team · 7hLook man write your own rules about who gets to bluster and what’s real or not. How would you decide and enforce on who gets to make a threat and when that is serious or not? Obviously that doesn’t work at scale. Imagine getting 250k reports in a day and trying to figure out who is not serious. 130
Aaron Rodericks @aaron.bsky.team · 8hIt was the first time we had graphic content sweep the site in a crisis so it’s hard to get it right with scaled human sweeps on the first try. It’s compared to cases like the Hortmans where there’s no equivalent violative content on the site as signs of bias. They aren’t comparable. 210
Aaron Rodericks @aaron.bsky.team · 8hThose were literally two false positives in the sweep that were errors and released - and we sent apologies to both users. Errors happen when trying to enforce at scale. 140
Aaron Rodericks @aaron.bsky.team · 8hOh I never meant to imply it didn’t apply. It applies to us right now. Just at a lighter level than full VLOP status due to our size. 110
Aaron Rodericks @aaron.bsky.team · 9hWell the mod inbox will help since people will see the outcome of their reports. Beyond that these things take a while to shift and there’s only so much a small team of humans can do. 150
Aaron Rodericks @aaron.bsky.team · 9hAreas where I attempt to discuss policy aren’t official announcements - they are trying to making things more accessible to the average user. If done through the safety account I just wouldn’t bother with comms approvals etc that’s just a personal goal to help people understand. 020
Aaron Rodericks @aaron.bsky.team · 9hWe have the safety account for updates. I discuss things related to T&S but all official announcement on behalf of the company go through there. Ie I’ll discuss what I’m thinking and engage with people on it but that’s what the safety account is for. 120
Aaron Rodericks @aaron.bsky.team · 9hIn one famous example someone said they were suspended for ‘wishing ill on a famous anti trans author’ when the actual wording was that they wanted to split her skull open with an axe. Users usually tell only one side of the story and excuse the behaviour of their friends. 5160
Aaron Rodericks @aaron.bsky.team · 9hI had a case yesterday where it was escalated internally that someone was defending their community. k. We go check. Perhaps we missed context. Ah they were advocating for burning down a school. Right. 1140
Aaron Rodericks @aaron.bsky.team · 9hYup. Thats why we are putting in toxicity detection at scale to catch accounts repeatedly doing this. Users often lie about moderation actions. It’s just part of the game. Nearly every high profile suspension is blamed on brigading, defence of values etc. 2111
Aaron Rodericks @aaron.bsky.team · 9hYes but I’ve tried many other options over the years here and now I’m trying this one. Perhaps it’s like democracy. It’s the worst option except for all the others. 020
Aaron Rodericks @aaron.bsky.team · 9hPeople have tried harassing me at home, via other accounts and networks, made multiple physical threats. Not sure why they would stop at two accounts on Bluesky and respect the line between them? 020
Aaron Rodericks @aaron.bsky.team · 9hI have considered it and still considering it. As far as I’m aware there aren’t too many T&S leaders with public accounts on their own site so I don’t have one specific model to follow. I have tried a number of things and that is certainly one of them. Doesn’t solve for abuse though. 220
Aaron Rodericks @aaron.bsky.team · 9hIe blacksky, northsky, wsocial, eurosky etc. depending on whether they host their users data and their own app - that wouldn’t actually be a responsibility of Bluesky. Either way I don’t thing our architecture and the DSA are incompatible. 120
Aaron Rodericks @aaron.bsky.team · 9hThere are elements to this that are right and elements to this that are wrong. Are there tons of obligations with the DSA. Yes. Would allowing communities to add layers of safety on top of Bluesky’s be a violation? Most likely not. If you are familiar with atproto some have their own apps entirely. 110
Aaron Rodericks @aaron.bsky.team · 9hI am blocking people who abuse me personally. There are tons of great folks from marginalized communities that can I can and do engage with to surface their concerns and those are the ones who I want to hear their perspectives than people who make death threats as part of their argument. 220
Aaron Rodericks @aaron.bsky.team · 9hA lot of the tension is because people want enforcement to be immediate, only done with humans, factor in deep context, and then align only with their values. These are an impossible set of requirements. 180
Aaron Rodericks @aaron.bsky.team · 9hThe failure of large platforms is that it’s one size fits all. Minorities get less protection under the same standard rules. We recognise that and that’s why the other layers of composable moderation at there. It can be as simple as a shared blocklist or more complex. 130
Aaron Rodericks @aaron.bsky.team · 9hWe enforce on the community guidelines across the site in all users. Now it’s largely a team of humans doing their best on a large volume of reports so there’s no getting it perfect all the time but I’d have to disagree with the premise that we aren’t trying to protect everyone. 110
Reposted by Aaron RodericksDarrin Loeliger @darrin.bsky.team · 21hWe want to make it easier for creators to link out and attract financial support, so we're exploring a new feature that lets you add links on your profile. They show up as buttons under your bio. 3239884
Aaron Rodericks @aaron.bsky.team · 11hThese are for blocks on my own account. I don’t see how blocks on my own account from direct abuse ends in death camps. 410
Aaron Rodericks @aaron.bsky.team · 11hThere’s been enough discourse this week on my account but that’s the point of going through this exercise - ie talk about what this site can be to people who want to listen :) 020
Aaron Rodericks @aaron.bsky.team · 11h100 and due to the nature of this convo our ban evasion infra isn’t and can’t be as good as the big incumbents. Ie we don’t collect phone number’s and IT wouldn’t work off of our infra. 060
Aaron Rodericks @aaron.bsky.team · 11hIt feels like people are deliberately trying to take things out of context at a certain point. 320
Aaron Rodericks @aaron.bsky.team · 11hbsky.app/profile/aaro... People took one post in my thread out of context - many deliberately. There was never anything to change in my stance. 260
Aaron Rodericks @aaron.bsky.team · 12hI’m fairly sure it will be launched before the end of the year for the in app mod inbox. 020
Aaron Rodericks @aaron.bsky.team · 22hI have nothing to add other than that’s a great display name! 👌 150
Aaron Rodericks @aaron.bsky.team · 22hOh I recognise that school would be a wonderful escape but unless I’m completely reskilling into a completely different career I’m a glutton for impossible challenges like this one. 130
Aaron Rodericks @aaron.bsky.team · 22hNow folks aren’t wrong! I don’t want them to have to report content. It’s just I started with barely any systems and 20 mods and we did everything manually. We do detect spam and CSAM at scale but I do want to get proactive detection in place so people don’t have to report everything. WIP. 120
Aaron Rodericks @aaron.bsky.team · 22hOh don’t get me started on the headaches I have had because folks don’t report content but yell instead. That being said it’s also on us for not having shipped the mod inbox but hopefully resolvable before end of year. 320
Aaron Rodericks @aaron.bsky.team · 23hI had to write the text of the post in my thread… because they were banned when I wrote the thread. So I don’t really know what I’m being accused of lying about in this case but whatever. 120
Aaron Rodericks @aaron.bsky.team · 23hYou know the block tool is available to everyone so this seems like a misguided, if not entirely bad faith argument. 8210
Aaron Rodericks @aaron.bsky.team · 23hReading your posts makes me want to go back to school for social network theory. I loved that class. I may have used mean girls in my final presentation. Truly an amazing movie for social dynamics. 140
Aaron Rodericks @aaron.bsky.team · 23hyeah so we had just issued numbered threads in app, so I, in my idiocy decided to not number my posts in thread, and found out the brutally hard way that the thread numbering drops for quote posts. So back to numbering it is. 160
Aaron Rodericks @aaron.bsky.team · 23hThe part that confuses me is that Bluesky probably has the highest proportion of trans users of any social network. Some among them are awesome, others are not. Just like people? I've tried to not paint one characteristic as an entire community, since not everyone makes it their whole identity. 160
Aaron Rodericks @aaron.bsky.team · 23hI worked with a team of quantum cryptographers/computation folks a few years ago in Canada. Most dumb questions/statements like mine to them would get a 90 min answer, so appreciate you saying - no, no. That doesn't work. 120
Aaron Rodericks @aaron.bsky.team · 23hcorrect, thank you for appreciating the nuance in T&S decision which seems truly impossible to convey via 300 characters. I will take a stab at blogging (when I find time) to see if long form improves reception. 1170
Aaron Rodericks @aaron.bsky.team · 01/10/2026Oh fair. I wondered whether there was an acceleration effect but if you say no that’s probably good! 140
Aaron Rodericks @aaron.bsky.team · 01/10/2026I think it will get better and simultaneously worse. Folks will troll the messages and use it to game the refs etc so there will be some growing pains once that’s released. 170
Aaron Rodericks @aaron.bsky.team · 01/10/20261. bsky.app/profile/aaro... 2. Yeah I tried that on the thread you are referencing and everyone took out of context quote posts to be the entire answer since the numbering didn’t show up there… so I’m still manually numbering. 020
Aaron Rodericks @aaron.bsky.team · 01/10/2026The person was suspended for harassment and the content was taken down. I could have phrased better, but I was responding to a question about what policy applied. People took policy A not being applied vs policy B being applied, as - this is okay to say. Not the case. 2191
Aaron Rodericks @aaron.bsky.team · 01/10/2026Between this and the enigma machine (that I haven't read in detail), I wonder how long it will take a combo of quantum + ai to start really making a lot of current day cryptography obsolete. 690