So uhhhh are there any moderation services left now, or did we throw every baby out with the bathwater? I only ask because since Aegis and Taurus both shut down, even lowbies like me are getting hit with more transphobic shit from reply guys who only come here to do pvp against queer people
currently nothing, and i don't know if anyone is interested in starting any new moderation labelers rn because the atmosphere is fucking horrible for it and nobody wants to jump in front of that bullet even from people i've talked to about it.
a key issue is not sticking to one lane. kairi was-
singlehandedly doing 90% of aegis operations which was news to me, and she patently was a fucking idiot who didn't know shit about jack. it should've stayed at transphobia. we don't need comprehensive labelers, we need niche labels who know their shit on one specific issue so we can mix and match.
the other issue is trust. this is the fundamental one that we will be reeling with for a while as to why the atmosphere is so bad. the labeler needs anonymity, but the community needs transparency. there's a cursed problem of trust of how to develop trust with users who should not trust you.
The Taurus labeler shut down because the guy running it got... Let's say criticized, for not running the anti-Semitism label in a way that some of the bigger voices were happy with.
That criticism increased when he (rightly) suggested that service would be better handled by people who weren't him.
I agree and him being willing to bow out was, in my view, absolutely the correct choice. It's a discussion he isn't really equipped to have.
I'm a little miffed that, even if he was a bit ignorant on anti-Zionism vs anti-Semitism, the response effectively killed his interest in trying to help.
he was right to bow out and it was likely overboard but that's precisely what i mean in that a person operating a labeler needs to be able to disconnect themselves from the labeler for operational efficiency in the first place tbh
…and one of the ways to do that is it needs to be run by multiple people who have equivalent authority and have to act by consensus, especially about appeals. At the very least, a mod manually labeling an account or post needs to be reviewed before it’s applied.
well, that opens up issues. rah's thread about it is great but that effectively doubles the work where they agree in 99% of cases but don't catch the 1% because they usually agree so they don't look as hard. the moment you scale to multiple people, you've got a lot of new things to worry about
For sure. I don’t think reactive labeling, ie user reported, needs that kind of oversight, that can likely default to labeling based on some threshold number of reports, but IIRC part of the Aegis drama was Amos decided to proactively label some accounts without sufficiently-justifying reports.
The flowchart I think would actually be fairly complex. Reports should be weighted for review priority, with a heat map of most-reported users and posts, and the appeals system should likely pre-filter for empty appeals or appeals that are just swearing or keysmashing.
That's not built into Ozone and I don't know if there is any impetus to include that kind of stuff.
I don't even know if those actions show up on the Firehose for people to build systems to do that.
sorry if this is a little too cross-examiney, but 1) what is Ozone? 2) did Aegis at any point publish a concise "Code of Conduct" that publicly stated which tags exist and a sentence or two why a mod would apply that label?
Yes, Aegis had that. But there’s always a human at some point, and the more specific your published rules are, the easier it is for abusers to take advantage of loopholes. The private rules should never be public, and need a lot of maturity to evolve and monitor for effectiveness.
Well that and I think they kind of miss a key point of human behavior. Societies and communities run on fundamentally different systems.
Societies run on rules. They're big, they're diverse, they have multiple elements who disagree with each other, and they have to scale to infinity.
1/x
Communities run on vibes. They're more focused, they can exclude, they are less fractious by being self selecting.
When these systems are built and their guidelines set, I think too many try to run society level rules when they want community level vibes.
2/x
bsky (and others, like Reddit) have offloaded most moderation to the communities on their platform. That makes things very susceptible to “vibes” because communities usually lack sufficient expertise to self-moderate at larger than small scales.
yes, subreddit moderation was always vibes based because
1) volunteer-driven
2) subreddits are vibes based
this is sort of different than the way that wikipedia page rules are vibes-based. but neither subbredits nor wikipedia pages are things i think anyone should call "communities"
Work forward from two principles:
1. Laws and rules aren't real
2. Enforcement of laws and rules are
The society level stuff is going to be strictly enforced in very limited cases.
The community level stuff gets enforced to a variable degree but can be done so in nearly every interaction.
why would i work from those two principles?
i don't really agree that there is a remotely coherent definition for "community level versus society level"
if you said something like "both the US & UK Constitutions are just Codes of Conduct, and Cops are the Mods" then i would agree / nod along
I think the rah thread provides a far more useful distinction of objective facts vs subjective characteristics.
Community vs society is too blurry of a line to be useful as differentiating criteria, whether or not it exists (also somewhat covered in the rah thread)
yeah, katie linked the tippy top of the thread, but i linked to the second tweet because these were the two questions i wanted answers to re: Aegis vs "BSky Official Labeler"
This thread covers the two fundamental things all labelers need to decide on up front and stick to:
1) Who is doing the moderation, what are their biases, and how are those biases mitigated?
2) Are you moderating/labeling objective actions/content, or subjective characteristics?
Publicly Aegis preached decent standards and approaches, yet fell into the trap of ignoring #2.
Pile on the inevitable personal conflicts becoming a huge problem = destroyed all credibility and trust
Bsky official T&S has been much more quiet, and is hard to evaluate externally for similar risks.
The head of Bluesky T&S is @aaron.bsky.team. He's doing a great job and has my complete confidence! He also is doing an extremely hard job and has 15,000 priorities at any given moment, and every time he tries to explain things people collectively lose their shit at him.
I would guess rahaeli mostly intended those two points to be questions for any new team getting going, and this isn't a totally official/authoritative statement, but w/r/t dvd's concern:
1) bluesky mod team employees do the work, following policies and authority of the overall company (a USA corporation)
2) working against the public policies, trying hard to keep it objective, consistent, understandable. context and nuance matter, but need to be formalized
the last (only) time i got emotional and yelled at aaron, was because of an account framed as a Palestinian refugee's (teenage?) child had been posting a GoFundMe link too much and getting spam complaints.
from a less immediate emotional reaction perspective... i have a better viewpoint... 1/2
Also, the characteristic vs. objective fact distinction
If a behavior meets the objective criteria of “spam”, is it? Most people arguing that one scenario were looking for an exception for that specific message, which changes the label to a “characteristic” decision judging the sender’s intent
