But the definition of "sex trafficking" was left vague, and the definition of "promote" probably borrowed from some other use of the word in the policy, which meant that it was time for mass misinterpretation.
-
Show this thread
-
So not only did some original FB post get taken down, but the Medium article it shared got marked as "promoting sex trafficking" because it argued for (misinterpretation of 'promote') the rights of porn studios (misinterpretation of 'sex trafficking').
1 reply 3 retweets 18 likesShow this thread -
Which would mean that anything sharing that article might get a high-level suspension! (NB: I haven't tested this yet. This is just one of the ways in which such a thing could go wrong, but it's a very realistic example case)
1 reply 1 retweet 11 likesShow this thread -
So what went wrong, and how could it be done correctly?
1 reply 0 retweets 8 likesShow this thread -
(1) Communication between the policy team and the reviewers: Companies tend to underestimate the need for expert translators between these two, and effectively dump the job on the managers of the review teams. This does not have the desired effect.
1 reply 0 retweets 21 likesShow this thread -
The "right" way to push a policy change is to write up the formal statement (for the legal team), write up the training statement, train a group of analysts, have them review stuff, check for inter-rater agreement, and update the training if needed.
1 reply 0 retweets 22 likesShow this thread -
This is generally considered way too expensive and time-consuming to do, especially for an urgent policy change. It's expensive because these people are hourly, training takes time, testing takes time, and there aren't enough raters for the language anyway.
2 replies 1 retweet 16 likesShow this thread -
This is part of a broader problem: the budget for abuse reviewers is *way* too low, by a factor of 3-5, compared to what they actually need to do. Solving this problem properly would require more time for training, more time per item, and a few more things below.
3 replies 3 retweets 27 likesShow this thread -
-
Replying to @RichFelker
I don't know. I think you could do it pretty well without an insane scaling; maybe 10x at the top. Beyond that, you're likely better off empowering communities to moderate themselves more effectively.
1 reply 0 retweets 0 likes
At the very least you should be hiring ppl with relevant degrees & cultural contexts at reasonable rates, not 1/10 minimum wage overseas to bypass labor laws.
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.