Content Moderation Is Impossible: You Can't Expect Moderators To Understand Satire Or Irony

The latest in our never ending series of posts on why content moderation at scale is impossible to do well, involves Twitter now claiming that a tweet from the account @TheTweetOfGod somehow violates its policies:

What the fuck, Twitter.
Seriously.
What the fuck.
What the fuck is this. pic.twitter.com/NQwcRXFWg7

- God (@TheTweetOfGod) June 11, 2019

If you're unfamiliar with that particular Twitter account, it is a popular account that pretends to tweet pithy statements from "God" that attempt (often not very well, in my opinion) to be funny in a sort of ironic, satirical way. I've found it to miss a lot more than it hits, but that's only my personal opinion. Apparently, Twitter's content moderation elves had a problem with the tweet above. And it's not hard to see why. Somewhere Twitter has a set of rules that include that it's a violation of its rules to mock certain classes of people -- and that includes making fun of people for their sexual orientation, which violates Twitter's rules on "hateful conduct." And it's not difficult to see how a random content moderation employee would skim a tweet like the one flagged above, not recognize the context, the fact that it's an attempt at satire, and flag it as a problem.

Thankfully, in this case, Twitter did correct it upon appeal, but it's just another reminder that so many things tend to trip up content moderators -- especially when they have to moderate a huge amount of content -- and satire and irony are categories that frequently trip up such systems.