[rmatt2000]

In the political Twitter neighborhoods that I inhabit, there have been many cases over the years of people having their accounts yanked, with the only message being "You have been banned for violating the Twitter rules. This decision is final and cannot be appealed."

When these bannings get publicized, however, the bad publicity often leads to the "permanent ban" being reversed, with Twitter simply explaining that the original ban "was a mistake". How does this sort of thing happen so often, unless random employees have access to powerful tools that they are not authorized or trained to use?

[SilverBirch]

Here's the obvious counter - social media companies are by nature trying to chase super-linear cost:return ratios. So they can't actualy accept that N users mean N/x moderators. So instead, you hire Y moderators, and any time anyone points out you've shit the bed you wave your hands in the air and whisper "AI" when really what's happening is Y < N/x so you have a shit moderation team.

[rmatt2000]

If there aren't enough moderators, wouldn't the likely result be fewer rule-breaking accounts banned rather than more non-rule-breaking accounts banned?

Maybe, instead of hiring N skilled moderators for $D, they could be hiring N * \delta unskilled moderators for the same price.