Imagine you suddenly getting banned after posting a picture of a chessboard

Or X11 default root weave pattern.

https://matttproud.com/blog/posts/x-window-system-boot-stipp...

It's an old problem: https://gizmodo.com/british-cops-want-to-use-ai-to-spot-porn...

I've been fairly convinced a lot of social media bans, where people are genuinely confused about the bans, are just automatic moderation problems.

It's for detecting porn? I genuinely don't get it. I regularly post actual nudes on discord and never so much as a warning, crazy that a chessboard or grids are doing it. I wonder what makes an algorithm view a grid pattern as pornographic.

Somebody said people were posting images of QR codes that contained (links to?) CSAM. But instead of using a QR code decoder, it sounds like Discord used perceptual hashing or AI to detect and ban all QR-like images.

Ohhh that makes a lot more sense. Well, their implementation doesn't, but it makes sense that grids were mistaken for QR codes.

> “When our staff reviewed and cleared those accounts, the same bug prevented the ban from being lifted automatically, so it just stayed in place,” Discord says.

It seems like a similar issue is affecting Twitter currently. People are being suspended for inauthentic behavior, they appeal, they get an email saying the appeal was accepted, and then, nothing, the account doesn't come back. Is there perhaps a third party moderation platform behind both of these services?

https://www.reddit.com/r/twitterhelp/comments/1sr7t1m/x_rest...

https://devcommunity.x.com/t/restoration-emails-for-suspende...

Optimistic future: Moderation scaling issues prevent monopolies and ensure people have alternatives.

Pessimistic future: There will be a couple major services and some people will be frozen-out victims of a system focused on cheap rather than correct.