Apparently Piefed OCR’s every image for the text " Anonymous" and “no” (horrible code by the way since it doesn’t look for it sequentially but just those two strings on the whole image) for 4Chan posts and then censor those with an error message and -1 the uploaders social credit score.
there’s hardcoded censorship?
edit: No it’s not it’s optional
I don’t see why people call it hard coded when it’s possible to disable. That’s not what hard coded means, it’s just coded.
Apparently Piefed OCR’s every image for the text " Anonymous" and “no” (horrible code by the way since it doesn’t look for it sequentially but just those two strings on the whole image) for 4Chan posts and then censor those with an error message and -1 the uploaders social credit score.
That’s wild. I mean that could be cool if it was a feature that instance admins or community mods can enable. But so stupid like this
It is a feature they can enable.
Last I understood you had to opt out and they are enabled by default.
I hope only admins can enable it as it could kill backend performance if on a low spec box
This reply to another thread highlights some interesting things in the code:
https://lemmy.ml/comment/23619001