mhoye

Working there must feel great.

Are they deleting the user accounts or just the clips that violate copyright?

From the look of it they're deleting the tweets and locking out the account for some amount of time (hours, not days.)

Yet I've reported multiple account for multiple violations of TOS (including copyright) and one month later, photos are still there.

Trying to get Twitter to do the right thing feels a lot like pushing a disgusting, greasy rock up a disgusting, greasy hill.

We should build a decentralized alternative — we won't be able to exclude abusers, but at least it won't protect Ⓒ

I like the idea of hiding tweets from brand new egg accounts if they get reported a lot. Keeps the sock puppets at bay.

It wouldn't take many options to make Twitter viable. Options to mute new accounts & share blocklists would go far.

Disallowing non-follower retweets, a few other things. @randileeharper has thought & written about this a lot.

not to quibble, nor to even really disagree, but you can algorithmically detect one with relative ease, but not the other.

Sure, but that relative ease is entirely because an existing investment of research and development.

A fraction of that effort put into ablating harassment would go a very long way (muting sockpuppet accounts & dogpiling, eg.)

as someone who works closely with Facebook's AI team, I can tell you that harassment is actually much harder & costlier to detect.

I would like to talk to you more about this, but Twitter is unusable for me right now. I'm completely snowed in.

Indeed. On this evidence, if only every harasser did so via easily matched copyrighted video clips we would be all set!

I laughed at your idea that automatching video clips is somehow much easier than plain text.

Dealing with harassment isn't a simple text matching problem, online harassment is about context as much as content.