Harassment on Twitter is a very real problem, not just celebrities, but for regular users too. In order to help curb abuse on the platform (a fact that has reportedly kept the company from being sold), Twitter introduced a new feature last week that would put abusers in “time out” if they tweet offensive content at other users.
So if you wanted to tweet “fuck you” at a specific user or tell someone to “fuck off,” you’d run the risk of having your account “temporarily limited” for 12 hours. Basically, you would be able to see your tweets, and your followers would be able to see your tweets, but no one else would.
In theory, that’s not a bad idea. As we’ve said, harassment on Twitter is a real problem and the company has struggled mightily to make things better.
But these new filters can be overzealous. Twitter user Victoria Fierce quoted a tweet from Mike Pence’s official @VP Twitter account and added the commentary, “fuck you, I gotta piss, and you’re putting me - an American - in danger of assault by your white supremacist brothers.” Her account was immediately put into time out for 12 hours.
— Emily of the State (@EmilyGorcenski) February 23, 2017
That seems a bit harsh. Gizmodo has learned that the way Twitter’s abuse algorithm is supposed to work is that users who repeatedly tweet abusive content (and what is “abusive” is up to Twitter) will be put in time out when they engage with other users. So if you repeatedly tweet “fuck you” to various users, you run the risk of having your account limited.
The problem is, those filters appear to be very sensitive. Looking through Fierce’s timeline, the only other content that could be potentially seen as objectionable was a tweet containing the word “piss” and the handle @POTUS.
I was curious about how sensitive these filters were, so I started tweeting various objectionable content to various verified accounts. Amazingly, as soon as I tweeted “fuck you @VP” or “fuck you @POTUS,” my test accounts were immediately limited for 12 hours. Other Gizmodo colleagues had similar results, always after tweeting @VP or @POTUS. In my tests, tweeting the same content to other verified users (including @RealDonaldTrump), did not result in any time outs. From an outside perspective, it certainly looks like the POTUS and VP accounts are given certain abuse protections that might not apply to others.
A Twitter spokesperson assured me that this is not the case and that every account is treated equally. The spokesperson added that there will be times when Twitter gets things wrong, but the company is working fast to solve a very real problem.
I have no reason to doubt Twitter’s sincerity in trying to address abuse, but it seems fair to say that the way the new filter currently works lacks nuance. The most problematic tweets aren’t those that say, “fuck you,” it’s the ones that say “I’m going to kill you.” In my tests, “I’m going to kill you” tweets were ignored whereas “fuck you” tweets sent to @POTUS or @VP almost immediately made an account vulnerable to being limited.
I know Twitter is working hard on this problem and I recognise that it isn’t easy to solve. Let’s just hope the solution isn’t limiting users who use language that some people find objectionable, while allowing others to make much more pernicious and dangerous threats unabated.