Early this month Twitter unveiled a new plan to crack down on harassment. They laid out three new tools to start fighting against abusive accounts and comments.
They neglected to mention, however, the real changes they would be making to punish users for using supposedly offensive language.
Last week we discovered Twitter was punishing accounts for using “offensive” language by removing account features for 12 hours. Now it appears they are “ghost” deleting Tweets they deem offensive.
When a Tweet is ghost deleted, the person who wrote the Tweet still sees it and does not know it is technically deleted. But everyone else trying to find the Tweet cannot see it, and even if you manually enter the Tweet’s URL, it will bring you to a page that says it was deleted.
This was first discovered by John Sweeney at SuperNerdLand whose offensive tweet was ghost deleted by Twitter.
Here is the original Tweet.
The actual text of the Tweet does not contain any foul language or targeted harassment. The meme, however, contains the term “autistic screeching” a 4chan joke, meaning a colossal freakout. For Twitter to have found offense with the term must mean they had to view the meme manually.
Sweeney conducted a few experiments to find out the cause of the deletion. When he Tweeted out the meme with no hashtag, Twitter would not delete it. But when another account @SuperNerdMike, Tweeted out the meme in conjunction with the hashtag, he got his Tweet ghost deleted as well.
— Be Just & Fear Not (@SuperNerdMike) February 19, 2017
Twitter is incredibly opaque when it comes to explaining how their “anti-harassment” tools work. Sweeney’s working theory is that Twitter was monitoring the popular hashtag #nottheenemy. When they found some highly retweeted offensive content, they surreptitiously deleted the Tweet.
#NotTheEnemy began as a anti-Trump hashtag, saying journalists should not be Trump’s enemy. Sweeney used the hashtag to make fun of journalists discussing Milo Yiannopoulos’ appearance on Bill Maher.