Twitter upends retweets in bid to stop spread of election misinformation

1

Twitter is making sizable changes to retweets and other features aimed at making it harder for politicians and other users to spread misinformation about the Nov. 3 election, the company announced Friday — the latest attempt by Silicon Valley platforms to protect the vote.

The changes are part of a push to protect the "critical dialogue" on Twitter "from attempts — both foreign and domestic — to undermine it," said Twitter’s general counsel Vijaya Gadde and head of product Kayvon Beykpour in a blog post announcing the moves.

The design changes, being rolled out in phases starting next week, are all meant to add friction to the distribution of misinformation, conspiracy theories and election interference. Among the most important:

Making it harder to share misinformation

Twitter will use warning labels to hide misleading tweets from some accounts, including those of U.S. politicians and people with more than 100,000 followers. Anyone attempting to retweet that post will get a notice indicating that the tweet is "disputed," and will be prevented from moving ahead unless they add their own commentary.

That could affect users like President Donald Trump, whom Twitter penalized just this week for a tweet Tuesday suggesting that Covid-19 is less dangerous than the flu. Under its existing rules, Twitter labeled that tweet as violating its policy against potentially harmful misinformation related to the coronavirus. Once the changes to the site roll out, users will first have to click through a warning label to get through to that tweet, and then will be prevented from reflexively retweeting the president unless they add a bit of skin to the game by attaching their own thoughts to the president’s message.

Hampering other retweets

From Oct. 20 through at least Election Day, anyone attempting to retweet any tweet will be nudged toward adding their own context before moving ahead. The changes, said Twitter, are meant to get users to stop and think before sharing what they find floating around on Twitter — a move, the company believes, that will significantly stop misinformation from going viral.

Policing the trends

Twitter will change its handling of its trending topics feature, which has drawn accusations of letting hate speech and other noxious content run wild. The company said it will make sure that topics that get the trending boost on the site have additional context — either a pinned tweet explaining the topic or a one-sentence write-up from the company. Twitter says that it has expanded the team handling the trending topics feature around the election and made it a 24-hour operation.

No calling the election early

Twitter also clarified its existing civic integrity policy to say it will label tweets that prematurely declare victory during the election and remove tweets that attempt to interfere in the vote, including through violence.

The backstory

Twitter, like all social networks, is scrambling to figure out how to cut the spread of misinformation, hate speech and conspiracy theories surrounding the U.S. election. (Twitter hasn’t faced the same pressure over political advertising that Facebook and others have because Twitter banned such ads last year.)

Friday’s announcements are a reflection of a realization that Twitter executives say they’ve come to in recent years — policy changes and even enforcement actions aren’t enough to clean up the service, but need to be paired with design changes to the platform can help nudge users behave in the way that Twitter would prefer. That’s why Friday’s news comes from both Gadde and Beykpour — the lawyer responsible for Twitter’s policy and safety operation and the engineer responsible for Twitter’s product, respectively.

Some researchers say that even the sort of mild-mannered prompts that Twitter is rolling out, like encouraging users to "Help keep Twitter a place for reliable info," are enough to cause many users to think twice.

What’s next?

Twitter’s new approach is going to quickly be tested — including the next time it labels one of Trump’s tweets.

View original post