Prior to the start of the U.S. elections this month, Twitter announced that it will be adding cautionary labels to misleading claims regarding the presidential race. The platform did this to some extent - even incumbent President Donald Trump"s latest tweet is labeled as "disputed" at the time of this writing -, it appears that work on this front is still in progress.
Now, Twitter has shared some statistics surrounding its efforts to combat the spread of fake news on the platform, details regarding product changes it made for this purpose, and plans for what"s next.
Twitter says that in the few months leading up to the elections, it encouraged people to vote and gave them prompts in their timelines and search results that voting by mail is legal and safe. It also informed people that election results might be delayed due to a large number of people opting to cast vote via mail. These prompts were seen 389 million times. Some further statistics from the time period of October 27-November 11 are summarized below:
- Almost 300,000 tweets were labeled as "disputed" because they contained content that could potentially mislead people. Overall, these constituted 0.2% of all tweets related to the U.S. elections.
- Out of the aforementioned 300,000 tweets, 456 had engagement features limited so people could only "Quote Tweet" them, not retweet, like, or reply.
- 74% of the people who viewed these tweets saw them after Twitter had already applied cautionary labels to them.
- There was a 29% decrease in Quote Tweets to these tweets containing disputed information following Twitter"s application of warning labels to them.
Regarding product changes that Twitter made to reduce the spread of misleading news, the company says that some of these modifications will be reverted while others will be kept as-is.
The company began recommending Quote Tweets to encourage thoughtful consideration before sharing content which saw an overall decrease in number of combined retweets and Quote Tweets. Twitter plans to analyze this impact further, so this product change will continue for now.
Regarding restricting content from people you do not follow in your timeline, the firm says:
We stopped providing “liked by” and “followed by” Tweet recommendations from accounts you don’t follow in the Home Timeline and through notifications. While we had initially hoped that this would help reduce the potential for misleading information to spread on our service, we did not observe a statistically significant difference in misinformation prevalence as a result of this change (nor any meaningful reduction in abuse reports). Instead, we found that pausing these recommendations prevented many people from discovering new conversations and accounts to follow. As of today, we are reverting this change.
In the "Trends" section, Twitter made a modification that would only show content after its team had manually added a description and some representative context so people could get an informed overview of publicly discussed topics. However, this resulted in people viewing topics they were not interested in. As such, the firm is backtracking on this change, and while context-rich Trends will be prioritized, Trends without context will not be restricted from this section.
Moving forward, Twitter says that it will continue its strategy not only to curb the spread of misinformation related to U.S. elections, but to elections all over the world. The company efforts "to protect the integrity of the election conversation" will continue in order to encourage healthy public conversation. A lengthier retrospective regarding Twitter"s work in the 2020 U.S. elections will be published in early 2021.