Twitter’s real-time updates makes it uniquely suited for broadcasting breaking news. However, the service seems to attract hackers and pranksters, ranging from a political impersonator‘s unusual policy suggestions to a bogus report of Obama’s supposed assassination. How does a reader determine what is true on Twitter?

A 2012 paper from Carnegie Mellon University and Microsoft Research presented at the annual Computer Supported Cooperative Work conference, “Tweeting Is Believing? Understanding Microblog Credibility Perceptions,” analyzes how users assess a tweet’s credibility. The researchers examine survey and experimental data captured from college-educated participants ages 18 to 60 who found tweets in a variety of ways; collectively, the survey group found them through “searches on search.twitter.com (84%), clicking trending topics on the Twitter homepage (84%), searching for tweets using Bing’s and Google’s social search functionality (72%), or serendipitously encountering tweets mixed into the results of general Web searches (81%).”

Key study findings include:

• Users assessing the credibility of an isolated tweet divorced from contextual markers such as the author bio and number of retweets are forced to rely on limited — and unreliable — cues such as subject matter or author prominence.
• Perceptions of the author’s influence, topical expertise and reputation all enhance a tweet’s credibility. Additional measures include the public profiles of tweeters and how often their posts are retweeted. “Features that often are obscured in the user interface, such as the bio of a user, receive little attention despite their ability to impact credibility judgments.”
• Typical users are not unduly concerned with the credibility of tweets on celebrity news and restaurant reviews, but are concerned with the veracity of breaking news and political content. They tend to most trust tweets from individuals they follow and trending topics listed on Twitter, and are very concerned about the credibility of tweets they find through Twitter searches and online search engines.
• While the perceived credibility of a tweet was linked to its author, it was not associated with the truthfulness of the tweet itself. This held true regardless of the assessor’s experience with Twitter; in fact, more experienced users typically rated tweets as more credible overall. “Those with more experience with a given technology view it as a more credible information source” than those with less experience.
• Tweets on science topics are rated as more credible than tweets on politics or entertainment topics; tweets with a topical user name relating to the discussed topic (i.e., ScienceNow) are rated as more credible than those with traditional (i.e., John Smith) or Internet (JSmith84) style user names.
• Users represented by the default Twitter icon are perceived as significantly less credible than users with any other type of icon image.

But even under the best circumstances, the authors caution, “Tweet consumers should keep in mind that many of these [trustworthy] metrics can be faked to varying extents.”

Tags: ethics, technology, news, cognition, Twitter