It’s probably not at all surprising that most content posted to Reddit is voted on more or less blindly. I’ll cop to liking articles that friends have shared on Facebook without reading, let alone evaluating them. I’d say there’s even sort of an aggregation myth that pervades our view of social media, that buried within discussions of fake news and social media corporate responsibility is this assumption that people are actually reading the articles, or at least that a lot of them are. The data, however, suggests that they aren’t.
According to a paper published in IEEE Transactions on Computational Social Systems by researchers at Notre Dame University, some 73 percent of posts on Reddit are voted on by users that haven’t actually clicked through to view the content being rated. This is according to a newly released dataset consisting of all Reddit activity of 309 site users for a one year period. In the process, the researchers identified signs of “cognitive fatigue” in Reddit users most likely to vote on content. Online aggregation is then somewhat a function of mental exhaustion.
As of 2017, reddit sees 234 million unique monthly users, according to Alexa. It’s the fourth most trafficked website in the United States. In light of that, a 309 user sample seems pretty small. But it’s small because the Notre Dame dataset offers something a bit different than most other reddit datasets. Rather than relying on reddit’s own data collection methods (via its API or by crawling the website with software), Notre Dame computer science researcher Tim Weninger and colleagues were able to come up with their own independent data collection architecture, thus allowing them to ask different sorts of questions. Like, are redditors even bothering to read the articles? Or, are they, in the words of Weninger and co., merely “headline browsers?”
They accomplished this by building plugins for the Chrome and Firefox web browsers that simply record complete snapshots of user activity while on reddit. Calls for study participants were then put out via postings on certain subreddits. (Weninger and his team note studying self-selecting pools of users is in itself problematic, but to ask these sorts of questions there wasn’t much of a way around it.)
“The data show that most study participants were headline browsers,” the study concludes. “Specifically, 84 percent of participants interacted with content in less than 50 percent of their pageloads, and the vast majority (94 percent) of participants in less than 60 percent of their pageloads.”
I think we can mostly agree that this is bad. As those of us that click through to the articles know well enough, headlines are very often poor representations of the substance of the content within. Moreover, it adds an interesting twist to discussions of fake news sites. We’re often befuddled by the traction that obvious, malignant bullshit gets online, but that obviousness—including literal satire disclaimers—doesn’t often percolate upward to headlines. One might even say that headline browsers are in some part responsible for giving the US its headline president.