View profile

How Facebook beat Twitter at fighting fake news, in two charts

September 14 · Issue #206 · View online
The Interface
Reality has had a tough year. When the president of the United States is denying that thousands of Americans died in a hurricane, journalists face an uphill battle. That battle has been particularly fraught on social networks, where malicious actors have spent the past several years peddling hoaxes and sowing division.
But today I bring you at least one reason for optimism. “Trends in the Diffusion of Misinformation on Social Media,” a new study from authors at Stanford University and New York University, analyzed the performance of stories posted on fake news sites from January 2015 to July 2018. Here’s what they found (emphasis mine):
Interactions with these sites on both Facebook and Twitter rose steadily through the end of 2016. Interactions then fell sharply on Facebook while they continued to rise on Twitter, with the ratio of Facebook engagements to Twitter shares falling by approximately 60 percent. We see no similar pattern for other news, business, or culture sites, where interactions have been relatively stable over time and have followed similar trends on the two platforms both before and after the election.
For the study, authors Hunt Allcott, Matthew Getzkow, and Chuan Yu assembled a list of 570 sites that had been identified as peddlers of false stories in previous stories. It then measured engagements for a range of publishers — big mainstream ones, small mainstream ones, and niche business culture sites, alongside the fake ones.
Here’s the key paragraph from the study’s findings:
The results show that interactions with the fake news sites in our database rose steadily on both Facebook and Twitter from early 2015 to the months just after the 2016 election. Interactions then declined by more than half on Facebook, while they continued to rise on Twitter. The ratio of Facebook engagements to Twitter shares was roughly steady at around 40:1 from the beginning of our period to late 2016, then fell to roughly 15:1 by the end of our sample period. In contrast, 2 interactions with major news sites, small news sites, and business and culture sites have all remained relatively stable over time, and have followed similar trends on Facebook and Twitter both before and after the 2016 election. While this evidence is far from definitive, we see it as consistent with the view that the overall magnitude of the misinformation problem may have declined, at least temporarily, and that efforts by Facebook following the 2016 election to limit the diffusion of misinformation may have had a meaningful impact.
So what are the caveats? The authors mention a few. One, new publishers of fake news pop up and disappear all the time. This study measures only the performance of longer-lasting sites — although, given their relative stability on the platform, they are likely some of the largest peddlers of misinformation. As the authors note, fake-news publishers often change their domain names to evade detection and further confuse people.
Here are a few more. The extent to which bad content is shared publicly is only one way to measure the health of a platform. Platforms also strike against misinformation preemptively by banning fake accounts as they are created, filtering hoaxes from search results and trends, and so on. The ratios identified in this study don’t take those into account. Moreover, sometimes people share bad content in order to debunk it — a quote-tweet saying “This is garbage,” for example. It does not appear that the authors took this kind of sharing into account, though I’m inquiring about it.
Finally, Facebook in particular has shrunk the reach of nearly all news sites, thanks to changes to the News Feed rolled out earlier this year. That shouldn’t affect how we view the ratios identified by the study, but it’s important to remember that most people are seeing less news in their feed, period.
Still, I appreciate the value of this study, and the work by the platforms that it represents. Often executives offer us only boilerplate statements about “making progress”; this study offers a look at what progress might look like.
It also offers a look at the magnitude of the problem ahead. On Twitter, false news stories get between 4 million and 6 million engagements a month, and have since the election, the authors found. On Facebook, fake-news engagement has fallen dramatically from 2016, when it hovered around 200 million — roughly the same as engagement on legitimate news stories. But fake news still gets 70 million engagements a month — more than enough to pollute the information ecosystem to the point of making it unreliable.
I’d still rather look for the silver lining here. Facebook demonetized publishers of fake news by preventing them from letting them use its advertising tools. Partnerships with fact checkers allowed trusted sources to evaluate whether news stories were false, and down-rank them accordingly. And it deprioritized news articles in favor of posts from friends and family — a mixed blessing, to be sure, but perhaps useful in the narrow case of discouraging hoaxes.
I’d love it if future studies examined the relative effectiveness of each of these approaches. I also think there’s an opportunity here for the platforms to share best practices with each other, and (in more limited ways) with the public.
In the meantime, I’m glad to see some progress that can be quantified. (Charts are below!)

From the NYU/Stanford study cited above
From the NYU/Stanford study cited above
Google China Prototype Links Searches To Phone Numbers
Twitter CEO Jack Dorsey interviewed by NYU professor Jay Rosen; podcast, transcript
Vietnam urges Facebook to open office ahead of controversial cyber law
Facebook Says This Post About A Firing Squad For A Philippine Senator Doesn't Violate Its Rules
Crisis Response
How Forlini’s Survives the Instagram Horde - The New York Times
Odell Beckham Jr. Docu-Series ‘I Am More: OBJ’ on Facebook Watch
Twitch Hires Its First Head of Diversity and Inclusion
Tech Giants Join Forces to Score AI Chips ($)
The problem with real news — and what we can do about it
And finally ...
Let’s end the week on a heartwarming note. While the social world is known for its ruthless copying, the lower-stakes world of internet browser building is positively collegial. What else to make of the apparently longstanding tradition in which browser makers celebrate one another with cake?
If there's one thing we can all agree on, it's cake! Thanks for the @GoogleChrome birthday present @MicrosoftEdge 🎂😀
Here’s to more companies sending one another congratulatory pastries. We’re all in this toegther, folks.
Talk to me
Send me tips, questions, comments, and weekend plans:
Did you enjoy this issue?
In order to unsubscribe, click here.
If you were forwarded this newsletter and you like it, you can subscribe here.
Powered by Revue