Over 70 thousand people shared a story about a totally fake Sarah Palin quote! Over 5 million people shared a hoax story that Macaulay Culkin had died! It gets depressing hearing how many people get fooled by these hoaxes, doesn’t it?
The problem is, the numbers in those reports are wrong! Often, wildly wrong. They’re exaggerations caused by the confusing way that Facebook reports engagement.
Now, the underlying problem is real – social media hoaxes and rumors are bigger than ever. As a result debunking these things has become a popular pastime, well beyond the circle of organized skepticism.
Even the Washington Post runs a regular feature on Friday called What was fake on the internet this week. The science fiction site IO9 regularly debunks fake images that are making the rounds. And of course there are the old standards such as Snopes and Museum of Hoaxes, still in the business of debunking this stuff.
Read on to see how many of these well-meaning debunkers are being misled by Facebook into over-reporting the problem.
I’ve written before about how those little Facebook “Like” buttons seen on websites report inflated numbers. Here’s an example from right here on this blog. Pictured below is an image of the share buttons attached to my 2011 post about David Mabus getting arrested, one of the most popular posts on this blog.
The buttons indicate that particular post had been shared on Twitter 759 times and on Facebook 774 times. But the Facebook number is misleading!
So how do you find a more correct number? If you’re a programmer you could find the answer by directly calling the Facebook API. But most people will find it easier to simply paste the URL into a handy form found at the free service SharedCount.com. After you do that you get this result, captured below/right in graphic form.
As you can see, the article in question was actually only shared on Facebook 311 times, but those shares attracted 210 Likes and 253 additional comments by other people. (The numbers may vary from the screen shot here by the time you’re reading this article).
Those three numbers are totaled in the Facebook button to get a total engagement figure of 774. In this case, it more than doubles the number of times this article was shared on Facebook. I believe Facebook does this deliberately. Perhaps it is to help sites gain popularity or perhaps to inflate Facebook’s own importance.
And as we’ve seen in a January post on this blog, even links sent in direct messages on Facebook were (at least at one time) used to inflate these numbers. This can also cause an interesting problem for skeptics, as it means the act of posting a link to debunk it causes it to seem more popular – to both other people and to Facebook’s news feed algorithm.
So on to the recent cases…
Over-Reporting Hoax Stories
An annoying trend in viral hoaxes is the recent flood of “fake news” sites. These are conceptually similar to The Onion, minus any shred of entertainment value. Most of them claim to be satire or parody, but really seem to be solely after clicks. Their content is rarely the least bit funny. Here are three examples from recent months, showing how the virality is consistently over-reported in the debunking community.
Sarah Palin Calls on Obama to Invade Ebola
Posted in mid-September on the fake news site National Report, this article was debunked by Snopes on October 3rd. Snopes reported that it had “shared nearly 27,000 times on Facebook.” I checked the statistics on this page via SharedCount on November 14, and found it only had 5,977 shares and a total engagement (likes plus shares plus comments) of 30,880. So it is likely it was shared only about one fifth as much as Snopes reported.
Disney Announces First Film With Openly Gay Characters
Posted on November 10 on another fake news site called Amplifying Glass, this article was debunked by Mike P. Williams on Yahoo Movies on November 14, with the claim that it had been “shared over 60,000 times on Facebook”. Checking the actual count via SharedCount later on the same day turned up just over 33,000 shares but a whopping total engagement of nearly 247K, most of which was likes and comments. So Yahoo over-estimated the sharing by a factor of about 2, but a standard Facebook button would over-estimate by a factor of 7.5!
Macaulay Culkin Found Dead
The biggest recent exaggeration I found was a garden-variety “dead celebrity” hoax which made the rounds last week. A blatant clickbait site with a URL designed to be confused with the MSNBC TV channel had reported the death. The bare-bones story featured a prominent advertising block to generate revenue. Here’s a screen shot of the story when I visited it on November 14:
Confusingly, there are three Facebook-branded buttons here, two of which show differing share numbers – that should raise some suspicions right there.
In a debunk earlier in the week, Jim Waterson at BuzzFeed had noted the share count based on the leftmost button, stating “Facebook users have shared the link to the hoax story over 700,000 times.” Later in the week Scott Meslow at The Week posted tips on how to avoid being fooled based on this story, but he himself was also fooled by the leftmost button. He reported that the story had been “shared more than five million times on Facebook.” As you can see by my screen shot, that button was displaying six million by Friday.
And yet, none of these numbers are correct!
I checked the numbers at SharedCount on November 14, just after Meslow’s story but well after Waterson’s, and the actual numbers at that time were:
- Likes: 44,671
- Shares: 44,947
- Comments: 143,311
- Total: 232,929
This story was only shared on Facebook about 45,000 times! That’s about 100 times smaller than the number reported by The Week, and 15 times smaller than the number on Buzzfeed.
The standard Facebook engagement buttons (and variations such as the ones on WordPress) inflate the number they show you. The number shown include likes, shares and comments all added together. I’m sure Facebook has their reasons for thinking this better reflects total “engagement.”
But for most content, the number wildly overestimates how viral something is. This is especially true when you consider that people often “like” things on Facebook for reasons other than agreeing with it, and often comment on things to dispute them.
Before reporting how viral something is, always use a neutral tool like SharedCount.com to double check the numbers direct from the Facebook API. And report only the share count – don’t include comments and likes.
And for crying out loud, if you think the site itself is posting fake news, don’t even consider using engagement numbers that were reported to you directly from that same site! Why should you trust a button if you don’t trust the text right next to it?