Yesterday, we published an item based on a former Newt Gingrich staffer's claim that Gingrich assembled his 1.3 million Twitter followers—a number that he's taken to bragging about—in part by buying fake Twitter followers. A lot of people did not think that was true! But today social networking search firm PeekYou announced that it had crunched the data and come to the conclusion that roughly 106,055 of Gingrich's million-plus followers are real people. The rest are fakes.
Our source yesterday told us that about "80 percent of [Newt's followers] are inactive or are dummy accounts created by various 'follow agencies'" paid by his campaign. That made sense to us largely because Gingrich's 1.3 million followers seems oddly inflated compared to, say, Sarah Palin's 600,000. But many folks quickly, and quite reasonably, pointed out that Gingrich was one of the first GOP politicians to be listed on Twitter's suggested user list—a perch that guarantees a torrent of Twitter followers and is as likely an explanation as any for Gingrich's high follower count. Of course, that doesn't disprove the claim that he buys fake followers—both can be true. The only historical data we could find on Gingrich's followers goes back a year, and—except for a small bump that coincides with the launch of his campaign—he's held pretty steady over that time, so we don't know whether there are any suspicious spikes in followers prior to that or whether they all came rushing when he was added as a suggested user in 2009. Short of direct evidence of follower-buying, there's not much more we can do to corroborate our source's claims aside from actually pore over all those 1.3 million followers looking for fakes.
As it turns out, the search firm PeekYou has done just that! And it doesn't look good for Gingrich. In a happy accident, PeekYou—which is sort of a people-oriented search index—has for the past year been conducting research on how to measure the quality of Twitter audiences, a project that included looking at politicians' Twitter followers. And by their count, just 8% of Newt Gingrich's followers are real people.
"We just started running the 2012 candidates' numbers three weeks ago," said Josh Mackey, PeekYou's general manager of business and product development, "and when we saw your story, we went back to pull the Gingrich numbers. The huge majority of his followers are either completely anonymous people who have no other web presence, or they are spambots."
Mackey said PeekYou actually scrubbed each and every one of Gingrich's 1.3 million followers, using 23 criteria—including name, location, and inbound and outbound links in their feed—to determine whether they were real people. "We usually find out that real people have real web identities," he says. For the vast majority of Gingrich's followers, that wasn't true. They were either business accounts, private accounts, anonymous accounts that had only a user ID and no other discernible connection to the internet, or spambots. The average Twitter user, Mackey says, has a follower count that consists of anywhere from 35% to 60% real people. At 8%, Gingrich's is the lowest PeekYou has ever seen. "When was saw it, we actually had our quality assurance people go over the numbers for two days to doublecheck," he says.
Using algorithms to determine whether an online presence is real or fake is obviously more art than science, and sometimes an anonymous or private account that leads nowhere else on the web is still real. "It's hard to determine what's real," says PeekYou CEO Michael Hussey. "We're looking at digital footprints. We're not perfect. We'll miss some people." But even if their algorithms are off, you'd expect them to be off by the same measure for the other politicians they've looked at—but Gingrich's count is way out of the ordinary, Hussey says. "The ratio speaks for itself." The firm is preparing to release similar data on other 2012 candidates.
Does this constitute hard evidence that Gingrich buys fake Twitter followers? No. Does it constitute hard (or at lease substantive) evidence that almost all of Gingrich's followers—the ones that he thinks make him a formidable presidential candidate—are fake? Yes. Of course, our Gingrich source was indisputably wrong about one thing—he claimed that 80% of Gingrich's followers are "inactive or are dummy accounts." The real number, it seems, is 92%.
UPDATE: PeekYou sent over its analysis of the other GOP 2012 contenders for comparison. Whereas Gingrich rates 8% real followers, Sarah Palin is closest with a 20% ratio of real followers, by the firm's analysis. Mitt Romney has 26%, Michele Bachmann 28%, and Tim Pawlenty 32%. In other words, Gingrich has by far a higher proportion of fake accounts following him than any of his competitors.
Meanwhile, the mellifluously named Center for Complex Networks and Systems Research at Indiana University's School of Informatics and Computing has analyzed a random sample of 5,000 of Gingrich's followers—as opposed to the full 1.3 million, which PeekYou did—and comes up with data that, while not as dramatic as PeekYou's, seems to support the contention that most of Gingrich's followers are fake. Fully a third of the CCNS sample have never posted to Twitter, and 76% had added no biographical information to their profile. Half had the default profile image. Keep in mind that the PeekYou weeded out business and brand accounts as not real people, which CCNS didn't and would change the ratios.
[Photo via AP]