Ferreting out Facebook falsehoods

Somini Sengupta / New York Times News Service /

SAN FRANCISCO — The Facebook page for Gaston Memorial Hospital, in Gastonia, N.C., offers a chicken salad recipe to encourage healthy eating, tips on avoiding injuries at Zumba class, and pictures of staff members dressed up at Halloween. Typical stuff for a hospital in a small town.

But in October, another Facebook page for the hospital popped up. This one posted denunciations of President Barack Obama and what it derided as “Obamacare.” It swiftly gathered hundreds of followers, and the anti-Obama screeds picked up “likes.” Officials at the hospital, scrambling to figure out who was behind it and how to get it taken down, turned to their real Facebook page for damage control. “We apologize for any confusion,” they posted on Oct. 8, “and appreciate the support of our followers.”

The fake page came down 11 days later, as mysteriously as it had gone up. The hospital says it has no clue who was behind it.

Fakery is all over the Internet. Twitter, which allows pseudonyms, is rife with fake followers, and has been used to spread false rumors, as it was during Hurricane Sandy. False reviews are a constant problem on consumer websites.

Gaston Memorial’s experience is an object lesson in the problem of fakery on Facebook. For the world’s largest social network, it is an especially acute problem, because it calls into question the site’s basic premise. Facebook has sought to distinguish itself as a place for real identity on the Web. As the company tells its users: “Facebook is a community where people use their real identities.” It goes on to advise: “The name you use should be your real name as it would be listed on your credit card, student ID, etc.”

Fraudulent “likes” damage the trust of advertisers, who want clicks from real people to whom they can sell and on whom Facebook now relies to make money. Fakery also can ruin the credibility of search results for the social search engine that Facebook says it is building.

Tackling the problem

Facebook says it has always taken the problem seriously, and recently stepped up efforts to cull fakes from the site. “It’s pretty much one of the top priorities for the company all the time,” said Joe Sullivan, who is in charge of security at Facebook.

Sullivan declined to say what portion of the company’s now 1 billion-plus users are false, duplicate or undesirable. The company quantified the problem in June, in responding to an inquiry by the Securities and Exchange Commission in the process of going public. At that time, the company said that of its 855 million active users, 8.7 percent, or 83 million, were duplicates, false, or “undesirable,” for instance, because they spread spam.

Sullivan said that since August, the company has put in place an automated system to purge fake “likes.” The company said it has between 150 and 300 staff members who apply machine learning and human skills to weed out fraud.

Flags are raised if a user sends out hundreds of friend requests at a time, Sullivan said, or likes hundreds of pages simultaneously, or — most obvious of all — posts a link to a site that is known to contain a virus. Facebook users are sometimes asked to verify their friends’ profiles. “Is this your friend’s real name?” they are asked. Suspected fakes are warned. Depending on what they do on the site, accounts can be suspended.

In October, Facebook announced new partnerships with anti-virus companies. Facebook users can now download free or paid anti- virus coverage to guard against malware.

“It’s something we have been pretty effective at all along,” Sullivan said.

Keeping ‘likes’ legitimate

Facebook’s new aggressiveness toward fake “likes” became noticeable in September, when brand pages started seeing their total number of fans dip noticeably. Rihanna lost 22,000 on one day, out of 60 million, according to an analytics company, PageData. An average brand page, Facebook said at the time, would lose less than 1 percent of its fans.

“When a page and fan connect on Facebook, we want to ensure that connection involves a real person interested in hearing from a specific page and engaging with that brand’s content,” Facebook wrote in a blog post.

The election season seems to have swelled the market for fakery.

In Washington state, two groups fighting over a gay marriage referendum locked horns over “likes” on Facebook. The pro-gay marriage side pointed to the Facebook page of its rival, Preserve Marriage Washington: It collected thousands of “likes” in a few short spurts, and during those peaks, the pro-gay marriage group said, the preponderance of them came from far-flung cities like Bangkok and Vilnius, Lithuania, whose residents would likely have little reason to care about a state referendum in Washington. The “likes” then fell as suddenly as they had emerged, as though they had been purged.

The accusations were leveled on the pro-gay marriage group’s website. Preserve Marriage Washington in turn denied them on its Facebook page: “We have told our vendors explicitly, ‘Do not buy likes.’ We are investigating these claims.” A spokesman for the group, Chip White, said this week that he had nothing more to add. Facebook declined to comment on the contretemps.

The research firm Gartner estimates that while less than 4 percent of all social media interactions are false today, that figure could rise to more than 10 percent by 2014. The temptations are too great, Gartner argued, as brands compete for popularity.

This image is copyrighted.