November 05, 2013, 3:26 PM —
Image credit: flickr/Nadya Peek
I know this may come as a shock, but: The Internet is a fakers paradise. Facebook estimates that roughly 7 percent of accounts – some 76 million – are phony. Twitter guesses 5 percent of its 200-million+ active users are bogus. (I’d lay money that both numbers are actually much higher.)
Of course, if you are one of the thousands hundreds 17 regular readers of TY4NS, you’d know that already. I’ve spent a fair amount of time over the last three years chronicling bots, sockpuppets, and other assorted fakers on Facebook, Twitter, and the Web in general.
But I may not have to do that much longer, thanks to a clever startup named BeehiveID, which promises to ferret out frauds using the magic of software algorithms.
Hive got a feeling
BeehiveID works by analyzing your social media accounts and assigning you a score, much like a credit score. The higher the number, the more likely you are a genuine person. You can look up your own score by visiting Beehive’s site and allowing it to connect to your Facebook account.
When I did it, I got the maximum score of 850. Turns out I’m a real boy after all. Who knew?
Beehive uses a number of techniques – including facial recognition and writing analysis – to parse thousands of data points about each person. Bots act differently online than real people, CEO Mary Haskett explained to me in a phone interview. Real people tend to have clusters of friends from both genders who live in the same town or work at the same place; fake accounts don’t. Real people tend to be more active than fakes and to write status updates and tweets in a consistent way. Real accounts also tend to produce a lot more data than bogus ones.
This week BeehiveID will announce its first customer, a dating service called TangoWire that pulls profiles from more than 50 other online dating communities. When you create a profile on TangoWire, you have the option of having your account “verified.” You can choose to use Beehive, which will analyze your Facebook account to determine if you’re real, or you can submit a photo showing you holding up a piece of paper with your handwritten account number on it.
Having your account verified is totally optional, says Haskett, but people with verified accounts tend to get more action and attention – so there’s a built-in incentive.
Honeypies and honeypots
How well does this actually work? I had to find out.
I created a fake account on TangoWire using my usual bogus Facebook personality, who underwent virtual gender reassignment surgery about six months ago. When I checked back a few hours later, I discovered that my profile could not be verified. Turns out that “Roberta” only got a BeehiveID score of 350 – below the minimum threshold of 400 – thanks mostly to the fact that she didn’t have enough activity on her Facebook account to rate as real.
(I also submitted a photo of model Raquel Zimmerman holding up a piece of paper with my account number Photoshopped onto it. That didn’t work either.)
Then I created a real profile using my actual information and tried to verify it using BeehiveID. But apparently I went too far; TangoWire spotted me trying to create multiple accounts of different genders using the same IP address and nixed them both.
Theoretically, if you’ve used a stock photo of Brad Pitt or Megan Fox as your profile pic on TangoWire, BeehiveID should be able to flag that. But it’s not a lie detector. For example, if you claim to be a buff 27-year-old surfer when you’re in fact a porcine 47-year-old couch potato, Beehive won’t be able to do much about that. Some fakes are harder to stop than others.
During its first six weeks in action BeehiveID has accurately spotted phonies on TangoWire more than 95 percent of the time, says Haskett. Dating sites are a natural place to start with this kind of software-based identity analysis, she adds, due to the inherent physical safety issues they present. You don’t want to go on a blind date with someone who’s pretending to be someone or something they're not. That’s like the plot of a bad movie on Showtime.
At the moment, BeehiveID is only using Facebook data, because that’s the most relevant source for a site like TangoWire. But Haskett says BeehiveID can perform similar analysis with Twitter, LinkedIn, and other accounts, and plans to expand well beyond dating sites.
The next logical step is to enter the sharing economy, says Haskett. Sites like Uber or Airbnb where you interact with total strangers are ripe for this kind of identity assurance. And then, of course, ecommerce sites where you want to make sure that the person on the other end of a transaction is in fact a real human, and not a sockpuppet or a bot out to defraud you.
“In the past, all of this was done via personal connections,” she says. “You are taking someone else’s word for the fact that I am who I say I am. The Internet totally messes that up. The way we usually do identity online is via an email address. Well, I can have a hundred email addresses and be a different person with each one. We figured there had to be a better way.”
Looks like there is. Internet fakes, your days may be numbered.
Got a question about social media or privacy? TY4NS blogger Dan Tynan may have the answer (and if not, he'll make something up). Follow him on Twitter: @tynanwrites. For the latest IT news, analysis and how-to's, follow ITworld on Twitter and Facebook.
Now read this: