Future Tense

Could the #10YearChallenge Really Improve Facial Recognition Tech?

Could you exploit the #10YearChallenge to make facial recognition more sensitive to age differences?
Could you exploit the #10YearChallenge to make facial recognition more sensitive to age differences? DAVID MCNEW/AFP/Getty Images

Over the past week, the #2009vs2019 meme challenge, alternately known as the #10yearchallenge and #HowHardDidAgeHitYou, has become the latest social media trend ripe for think piece fodder. While the challenge inspired a host of discussions about social media narcissism and gendered norms, author and consultant Kate O’Neill put her own spin on the meme in a tweet raising the privacy implications of posting age-separated photos of oneself on Facebook.

The post generated enough buzz and discussion on Twitter that O’Neill expanded it into an article in Wired, in which she argued that Facebook or another data-hungry entity could exploit the meme to train facial recognition algorithms to better handle age-related characteristics and age progression predictions. She noted that the clear labeling of the year in which the pictures were taken, along with the volume of pictures explicitly age-separated by a set amount of time, could be quite valuable to a company like Facebook. “In other words, thanks to this meme, there’s now a very large data set of carefully curated photos of people from roughly 10 years ago and now,” O’Neill wrote.

While O’Neill’s perspective seems to have elicited generally favorable reactions on Twitter, some tech journalists and commentators questioned the premise. Writing in New York magazine, Max Read argues, “If you’re one of the 350 million people or so who’s been on Facebook since 2009—or if you’ve uploaded older photos to the platform after joining—the world’s biggest social network already knows what you look like now, in the past, and probably in the future, too.” Indeed, Facebook automatically collects data from pictures that users upload for its facial recognition features. You actually have to opt out of the service to keep this biometric info private. Read further posits that if Facebook were to engineer a meme trend like #2009vs2019, the goal would more likely be to boost its stagnant user engagement numbers, rather than to improve its facial recognition software.

Facebook has also denied that this challenge was part of a larger scheme:

To be clear, everyone in this debate agrees that there’s no evidence at all that Facebook engineered this meme trend with the goal of harvesting data for its facial recognition software. O’Neill writes that her tweet was meant to be “semi-sarcastic” and that the point of the article was not to specifically accuse Facebook of any impropriety, but rather to “raise awareness about the importance of human data and how it can be used.” So it’s best to evaluate O’Neill’s postulate as a thought experiment. Would the #2009vs2019 challenge be particularly useful to Facebook’s facial recognition software?

Most companies that employ facial recognition technology, including Facebook, find it useful for matching images of people with their identities. Quantity is usually more important than age separation in this case. “For an entity like Facebook … the more images of a person you can get, the more effectively you can build a profile for them,” says Jake Laperruque, who serves as senior counsel at the Constitution Project and does work with facial recognition and privacy. “[Facebook] already has a mass database of photos, which is what they really want.” He also noted that recent pictures would be more useful.

The question becomes more complicated if we take as a given that Facebook is developing age recognition and progression technologies. Theoretically, the data extracted from this meme could help such technologies become more sophisticated. “Any time you have multiple pictures of an individual in different poses, indoors and outdoors, that’s really useful,” says Anil Jain, a computer science professor at Michigan State University who studies biometric recognition. “But to make the system tolerant to time differences, we also need multiple faces of people at different ages.” Jain adds that this would be most useful for identifying other old pictures of a person, as face recognition technologies tend to see a significant dip in performance in trying to match images of people taken more than seven years apart.

But how much would this actually help in Facebook’s case? Hard to say. In truth, we don’t really know much about any specific age-related facial recognition initiatives at Facebook. Alexis C. Madrigal writes in the Atlantic, however, “Facebook isn’t building an age-progression machine-learning system; it almost certainly already has one.” The idea, again, is that having people post age-separated photos in this limited meme context really isn’t going to augment what’s already possible with Facebook’s massive trove of user pictures.

We do know that Microsoft developed a site in 2015 that tried to guess the ages of people who submitted photos, though there were reports that it was missing some ages by decades. A startup in the U.K. is also currently piloting a facial recognition system at grocery stores that can supposedly detect whether a customer is old enough to purchase alcohol, but the results of the trial have not been publicized.

While the plausibility of a #2009vs2019 facial recognition scheme is still under debate, the impulse to be wary of biometric data collection is a step in the right direction. Says Laperruque, “People are taking the right approach to start wondering about these things when they give companies their data and biometric information, but Facebook has been without permission opting individuals into their facial recognition for many, many years.”