In 2014, I began to notice a curious Facebook trend: drag queens posting screenshots wherein the platform’s facial recognition incorrectly tagged their photos as one another. As a drag queen myself—who’d recently organized a campaign challenging Facebook’s “real names” policy—I found these posts both comical and intriguing: they both shadily read other queens’ appearances and poked fun at Facebook’s own technological failures. And while most of my #MyNameIs work focused on securing Facebook access for marginalized communities, I was also curious what value there might be in using the tricks of drag’s trade to purposefully play with assumptions about identity, authenticity, and reality. Put another way, rather than fighting to make Facebook more inclusive, what possibilities might exist if all users looked and acted like fabulously messy drag queens?
In this project, I propose drag as both a conceptual framework for rethinking the users and uses of popular technologies, as well as a set of speculative practical techniques for mitigating the harms sustained by data-driven surveillance. Though drag is often conceived as a performance that transgresses gender binaries, I push that definition to consider how it disrupts dichotomies of visibility/concealment, truth/fiction, and mutability/consistency. In a digital media context that assumes users to have trackable interests, behaviors, and faces, drag offers a radically different approach to privacy that keeps information “hidden in plain sight”—and always with a knowing wink of a heavily made-up eye.
I specifically explore how drag makeup functions as a tactic to confuse facial recognition algorithms through contouring and embellishing facial features like cheeks, chins, noses, and lip and lash lines. In an initial phase of the project I have produced a dataset of 25 drag queens, each photographed in three separate looks: 1) non-drag, 2) a typical drag look for that performer, and 3) a consistent look inspired by the drag icon Divine that each performer recreates. I upload portraits to Facebook and other platforms to test their efficacy and explore whether specific techniques generate incorrect tags or false matches.