The Zooms

In November of 2021, an anonymous source leaked 600 hours of police aerial surveillance footage to the non-profit transparency organization Distributed Denial of Secrets.

The videos are mostly from police helicopters in Dallas, Texas, and the total size of the collection comes to around 1.9 terabytes.

When I heard about the leak I purchased a large external hard drive and got to downloading. By chance this was the first video I looked at. In the footage, a police helicopter closely follows a pro-Palestinian protest as it snakes its way across Dallas.

Two things are immediately striking to me about the footage. First, I am amazed and disturbed by the level of detail that the camera is able to capture, by the proximity it achieves. The police helicopters are not merely tracking groups of people, they are nearly able to identify individual faces. Second, I am struck by the design of the interface itself: the sci-fi overlay of street names, borders, parcel numbers, target distances, and so on.

It’s as if the makers of this camera system have used the fantasy surveillance aesthetics commonly found in cop movies and TV shows as a design guide for their real system. They have transformed those clich├ęd surveillance tropes into reality.


In a way I was fortunate to stumble upon the protest first in those 600 hours of footage. Although the corpus does contain moments where “things happen,” the vast majority of it is a kind of aimless wandering, an aimless looking.

In my work I frequently try to find ways to reduce and filter vast troves of material. In this case I started to explore the idea of reducing the footage from the aimless wandering eye of the camera to the moments when something catches the eye’s attention. So I wrote a script that analyzed all 600 hours of footage, and automatically re-cut it to only include shots where the camera zooms in.

Here are the results:

I was expecting to discover moments similar to the ones in the protest footage. Moments that showed how the police were targeting particular groups and individuals. Instead what I discovered was that most of the zooming shots were as aimless as the panning shots: most seem to zoom in to nothing in particular.

As I watched this reduced footage, which now clocks in at one nauseating hour, and slowly realized that nothing was going to “happen,” I kept returning in my mind to the question of the interface itself. To how the fictional, fantastical cop-movie interface somehow became a reality.

This slip between fiction and reality seems to speak more broadly to the role of law enforcement itself, and to how self-mythologizing police narratives go on to shape the world. To put it another way, the footage is a product of the fantasy that the role of the police is to protect us from ubiquitous hidden danger, a fantasy generated in no small part by the police themselves.

It’s a fakery to help the police feel OK about themselves, and more importantly to coerce us into accepting the need for their existence in the first place.


Note: I’ve made the code I used to detect zoom-ins available on github. It works on panning shots too!