people walking on the strain station
You may have to prepare for a different commute once school begins. Credit: Miles Rothoerl

After much gnashing of the teeth and public criticism, Mayor Eric Adams this week appears to be moving forward with the rollout of so-called gun detection scanners manufactured by Evolv Technology in certain subway stations around the city, calling it a pilot program. In theory, the machines work like metal detectors that can specifically detect weapons like guns and knives, as opposed to just any piece of metal.

Now I have to be upfront about something: I have a somewhat personal stake in this story, and that’s because I was the person who wrote the article for Hell Gate revealing the results of the city’s 2022 Evolv pilot program at Jacobi Medical Center in the Bronx, which I had obtained via a records request. The city had not previously released the results, and Adams didn’t even mention this earlier test during his big announcement of the new subway pilot in late March. (As an aside, City Hall also strenuously denied to me that Evolv had already been selected as the vendor for this program despite Adams showing off one of their scanners during this launch; I doubt there was much of a competitive process.)

The results weren’t just mixed, they were terrible. Over the seven months that the devices were installed at the entrance to Jacobi, they threw up alarms for over a quarter of the nearly 200,000 times someone passed through them, an incidence just half as likely as flipping a coin. The vast majority of these alarms turned out to be false positives, with just about 15% correctly identifying that someone had a weapon; of these, almost all were law enforcement. In the end, only 295 alarms out of 50,000 over seven months turned out to be non-law enforcement weapons, which is just over half a percent of total alarms. As Daniel Schwarz, senior privacy and technology strategist for the New York Civil Liberties Union, told me at the time, “it’s essentially a coincidence.”

person wearing black top and red bandana walking on subway
Photo by Ramil Ugot on Pexels.com Credit: Ramil Ugot

If you wanted to take a more dim view of it, you could say that the scanners were just a very expensive and high-tech excuse for the authorities to search whoever they wanted on the pretense that the machine had signaled them out, even though it was “singling out” one in four people. Basically, no different than the NYPD and National Guard at subway entrances with folding tables randomly grabbing people to poke through their bags, except the scanners cost $2,500 a month each and have the veneer of technological certainty.

The data I got were the results of the only long-term test of this specific technology in NYC (the scanners were also briefly put at City Hall but the NYPD denied my records request for that and I believe they were only up for less than a month). But the scanners have been tested plenty elsewhere, and the results have been similarly discouraging, though as far as I can tell what I got are the most granular real-world use statistics around. 

It wasn’t just my story that cast doubt on this program. The Daily News discovered that Evolv’s own CEO had said “subways in particular are not a place that we think is a good use-case for us,” just two weeks before Adams’ announcement in an investor call. An investigation by WIRED surfaced links between Adams administration officials and Evolv executives, as well as the fact that the company had dazzled City Hall in part by referencing its work with Disney. Evolv was sued by its own shareholders, who argued that the company misrepresented its technology and capabilities. The company is under investigation by both the Securities and Exchange Commission and the Federal Trade Commission.

Perhaps a decade of reporting on local policy have not left me jaded enough quite yet, because I really, truly thought that there was no way that after all this — the shady connections, the CEO straight-up saying this was a bad use case, the technology horribly failing in a NYC test, in black-and-white numbers — the city would move forward. Forget the surveillance state angle, which longtime readers will know is always a concern for me. Forget the logistical difficulties of trying to make everyone go through the machine at a busy subway station during rush hour, and the misunderstandings, altercations, and escalations that are sure to flow downstream from that. It doesn’t work. It simply does not perform its sole intended function. We already tried it! I mean, what are we even doing here? Shouldn’t this be the end of the conversation?

I guess you can try to argue that the technology has changed in the intervening couple of years, though I certainly haven’t seen any evidence to that effect. In fact, all evidence I’ve actually examined points to the thing not working. In a lot of ways, this reminds me of the so-called “Quadro” tracker, a 1990s device that was essentially just a hunk of empty plastic with an antenna poking out but which was nonetheless purchased by school districts, police departments, and other entities nationwide as some kind of drug- and bomb-detecting miracle device.

person walking on road
The machines are supposed to work like metal detectors. Credit: Matthis Volquardsen

The makers of the device were eventually indicted, but a version of it popped back up internationally, literally assembled in a shed by a couple in England after they bought cheap parts from China. From there, the devices made their way to active use in Iraq, Mexico, Thailand, and elsewhere, used in life-or-death security situations despite the fact that they were — and I really cannot stress this enough — basically empty hunks of plastic, a fact that police and security services around the world would have instantly noticed if they just bothered to open one up. They paid real money for these. Hell, they basically staked their lives on them.

Now, before the lawyers perk up, let me say explicitly that I am not claiming that Evolv is a fraud or that its tech is at the level of the Quadro stuff. It does seem to be real technology, using a mixture of standard electromagnetic detection and machine learning to try to specifically identify weapons. It just doesn’t seem to be particularly good at this, and I fail to see how this can be billed as a net boon for New Yorkers or even really justified given the cost and potential for negative externalities.

It’s pure security theater, and I’d argue that security theater isn’t often really that popular. People were pretty ambivalent about the National Guard deployment, especially given that crime in the subways really isn’t particularly high. If anything, I wonder if this in tandem with the the death of congestion pricing and the attendant collapse of billions of dollars in investments into MTA infrastructure will just highlight for riders that their leadership’s priorities are a little out of whack. Now it also seems like City Hall might be doing it in violation of a city privacy law that required the NYPD to roll out a use policy after public comment. None of that seems to matter, but it should.

Visit here to see more of our coverage on the subways.

Felipe De La Hoz is an immigration-focused journalist who has written investigative and analytic articles, explainers, essays, and columns for the New Republic, The Washington Post, New York Mag, Slate,...

Leave a comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.