California's Ban Of Facial Recognition Tech Killed Off San Diego's Mostly Useless Biometric Program
from the nothing-of-value-was-lost dept
California's ban on use of facial recognition tech by law enforcement showed the state's government is willing to get out ahead of potential privacy issues. The tech is as popular as it is unproven. Law enforcement agencies strongly believe facial recognition will help it apprehend criminals more efficiently, but the available data simply doesn't back up this belief.
The tech is still pretty lousy at recognizing faces, kicking out false positives at an alarming rate. This also means it's not recognizing faces it should be recognizing -- the criminals and terrorists government agencies are convinced it will catch. They're also prone to bias, more likely to misidentify minorities, which results in increased targeting of demographics already heavily over-represented in most law enforcement agencies' paperwork.
California's new ban affects mobile tech used by state agencies. This includes body cameras, smartphones, and tablets. That was enough to kill the San Diego's multiple facial recognition programs. The San Diego PD used 1,300 cameras to build a database of 65,000 images over the course of seven years. These images were matched against a much larger database run by the San Diego Sheriff's Department
The program went into effect without the public being informed.
Introduced in 2012 by the countywide San Diego Association of Governments (SANDAG) without any public hearing or notice, the Tactical Identification System (TACIDS) gave law enforcement officials access to software that focuses on unique textures and patterns in the face—ear shape, hair, skin color—using the distance between the eyes as a baseline. In less than two seconds, the software compares those unique identifiers to a mugshot database of 1.8 million images collected by the San Diego County Sheriff’s office.
The San Diego PD used this database more frequently than any other agency with access to it, which included federal agencies like ICE. According to the PD, its increased access rate was mostly altruistic.
[T]he department used facial recognition scans more than 8,000 times in 2018, almost double the number in 2016, which it says was largely due to the formation of a new division, Neighborhood Policing Division (formed in March of 2018), aimed at addressing the issue of homelessness. SDPD equipped officers in the new division with TACIDS devices to help identify homeless people, who often do not have identification.
The SDPD did not say whether this use of the database actually resulted in correctly-identified homeless people. But the statement it did give Fast Company suggests its thousands of queries were mostly dead ends.
“While facial recognition could be a useful tool, we do not foresee the three-year moratorium on mobile face recognition use by law enforcement as having much of an impact on our department,” says Lieutenant Justin White, media relations director for the San Diego County Sheriff’s Department.
This statement is far more telling. While certain California law enforcement officials are bemoaning the ban, the San Diego PD must have already known the tech was over-hyped. If it worked as advertised, the local law enforcement agencies would have delivered a steady stream of success stories. But when something is a constant disappointment -- and you've decided you're going to use it anyway -- you avoid creating a paper trail.
Lieutenant White of the San Diego County Sheriff’s Department said they have not been tracking TACIDS successes in arrests and prosecutions, so they do not have any statistics. Neither does the SDPD, according to spokesperson Lieutenant Shawn Takeuchi.
That's the burial of bad news. There's no reason to give critics of sketchy surveillance programs any more ammunition than they already have to work with. The moratorium will have almost no effect on law enforcement in the state -- something that few agencies are willing to publicly acknowledge. They want the tech even though it has proven -- over the course of seven years -- to have added nothing to their ability to do their job.
Thank you for reading this Techdirt post. With so many things competing for everyone’s attention these days, we really appreciate you giving us your time. We work hard every day to put quality content out there for our community.
Techdirt is one of the few remaining truly independent media outlets. We do not have a giant corporation behind us, and we rely heavily on our community to support us, in an age when advertisers are increasingly uninterested in sponsoring small, independent sites — especially a site like ours that is unwilling to pull punches in its reporting and analysis.
While other websites have resorted to paywalls, registration requirements, and increasingly annoying/intrusive advertising, we have always kept Techdirt open and available to anyone. But in order to continue doing so, we need your support. We offer a variety of ways for our readers to support us, from direct donations to special subscriptions and cool merchandise — and every little bit helps. Thank you.
–The Techdirt Team
Filed Under: biometrics, california, facial recognition, san diego, surveillance
Reader Comments
Subscribe: RSS
View by: Time | Thread
Facial recognition tech needs to die already. It's obviously garbage and either way too immature -- not ready for prime time -- or never will be viable. The vendors selling this crap should be charged with false advertising and their wares relegated to the same fate as snake oil.
But this bit is still unproven, depending on how you define "over-represented":
Either the ratio of criminals by race does or does not match the rate of arrests/incarceration. I've seen no studies indicating either way. That doesn't slow down the overuse of this claim, however, when it suits a narrative. I'd love to see some actual science on this point but it's unlikely to happen simply because, by nature, it will be perceived as anti-minority. Just as this post probably will despite no such bias.
[ link to this | view in chronology ]
“Over-represented”, in this context, likely refers to the percentage of people in a given racial demographic who have been arrested/incarcerated. Some such demographics have proportionally higher percentages of arrests/incarcerations than others.
Per the ACLU, Black people and White people use marijuana at roughly the same rates, but Black people are about four times more likely to be arrested for possession than White people. In 2010, nationwide arrest rates for pot were 192 per 100,000 White people and 716 per 100,000 Black people. If the proportional percentages were equal, about 380,000 White people and 103,000 Black people would be arrested for using pot, as opposed to 380,000 White people and 385,000 Black people.
Saying “certain racial demographics are over-represented in the American criminal justice system” isn’t stating an unproven fact. It’s pointing out how the system is working exactly as intended.
[ link to this | view in chronology ]
But we tested it...
We loaded it with the headshots of all 20 white males in the office, and it's able to recognize them consistently... so that means it should work for everyone everywhere, amirite?
[ link to this | view in chronology ]
Yeah
Because it's hard to enter zero entries in a database. Gotta have at least one. ;)
[ link to this | view in chronology ]
Facial recognition on TV
It does not help that TV "law enforcement" type shows portray facial recognition as perfect, leading the general public to believe that this is so. This has also been the case with identification of partial fingerprints and other CSI info. Of course, on TV, all cops are honest and extremely competent, too.
[ link to this | view in chronology ]
Re: Facial recognition on TV
There's a LOT on crime shows that are little more than wishful thinking. Prosecutors and defenders alike hate it - prosecutors because people think there should always be "infallible proof" of several sorts for every crime, and defenders because people think that many types of "evidence" are infallible and perfect proof of guilt.
[ link to this | view in chronology ]
FU Nancy
[ link to this | view in chronology ]
This is good news. I think the officials who sponsored the biometric program in San Diego just don't know what it really is. They could read at least this article before making a decision. Sadly, but I think that sooner or later they will put this system into operation (despite the ban).
[ link to this | view in chronology ]