Flock, Mass Surveillance, and the Collapse of Crime
The benefits of automatic license plate readers outweigh the dangers
There are now around 120,000 Flock cameras in operation across the US. These devices automatically read license plates and make their locations available to law enforcement. Flock now also makes it possible to look up cars by their descriptions, rather than just relying on plates. As one might expect, this has proved very useful in solving crimes. Just as predictably, it has caused a popular backlash.
Tucker Carlson is outraged, having made resistance to Flock cameras one of the messages he prints out on the hats he sells for allowing his fans to broadcast their low socioeconomic status. Groups that are worth taking more seriously, like the Institute for Justice and the ACLU, have also expressed concerns. How much should we worry? And where should we come down in this particular iteration of the old debate of liberty versus security?
Flock Cameras Likely Explain the Recent Crime Collapse
On the plus side of this technology, a 2023 study from the company itself estimated that Flock cameras are used in solving 700K crimes each year, or approximately 10% of all reported crime in the US. This study involved surveying law enforcement agencies that were Flock customers, ending up with 123 respondents that provided data researchers considered adequate for analysis. They were asked “How many arrests have you made that can directly be attributed to the use of a Flock Safety camera?” and asked to provide corresponding case records.
From there, the authors make broader estimates of society at large. More recent research from Flock boasts about last year seeing 1,000,000 “cases supported” and helping to locate approximately 10,000 missing persons. The company estimates that its technology was involved in roughly one in five cases cleared in its customer jurisdictions.
These are absolutely massive benefits. Can we trust research conducted by Flock itself? We should be skeptical, but keep in mind that a lot of academics investigate these questions with ideological biases going in the other direction.
One way we might be able to check whether the results are plausible is to look at the overall trajectory of crime in the US. One benefit of this measure is that it cannot be manipulated by Flock or its opponents. If, as the technology has become more pervasive, crime has gone down, this would be a reason to suspect that the results that the company is presenting are at least in the right ballpark. If, in contrast, crime has increased, we would have reason to doubt that law enforcement recently got its hands on a new highly effective tool.
Violent crime reached a peak around the early 1990s before falling dramatically over the next three decades. Murder and aggravated assault then jumped in 2020 during Covid and the Summer of Floyd, remained roughly flat in 2021, and began falling thereafter. That decline accelerated dramatically in 2023 and has continued through the latest 2026 data. Last year, the United States likely had the lowest murder rate since we started reliably recording such data in 1960, and 2024 to 2025 may have seen the largest one-year drop ever.
You would expect Flock cameras to have the biggest impact on auto theft, and here the results are particularly striking. In 2023, the crime reached a recent high. But in 2024, the Council on Criminal Justice, which relies on a sample of cities, estimated a 24% decline over the previous year. That was followed by an even larger drop of 27% in 2025, and a further 20% fall in the first half of 2026. Overall, this corresponds to a decline of over 50% in just three years.
There’s no reason this shouldn’t eventually get to approximately zero. I’m actually curious how anyone at all can get away with car theft in an area with Flock cameras everywhere. Even if a thief pulled over right away and changed the license plate out of sight, as mentioned before searches can now go by general description of a vehicle. The same is true of kidnapping when it involves a vehicle, which will be just about always. When I was at the University of Chicago, locals would sometimes jump out of a van and mug a student before getting back in and driving off, and the days of this type of crime should be numbered too.
Flock isn’t the only form of surveillance that has been increasing in recent years. There has also been, for example, the rise of private camera registries and the use of drones. Flock therefore probably shouldn’t get credit for the full crime decline, but it is likely that the combined effect of all surveillance technologies is the predominant cause of the post-Floyd collapse. This is far and away the leading theory as to what changed so quickly over the last few years.
Totaling the Benefits and Costs
So the benefits of Flock cameras seem high. What are the costs?
When The Free Press posted a video praising the technology, a Community Note popped up linking to a Washington Post report on cases of police using the cameras to stalk women in their lives. That note has now been changed, but you can read the article here. It reports a grand total of at least…50 cases of law enforcement officers accused of or charged with misusing automatic license plate reader (ALPR) systems, 46 of them involving Flock. Twenty-six involved officers spying on women who were in their lives or that they wanted to meet, including new partners of exes. The worst crimes that have come of this are two instances of assault where officers grabbed or pushed their partners while making threats.
The article includes suggestions for safeguards like making police log every search. While that sounds good in theory, in practice paperwork requirements tend to make government less efficient. Fifty known cases of stalking strike me as something that is barely worth mentioning in the context of plummeting crime rates. Moreover, these are all rogue actors, and they have been punished in cases where there is sufficient evidence that they’ve committed crimes, indicating that none of these instances are part of a larger plot to deprive large numbers of citizens of their rights.
The Washington Post article discusses the psychological suffering that women have felt as a result of being spied on by cops using Flock cameras, but this of course has to be read while keeping in mind that crime itself is also traumatizing, and we’re now preventing a lot more of it due to mass surveillance.
It’s true that there’s no reason to believe that we know about every instance of misuse, and we don’t know how many of the 700K crimes would have been solved even without Flock cameras. But let’s just imagine that the estimate of 700K crimes solved was too high, and the number of 50 abuses was too low, each by an order of magnitude. Then, we’d have Flock cameras solving 70K crimes a year, and facilitating 500 cases of abuse. We’re still not even close to the scales being balanced! You need all estimates we have about real-world effects to be off by about two orders of magnitude in the directions favorable to the critics of the technology to get anywhere near having a case for eliminating Flock cameras.
Actually, wait, after just writing that paragraph, I’ve now realized that the 700K crimes solved estimate was yearly, at a time when the cameras were less pervasive, while the 50 abuses represent an all-time number. There’s no need to go back and do this analysis over again, because this mistake actually helps the Flock skeptics. Who says I’m not a fair guy? I am bending over backwards to give them every possible benefit in the debate, and it only serves to highlight how divorced from any kind of empirical grounding the anti-Flock campaign is.
Moreover, all this assumes that each case of letting a crime go unpunished and each instance of Flock abuse are equally bad, or even that the abuses are worse. But imagine a man is curious what his ex-girlfriend is up to, finds her driving to some other man’s house, becomes discouraged, and then just moves on with his life. This is likely the modal case of Flock abuse. How many extra carjackers, kidnappers, and murderers on the street are we to accept in exchange for never allowing something like this to happen? Not only is the crimes solved-to-abuse ratio of Flock likely astronomically high, but once we weight the two kinds of harm by their severity, the case for embracing the technology becomes even stronger.
“But We’re Americans, Dammit!”
On X, I’ve seen some argue that costs and benefits is the wrong way to look at this. They occasionally say this is a matter of “constitutional rights,” which is wrong, since the Constitution simply prohibits unreasonable searches and seizures, and courts have historically not found it illegal to surveil people in public as part of regular law enforcement. The government doesn’t need a warrant to follow your car, for example. Some lawyer could make an argument that distinguishes Flock cameras, but it’s one of those things where your politics are likely to determine how convincing you think their case is. If you want to ban Flock cameras, you should be held to the same standards as people arguing for any other policy position, meaning that you need to present logic and evidence.
If you want to go by a general principle that it’s never worth sacrificing liberty for safety, that obviously can’t provide any guidance, since it would suggest a need to abolish the police. Here’s one tweet making such an argument.
Joshua Reed Eakle says that “in America we believe that liberty is more important than security.” Do we? This isn’t a statement I find meaningful. We clearly don’t maximize liberty in every possible situation, but rather usually consider tradeoffs. Most of us would be happy for the police to detain us for a few minutes in order not to be hacked to death. And nobody says get rid of jails because, even if we are sure most people there are guilty, some innocent person is surely being locked up somewhere, and we must always choose liberty over safety. What is relevant is what kind and how much liberty and security we’re talking about.
That is to say that there’s no way to avoid cost-benefit analysis in situations like this. They pretend this is about larger principles, but it seems that anti-Flock advocates are going based on their own mood affiliation, which has a bias toward the status quo and against new forms of technology. We barely register large, multiyear drops in the murder rate, even though they cumulatively represent thousands of lives saved. At the same time, a couple dozen stalking cases facilitated by Flock will become national news, and while regular victims of crime are collapsed into unanimated numbers, those victimized in a novel way get the complexities of their inner emotional lives put on full display. When we’re dealing with status quo problems, we express harms in terms of statistics, while moral panics are written about in the language of poetry.
Cost-benefit analysis is a tool for making rational thought possible by judging both sides in a debate by comparable metrics. You’re not going to get far in a debate over whether one population is on average taller than another if one country records height according to the metric system and the other uses the imperial system, and nobody is aware of how measurements from one are converted into the other. The Flock debate is kind of like that. Or not really. It’s more like one side is using the metric system, and the other is just pointing and screaming about how they feel that Population A must be taller than Population B because it feels right.
Learning the Lessons of History
Crime is mostly a solvable problem. As mass surveillance has increased, it’s remarkable how little it has interfered with the day-to-day lives of law-abiding citizens. This suggests that a model of tyranny that sees expansive police power as a threat to liberty has fallen short in terms of explaining the world. Checks on abuses can come through mechanisms like democratic accountability, a free press, and power distributed within government across prosecutors, judges, and inspectors general and others tasked with rooting out abuses and corruption. I see little to indicate that the instincts of technophobic activists, conspiracy theorist podcast hosts, and those they rile up have much to contribute to the project of preventing tyranny.
Before the Warren Court, state and local law enforcement was not subject to many of the constitutional restrictions that we take for granted today. Many of the rights now associated with the Fourth, Fifth, and Sixth Amendments had either not yet been applied to the states or were enforced much less stringently. State courts could often admit evidence obtained through unconstitutional searches, and police could interrogate suspects in custody without first informing them of a right to remain silent or to have a lawyer present. Some states did not even provide attorneys to indigent defendants. It was only in 1949, in Wolf v Colorado, that the Supreme Court held that the Fourth Amendment protection against unreasonable searches and seizures applied to state and local governments. Even after that, the Court did not require states to exclude illegally obtained evidence until Mapp v Ohio (1961).
Most Americans in the 1930s, 1940s, or 1950s did not believe they were living in a fascist state because police were relatively unrestrained by modern standards. If you’re going to compare how much extra Americans had to fear law enforcement before the Warren and Burger courts to how much more they came to be threatened by crime after the judiciary had done its work, it’s not even close.
History suggests that, in a well-functioning advanced democracy, it is possible to give law enforcement strong surveillance powers and wide discretion in trying to solve crimes without causing much undue harm to innocent citizens. The overwhelmingly positive benefit-to-cost ratio that we see in the case of the use of Flock cameras is yet another demonstration of this.
Those worried about government abuses of power should stop targeting surveillance itself, and instead think about how to make sure it doesn’t go wrong. Our only choices aren’t to either trust government completely or blind it so it can’t maintain public order.
American law enforcement already has safeguards to prevent abuse, and they can always be made better. Police departments have formal complaint and internal-affairs processes. Depending on the jurisdiction, there may also be inspectors general or oversight review boards. Citizens can sue law enforcement over constitutional violations, and defense lawyers are able to challenge evidence that was obtained illegally. None of these mechanisms is foolproof, but they seem to generally work well, as can be seen in the fact that the vast majority of Americans who aren’t criminals have practically no fear of being mistreated by law enforcement. All of this suggests that the sensible response to the possibility of abuse is to strengthen accountability and oversight rather than deprive the government of useful tools for enforcing the law.
This post is sponsored by Mechanize. They are hiring software engineers at a base salary of $300K-$400K/year, plus substantial bonuses. They also have openings for other positions, including Research Engineer ($500K base pay), Recruiter ($200K), Growth Manager ($250K), Interns ($150/hr), and legal counsel ($400K). You can even “build your own role.” Here is their website, and you can apply here. If you apply, make sure to use the links provided here so I can receive credit, and name me as a reference in the process.