Libertarians Shouldn’t Want Perfect Security–Reply to Professor Epstein

I was pleased to see last week that Professor Epstein had penned a response to my criticism of his recent piece on Hoover’s Defining Ideas in which he argued against treating protection of civil liberties and privacy as "nonnegotiable" in the context of counterterrorism. It is not the disagreement that is pleasing, of course, but the opportunity to air it, which can foster discussion of these issues among libertarians while illustrating to the broader world how seriously libertarians take both security and liberty.

What’s most important in Professor Epstein’s rejoinder is what comes at the end. He says that I should “comment constructively on serious proposals” rather than take an a priori position that civil liberties and privacy will often impede expansions of government power proposed in the name of counterterrorism.

I believe that Professor Epstein and I share the same prior commitments--to limited government, free markets, and peace. Having left it implicit before, I'll state that I, too, believe that protection of life and property is the primary function of the state. But I also believe that excesses in pursuit of security can cost society and our liberties more than they produce in benefits.

Some years of work on counterterrorism, civil liberties, and privacy bring me to my conclusions. I had put in a half-decade of work on privacy before my six years of service on the Department of Homeland Security’s privacy advisory committee began in 2005. While interacting with numerous DHS components and their programs, I helped produce the DHS Privacy Committee’s risk-management-oriented “Framework for Privacy Analysis of Programs, Technologies, and Applications.” From time to time, I’ve also examined programs in the Science and Technology Directorate at DHS through the Homeland Security Institute. My direct knowledge of the issues in counterterrorism pales in comparison to the 30+ experts my Cato colleagues and I convened in private and public conferences in 2009 and 2010, of course, but my analysis benefitted from that experience and from co-editing the Cato book: Terrorizing Ourselves: Why U.S. Counterterrorism Policy is Failing and How to Fix It.

Whether I'm operating from an inappropriate a priori position or not, I don’t accept Professor Epstein’s shift of the burden. I will certainly comment constructively when the opportunity arises, but it is up to the government, its defenders, and here Professor Epstein to show that security programs are within the government’s constitutional powers, that such programs are not otherwise proscribed by the constitution, and that they cost-effectively make our society more secure.

The latter two questions are collapsed somewhat by the Fourth Amendment’s requirement of reasonableness, or "fit" between means and ends when a search or seizure occurs. And to the extent I can discern the program that Professor Epstein prefers, I have commented on it as constructively as I can.

Professor Epstein apparently believes that there is some kind of surveillance that is at once general, comprehensive, and non-intrusive. Adjoined to racial or ethnic profiling, it works to deter terrorism and more immediately to apprehend terrorists.

Let’s dispense with the profiling first. A “disproportionate number of terrorists are Muslim,” Professor Epstein wrote in his original piece, and in his rejoinder he said race and ethnicity are “valuable information that could help tweak the design of surveillance systems of searches.”

In 2006, IBM distinguished engineer and chief scientist Jeff Jonas and I published a paper entitled: “Effective Counterterrorism and the Limited Role of Predictive Data Mining.” We found that sophisticated algorithms based on variables far more salient than race or ethnicity will not turn up terrorism because there are insufficient instances of terrorism on which to build models that predict it. Sifting vast stores of data can turn up credit card fraud because there are thousands of examples per year on which to model this particular type of wrongdoing, and the costs of false positives are low--inconvenience to credit card customers, for example.

This is not the case with terrorism. “The possible benefits of predictive data mining for finding planning or preparation for terrorism are minimal,” we concluded. “The financial costs, wasted effort, and threats to privacy and civil liberties are potentially vast. Those costs out-strip any conceivable benefits of using predictive data mining for this purpose.” Data mining doesn't work in this area, and tracking Muslims doesn’t help.

So what about mass surveillance? In his original essay, Professor Epstein cited the need for “improved software in such key areas as facial recognition detection” and apparently the use of aerial drones for visual data collection regarding all potential terrorists, which is to say everyone. His preferred surveillance system would “collect troves of information about the conduct of people in public places,” he emphasized, “which can then be stored for future use.” The idea of a secret court system governing access to this data does not put the privacy-minded at ease.

As I also said in my original piece, these systems would not work. By “work,” I mean “cost-effectively secure the country and its inhabitants.” The number of ways to do damage and the number of “soft targets” in the country would require an utterly comprehensive national surveillance system. National security expert and Cato senior fellow John Mueller has distilled this fact to a jibe about the terror threat to the Weeki Wachee Springs water park in Florida. It's a symbol of the thousands of locations around the country all seeking homeland security money because they all could be attacked. They are all indeed potential targets, equally likely and unlikely, and it would take billions of dollars to implement the surveillance system Professor Epstein envisions if he wants visual surveillance to serve as a preventive or terrorist apprehension tool for them all.

As I said, doing so would not cost-effectively deter terrorists. And Boston is not the only guide. I have been pointing out since at least 2007 that terrorists are not terribly concerned with worldly justice and thus not as deterrable as ordinary people. Audrey Kurth Cronin shows in her excellent monograph "Ending Terrorism" that a goal of many terrorist acts is to signal other terrorists (imagined or real, and however pathetically). Notoriety is in the interest of terrorists to some degree. This helps explain lame escape attempts by the terrorists who don't do themselves in. Their relative indifference to capture given their mixed motives renders security measures like national identification and mass surveillance rather impotent in comparison to its effect on ordinary criminals whose goal is truly "getting away" with something.

We aren’t relitigating Boston here. Security guru Bruce Schneier had an excellent piece on CNN.com last week exploring how hindsight bias makes the Boston attack appear obvious when it was not. The problem after Boston is what will secure against the next attack, coming, as it will, from an unknown actor or actors, anywhere in the country, using a wide variety of tools and methods.

That’s a real security problem, though it’s far from existential. Terrorism is well under control and remains a far lower cause of mortality and morbidity in the United States than very exotic dangers like bees and fireworks accidents. Terrorism is indistinguishable from the x-axis in any reasonably sized cause-of-death chart that includes overeating and smoking. The objective facts about causes of death and property damage in the United States today counsel keeping terrorism in its place on the list of priorities, below automobile accidents, heart disease, domestic violence, and the Drug War.

Returning to the constitutional issues in the light of the security challenge, there is certainly such a thing as surveillance that doesn’t implicate the Fourth Amendment. But what Professor Epstein must be defending if he thinks this program is going to catch terrorists before they act is not in that zone. United States v. Jones last year showed that the Court is not going to embrace the facile conclusion descended from dictum in Katz (and rejected in the holding) that all bets are off simply because a person is in public.

When facial recognition software arrives to the point where it works at scale—it’s not there yet—the issue will be joined whether it is a search to use high-definition camera scanning of each person’s visage, highly tuned algorithms that convert each face to a signature, and time-and-location databases that record every person’s facial signature each time it is recognized outside the enclave of the home. “Stand-off” identity detection by laser-scanning of the iris holds out more hope for reliability, I think. I look forward to the case in which the government argues that these kinds of high-tech “examinations” are not searches.

Whether he has in mind the logistics of his surveillance system or not, Professor Epstein rejects my comparison of it to clearly unconstitutional crotch-searches in response to a nuclear threat because:

The Fourth Amendment treatment of unreasonable searches and seizures rests on a critical distinction between investigation of particular suspects and the stopping of dangers from unknown quarters. There is a lot more information in the first case, so that a dragnet search makes no sense, which is why particularized evidence is required. But general surveillance at unknown targets has to spread its net far wider. It is both less intrusive and more comprehensive, and it can and does work.

The distinction he talks about doesn’t arise from the language of the Fourth Amendment, which requires all searches and seizures of protected things to be reasonable, and I don’t know of case law relying on these poles. To my mind, he is simply describing two distinct points on the continuum along which suspicion and reasonable government action run. Courts administer the reasonableness of searches and seizures in a step-wise fashion, of course: Searching or seizing is unreasonable when suspicion is non-existent or low, a modicum of seizing and searching is reasonable when there is “reasonable suspicion based on articulable facts,” and a full warrant search is reasonable when facts, inferences, and identifiers about crimes and criminals have reached the level of probable cause.

If there is a difference in kind between the two situations, I don’t see how they are on opposite sides of the line. In both my crotch-searches-for-nuclear-weapons hypothetical and mass surveillance for terror threats, law enforcement has no idea who the bad actor is and no idea where he, she, or they will strike--if anyone ever does. The government’s power to search and seize our persons, papers, houses, and effects is low—and rightly so: Even great danger doesn’t justify essentially random government action.

What we are looking at post-Boston is well-summarized by the statement of Massachusetts state senator Robert Hedlund (R - Weymouth) that Professor Epstein attacked. Hedlund appears to be the rare example of a politician who avoids the a priori position that we can get more perfect defense against terrorist attacks if we just give more power to the government. It brings him to thinking that I find welcome, and that I think libertarians should generally endorse:

It’s not surprising that you have law enforcement agencies rushing out to use [the Boston bombing and subsequent manhunt] as pretext to secure additional powers but I think we have to maintain perspective and realize that civil liberties and the protections we’re granted under the Constitution and our rights to privacy, to a degree, are nonnegotiable. . .

You don’t want to let a couple of young punks beat us and allow our civil liberties to be completely eroded. I don’t fall into the trap that, because of the hysteria, we need to kiss our civil liberties away.