Tag: behavior detection

Behavior Detection as Interrogation

With the Department of Homeland Security constantly spinning out new projects and programs (plus re-branded old ones) to investigate you, me, and the kitchen sink, it’s sometimes hard to keep up. But I was intrigued with a report that behvaior detection officers are getting another look from the Transportation Security Administration. Behavior detection is the unproven, and so far highly unsuccessful (Rittgers, Harper), program premised on the idea that telltale cues can reliably and cost-effectively indicate intent to do harm at airports.

But there’s a new behavior detection program already underway. Or is it interrogation?

Due to a bottleneck at the magnetometers in one concourse of the San Francisco airport (no strip-search machines!), I recently had the chance to briefly interview a Transportation Security Administration agent about a new security technique he was implementing. As each passenger reached him, he would begin to examine the traveler’s documentation and simultaneously ask the person’s last name. He confirmed to me that the purpose was to detect people who did not immediately, easily, and accurately respond. In thousands of interactions, he would quickly and naturally learn to detect obfuscation on the part of anyone carrying an ID that does not have the last name they usually use.

As a way of helping to confirm identity, it’s a straightforward and sensible technique. Almost everyone knows his or her last name, and quickly and easily repeats it. The average TSA agent with some level of experience will fluently detect people who do not quickly and easily repeat the name on the identity card they carry. The examination is done quickly. This epistemetric check (of a “something-you-know” identifier—see my book, Identity Crisis) occurs during the brief time that the documents are already getting visual examination.

Some people will not repeat their name consistent with custom, of course. The hard of hearing, speakers of foreign languages, people who are very nervous, people who have speech or other communication impediments, and another group of sufferers—recently married women—may exhibit “suspicious” failure to recite their recently changed surnames. Some of these anomalies TSA agents will quickly and easily dismiss as non-suspicious. Others they won’t, and in marginal cases they might use non-suspicious indicia like ethnicity or rudeness to adjudge someone “suspicious.”

The question whether these false positives are a problem depends on the sanction that attaches to suspicion. If a stutterer gets a gauntlet at the airport each time he or she fails to rattle off a name, the cost of the technique grows compared to the value of catching … not the small number of people who travel on false identification—the extremely small number of people who travel on false identification so as to menace air transportation.

We used this and closely related techniques, such as asking a person’s address or the DMV office where a license was issued, at the bar where I worked in college. It did pretty well to ferret out people carrying their older friends’ IDs. Part of the reason it worked well is because our expert doormen could quickly escalate to further inquiry, dismissing their own suspicions or denying entry to the bar very quickly. The cost of getting it wrong was to deny a person entry to the bar and sometimes possession of a license. These are relatively small costs to college students, unlike the many hours in time-costs to a traveler wrongly held up at the airport. According to my interview, suspicion generated this way at the airport requires a call to a supervisor, but I did not learn if secondary search is standard procedure, or if cases are handled some other way.

TSA agents are not doormen at bars, of course, and the subjects they are examining are not college kids out to get their drink on. These are government agents examining citizens, residents, and visitors to the United States as they travel for business and pleasure, often at high cost in dollars and time. The stakes are higher, and when the government uses a security technique like this, a layer of constitutional considerations joins the practical issues and security analysis.

I see three major legal issues with this new technique: Fourth Amendment search and seizure, the Fifth Amendment right against self-incrimination, and Due Process. When questioning joins an ID check at the airport, it’s a deepening of a search that is already constitutionally suspect. The Fifth Amendment issues are interesting because travelers are being asked to confess through their demeanor whether they are lying or telling the truth. It would seem to cross a Fifth Amendment line and the rule against forced self-incrimination. The Due Process issues are serious and fairly straightforward. When a TSA screener makes his or her judgment that a person is not responding consistent with custom and is therefore “suspicious,” these judgement calls allow the screeners to import their prejudices. Record-keeping about suspicion generated using this technique should determine whether administration of this epistemetric check violates constitutional due process in its application.

In its constant effort to ferret out terrorist attacks on air transportation, the TSA is mustering all its imagination. Its programs raise scores of risk management issues, they create constitutional problems, and they are a challenge to our tradition of constitutionally limited government. The threat that a person will use false identification to access a plane, defeating an otherwise working watch-list sytem, to execute some attack is utterly small. At what cost in dollars and American values do we attack that tiny threat?

The founding problem is the impetuous placement of federal government agents in the role of securing domestic passenger aviation. There are areas where government is integral to securing airports, airlines, and all the rest of the country—foreign intelligence and developing leads about criminal plots, for example—but the day-to-day responsibility for securing infrastructure like airports and airplanes should be the responsibility of its owners.

If the TSA were to go away, air security measures might be similar in many respects, but they would be conducted by organizations who must keep travelers happy and safe for their living. The TSA hasn’t anything like private airports’ and airlines’ incentives to balance security with convenience, privacy, cost-savings, and all other dimensions of a satisfactory travel experience. Asking people their names at airport security checkpoints is an interesting technique, and not an ineffective one, but it should probably be scrapped because it provides so little security at a relatively great cost.

TSA: If You Object to Giving Up Your Rights, We Should Take a Closer Look at You

TSA screeners and behavior detection officers may give you extra attention if you complain about security protocols (video at the jump). Former FBI agent Michael German sums up my feelings pretty well:

It’s circular reasoning where, you know, I’m going to ask someone to surrender their rights; if they refuse, that’s evidence that I need to take their rights away from them. And it’s simply inappropriate.

In related news, the GAO recently told Congress that the TSA’s Screening Passengers by Observation Technique (SPOT) is not scientifically grounded. The GAO testimony is available here.

More Cato work on TSA screening here, here and here.

GAO’s Damning Report on ‘SPOT’

Via the Identity Project’s “Papers, Please” web site, and despite my colleague David Rittgers’ excellent post from yesterday, I note last week’s utterly damning Government Accountability Office report on the SPOT program. “SPOT” stands for “Screening Passengers by Observation Techniques.” In the program “BDO’s,” or “Behavior Detection Officers,” observe travelers in airports, pulling them out of line if a secret list of behaviors signal that they’re a likely threat.

The thing is:

TSA deployed SPOT nationwide before first determining whether there was a scientifically valid basis for using behavior and appearance indicators as a means for reliably identifying passengers as potential threats in airports. … TSA state[s] that no other large-scale U.S. or international screening program incorporating behavior- and appearance-based indicators has ever been rigorously scientifically validated. While TSA deployed SPOT on the basis of some risk-related factors, such as threat information and airport passenger volume, it did not use a comprehensive risk assessment to guide its strategy of selectively deploying SPOT to 161 of the nation’s 457 TSA-regulated airports. TSA also expanded the SPOT program over the last 3 years without the benefit of a cost-benefit analysis of SPOT.

The Israeli airline El Al uses behavior detection, counters the TSA—as did DHS Secretary Janet Napolitano when I asked her about this report at a meeting of the DHS Privacy Committee Tuesday.

The GAO report notes that El Al’s processes, which are different from the TSA’s, have not been scientifically validated. As of 2008, El Al had 34 aircraft, operating out of one hub airport, Ben-Gurion International. There are 457 TSA-regulated airports in the United States. In 2008, El Al had passenger boardings of about 3.6 million; one U.S. airline, Southwest, flew about 102 million passengers that year.

From late May 2004 through August 2008, BDOs referred 152,000 travelers to secondary inspection. Of those, TSA agents referred 14,000 people to law enforcement, which resulted in approximately 1,100 arrests. TSA officials did not identify any direct links to terrorism or any threat to aviation in these cases. GAO noted its inability to determine if this is a better arrest rate than would occur under random screenings.

GAO also determined that at least 16 individuals allegedly involved in terrorism plots have moved at least 23 different times through eight airports where the SPOT program has been implemented. SPOT caught none of them.

The Government Accountability Office is a master of understatement, leaving conclusions for readers to draw. Mine is that the $1.2 billion in planned spending on the program over the next five years will be a wasteful producer of civil liberties violations.