The Associated Press is reporting on the Department of Homeland Security’s Automated Targeting System (ATS), which assigns international airline passengers “risk scores.”
All international airline passengers do not pose an equal terrorism threat. In principle, risk scores help allocate scarce homeland security resources on more serious threats. as with all scoring systems, risk scores provide rankings that are categorical (e.g., “low,” medium, “high”) or ordinal (e.g., 1 to 100 points). Errors will occur. For example:
- Some travelers will be ranked riskier than they really are (“false positives”)
- Some travelers wll be ranked less risky that they really are (“false negatives”)
- Rankings may be systematically biased.
Privacy advocates generally object to all such systems unless, at a minimum, travelers can learn their ranking and protest errors. If such an appeal process could be devised (and for sake of argument, imagine that it was costless to implement), it almost certainly would solve error type #1, do nothing to solve error type #2, and increase error type #3. Those who believe their risk has been overstated will petition for correction, but those whose risks have been understated will not. The database will consist of two groups: (1) Those whose risks are accurately estimated and (2) those whose risks are underestimated.