Right, the data. So the proper way to do this would be to set up a massive A/B test, with everyone entering the US to be randomly assigned to one of two groups

Right, the data. So the proper way to do this would be to set up a massive A/B test, with everyone entering the US to be randomly assigned to one of two groups - either you're put into the hierarchical filter and you're profiled based on broad first-level parameters, or you're put into the fine filter and you're forcibly opted in into intensive surveillance for the next couple of years or so. Then measure the false positives, hits, and misses and take a decision. Seems to me that this step is probably the first of several by the US immigration to get this data and justify the move away from broad profiling (which they've taken flak for), towards the need for more intense profiling (which they're also taking flak for).