Lots has been written in the last few days about tightening airport security, so I figured that I would add my thoughts.
Airport screening serves two goals.
(1) It is the last line of defense. It is meant to detect and prevent terrorists from bringing the tools of their trade onto aircraft.
(2) It is a deterrent. It is meant to make it more complicated or difficult to bring those tools on the aircraft, thereby influencing terrorist to seek softer targets.
Neither of these goals can realistically ever be achieved with 100% effectiveness. (Well, maybe number one if we fly naked after full body scans. Gross, and I'm critical of people who fly in shorts. I don't see how they're good with bare skin where hundreds of people sit each day - in the airports and on the planes. Do you want to sit naked where the last stranger sat naked? I don't. But I digress.)
The perceived effectiveness of goal one directly influences the real effectiveness of goal two. That is, the better they think we are at screening, the higher deterrent value it has.
So how can we increase the percieved effectiveness of the screening? We must be more targeted in our approach. 100% scans are fine where feasible, but let's face it - the most recent terrorists have several things in common: they are young, ages 20-40. They have a history of international travel - often to Yemen or Pakistan. They've also spent time in London. (If they haven't traveled there, they've got contacts.) They come from middle class or even wealthy families. Oh yeah, they're hold the view that Islam is at war with America and they are soldiers in that war.
This works because of how sampling and probability work. Let's say 5000 people a day pass through my checkpoint. I have resources to thoroughly screen 50 passengers a day. If I randomly select those 50 from all 5000, a terrorist has a 1% chance of being selected for follow on screening. He knows that and it looks like that's an acceptable level of detection risk. If I select my 50 from the 100 or so men who match the profile, assuming it remains stable, I've increased my chance of detection to 50%. Is that risk of detection still acceptable? The terrorists will then have to start looking for people who don't fit the profile, women, children, and old men, but that in itself contributes to the deterrent effect, acheiving goal number two. For them to change the profile is harder than for us. Martyrdom doesn't appear to be as attractive to women and old men.
I keep hearing that in running security for El Al, the Israelis don't look for the bomb as much as they look for the terrorist. We should do the same.