Suppose you produce automobiles and offer the purchaser the choice of different colors. Firebird Red is popular right now, so you order a hundred gallons of it. Will it be enough? That depends on its continuing popularity. If it is popular next week, why, we'll order another hundred gallons. They will arrive "just in time" to apply to the incremental car ordered in Firebird Red.
Except that the paint won't now arrive just in time if the truck transporting it is held up on the Canadian border for six hours. And it will be more expensive, because the truck driver's efficiency has been cut by one-third, as he sits in an immobile truck, waiting in line.
What does this tell us about the theory of security practices? What it tells us right away, and easily, is that the cost of security extends beyond the cost of the people hired to look into trucks crossing the border, searching for explosives or vats of poison. Does this inform us as regards security policy? Of course it does, because reasonable compromises need to be made. If we subjected everybody who wanted to fly somewhere to a strip search, we could be very confident that no weapons got aboard, and pretty confident that no passengers would want to fly. What is a reasonable level of security?
I think back to 1964, to J. Edgar Hoover appearing before the Warren Commission investigating the assassination of President Kennedy. Kennedy was shot by Lee Harvey Oswald. It was known that Oswald had been a Soviet sympathizer who had considered immigrating to Russia. Why hadn't Hoover, anticipating the journey to Dallas of Mr. Kennedy, detained Oswald, keeping him away from the presidential itinerary?
Hoover replied that if he had had to search out everyone in Dallas whose dependability was at the problematical level of Oswald's, he'd have had to round up 150 people. If the president were going to Chicago, Mr. Hoover said (the numbers here are from memory), he'd have needed to detain 450 people -- and "the American people wouldn't put up with it."
Let us, to make more graphic the theoretical point, stipulate that at Security Level 10, Oswald would be left alone. At Security Level 8, he'd have been brought in. At Security Level 12, we are allowed to check our baggage with the porter outside the airport. At Security Level 11, we have to take it in to the airport proper.
What we are feeling our way toward is an adjustment to an appropriate level of caution. At the plagued post office in Washington, Security Level 12 in handling mail seems insufficient, because two human beings have died. But what do we do if we change the security level to 10? Bring in dogs to sniff about before opening the mail sacks? Suppose the emergency worsened, to the point where mail was rejected unless it bore a sticker of some sort, tracing the bona fides of the mailer?
A statistical look at the current level of safety is certainly reassuring. And statistical probes can be greatly illuminating in practice. Somebody out in California a dozen years ago had a bright idea: 99.9 percent of vehicles crossing San Francisco's Golden Gate Bridge heading north are, at some point, going to cross the bridge heading south. So? Why not just let them go north unhindered by a toll bridge -- and double the toll coming south? Result: the same revenue, net, and half of the toll-takers let go to find other work.
The Golden Gate principle is, so to speak, actionable here. If permitting trucks to cross the Canadian border with pre-Sept. 11 security inspection increases the flow of (just what are we looking for in those trucks?) damaging materials to the point where we have a .00001 increase in Oklahoma City-style bombs, have we acted reasonably?
What security level can we live with? Just that one level higher that would have tripped up Lee Harvey Oswald? Certainly, a higher security level than obtained on Sept. 11. But homeland security chief Tom Ridge should lay it out for us, and let us decide, as we do every day in our own lives, what is the right level of security.