Terrorism continues to be a global threat but measuring the extent of risk necessitates a more subtle approach, writes Bart Kosko
Deaths due to terrorism have increased lately (especially after the Beslan attack), but the number still remains on the order of about a thousand deaths a year, according to US State Department figures – a small fraction of the 15,000 or so murders each year in the US or the 40,000 car accident deaths there.
The Bush administration and many others interpret these facts as proof that the government is winning its "war on terror" (though Osama bin Laden still roams free). And they may be right. It’s conceivable that there would have been attacks by now if not for the US government’s stepped-up security at home and its vigorous anti-terrorism efforts abroad.
We don’t know. We do know that studies of our statistical competency show that we systematically overestimate the probability of high-profile threats such as shark bites and terrorist bombings and that we poorly estimate the probability of dangers such as highway fatalities. The comparative absence of terrorism could just as easily support the conclusion that we have overestimated the terrorist threat. We may be "winning" a war against terrorism simply because there are few terrorists out there posing a serious threat to the US.
Citizens may have traded substantial civil liberties and international goodwill for a lot more security than we need. Answering these questions involves a subtle type of formal reasoning called negative evidence: sometimes a search that finds nothing is evidence that there is nothing. Suppose you shop in a store and then can’t find your car keys. How much of the store must you search before you conclude the keys are not there?
The negative evidence for this conclusion grows as the search widens and finds nothing. The strength of the negative evidence depends on the size and complexity of the search area. For instance, we have good negative evidence that there is no Loch Ness monster because no sonar sweep of the Scottish lake has found such a creature.
And we have no good negative evidence that we are alone in the cosmos because we have just started to search the vast heavens for signs of structured energy. The war in Iraq gives a telling example of negative evidence. The coalition forces still cannot find the alleged stockpiles of weapons of mass destruction. The negative evidence grows stronger each day as a wider search finds nothing.
The Bush administration has said that it is better to be safe than sorry. The trouble is that all bureaucracies have a well-known incentive to overrely on being safe than sorry. No one wants to risk approving a new drug or airplane design that has even a slight chance of killing someone, even if the drug can save lives or the design can increase flight efficiency.
A related problem is that terrorists have an incentive to exaggerate their strength in order to frighten their opponents and to attract recruits and donations. The result is an inadvertent global equilibrium where governments play it safe by overestimating the terrorist threat, while the terrorists oblige by overestimating their power. A tight presidential race in the US, for example, only heightens these perverse incentives all around.
The bottom line is this: there will always be terrorists and legitimate efforts to catch them. But meanwhile, the bigger statistical threat comes from the driver next to you who is talking on the mobile phone.
– (LA Times/Washington Pos service).
Prof Bart Kosko teaches probability and statistics at the University of Southern California.