Saturday, January 9, 2010

Low Probability; High Consequence

An essay in today's WSJ written by Paul Campos begins with this,

...I could definitely beat LeBron James in a game of one-on-one. The game just needs to feature two special rules: It lasts until I score, and when I score, I win.

As this was an article discussing terrorism, I thought I understood the point Campos was about to make. Namely, we (LeBron James in his analogy) could ring up hundreds, even thousands of consecutive successes (points) in the war against terror (the game), but if the terrorists (Campos) are successful just once (and score) killing tens of thousands or more, then we lose the game.

Unfortunately Campos took his argument in a different direction.

The world's greatest nation seems bent on subjecting itself to a similarly humiliating defeat, by playing a game that could be called Terrorball. The first two rules of Terrorball are:

(1) The game lasts as long as there are terrorists who want to harm Americans; and

(2) If terrorists should manage to kill or injure or seriously frighten any of us, they win.

Point one is correct. Point two is not. Terrorists don't win if they kill any or even a limited number of us. They certainly don't win if they seriously frighten any of us. There was plenty of killing and many frightening moments in World War 2 and the Cold War. Our enemies didn't win those. What would constitute a win for our present enemy? Well, someone wrote some time ago that Islamic terrorists will win when a giant mosque occupies the former World Trade Center site. That aint gonna happen, unless we let it.

Campos goes on to ridicule what he sees as our current obsession with preventing terrorist attacks. He cites statistics that reflect the wildly irrational fears among the populace regarding everyday risks it faces. is full of risk, and that of all the risks we confront in America every day, terrorism is a very minor one. Taking prudent steps to reasonably minimize the tiny threat we face from a few fanatic criminals need not grant them the attention they crave.

Consider that on this very day about 6,700 Americans will die.

...that around 1,900 of the Americans who die today will be less than 65, and that indeed about 140 will be children. Approximately 50 Americans will be murdered today, including several women killed by their husbands or boyfriends, and several children who will die from abuse and neglect. Around 85 of us will commit suicide, and another 120 will die in traffic accidents.

The chances that one of those horrible things will be that you're subjected to a terrorist attack can, for all practical purposes, be calculated as zero.

Nate Silver in an accompanying article presents statistical support for Campos' point. the decade of the 2000s, only about one passenger for every 25 million was killed in a terrorist attack aboard an American commercial airliner (all of the fatalities were on 9/11). By contrast, a person has about a one in 500,000 chance each year of being struck by lightning.
Relative to the number of commercial departures world-wide, passenger deaths resulting from what I term "violent passenger incidents"—bombings, hijackings, and other sabotage—were at least five times less common in the 2000s than in any decade from the 1940s through the 1980s.
The years between 2005 and 2009 (313 fatalities), in fact, represents the second safest period on record since at least 1970. Surely some of this is because of improved vigilance and intelligence.

Surely it is.
Campos' and Silver's thesis is that we expend far too much effort, time and money to prevent an almost non-existent threat. We should be trying to minimize more prevelant threats - murder and auto fatalities - by outlawing handguns and mandating long prison terms for the former; lowering speed limits and promoting high tech auto safety features for the latter.

Let's not quibble about some of Compos' questionable assertions. A few fanatic criminals? I'm not sure just how few of these there are. Criminals? I guess that's one way of putting it. Just not an accurate way. How about calling them soldiers? Warriors? Also, note that we could implement Campos' suggestions regarding murder and traffic deaths without any changes to our current antiterror practices. They're not mutually exclusive. And put aside whether or not the recommended measures would actually achieve their desired effects. That's a subject for a separate discussion.

The main problems with the Campos-Silver analysis are as follows. First there's the apparent misreading of cause and effect. We are highly focused on preventing terror attacks and terror attacks are rare. True. And your point is what? The New York Times a while back bemoaned the increased level of prison incarceration. With a decreasing crime rate no less. Makes no sense at all! This thinking has the perverse effect of punishing success. President Bush was maligned for his pre-occupation with preventing another 9/11. Well, there were no other 9/11s (or any other terror attack) during his tenure - what a waste of time and money!

And how about this enlightening observation by Silver,

The chance of a Westerner being killed by a terrorist is exceedingly low: about a one in three million each year, or the same chance an American will be killed by a tornado. (The Department of Homeland Security's budget is 50 times larger than that of the weather service).

So maybe if we bring the level of the DHS budget down to that of the NWS, deaths by terrorism will only rise to 50 per three million each year.

The only way to know if the low incidence of terrorism is due to the measures we're taking is to stop taking them. Does anyone want to be responsible for making that call? Campos' article derides the "increasingly pointless and invasive searches and the resultant delays", that our security system imposes on travelers. What's the alternative? Campos suggests
"Taking prudent steps to reasonably minimize the tiny threat we face from a few fanatic criminals..."
What prudent steps might these be? Armed marshalls on some flights? Ethnic profiling of the most likely terror suspects? (Good luck with that). He doesn't say.
This isn't to say that the system as it exists now is as efficient or effective as it could be. Improvements are needed. By all means fix what's not working.

But the biggest problem with the Campos-Silver calculus is not the (alleged) miniscule benefit in reducing "the risk from an extremely small nonzero number to a slightly smaller nonzero number" as Silver puts it. If the price of delaying and inconveniencing the flying public is preventing only 1 or 2 (or 3 or 4) blown up passenger jets a year - then it might be worth not paying it (though I disagree). But what if one of these attacks, maybe involving WMDs, is targeted in such a way as to cause great loss of life (tens of thousands or more) and/or widespread destruction? Outcomes that would have devastating and long lasting effects on our economy? Our way of life? Silver tries to dismiss this possibility, "...if history is any guide, the next attack will probably not be like 9/11—it will be like NWA 253, something which threatens the lives of dozens or hundreds of people, not thousands..." Note the caveats sewn into this statement - if history is any guide...the next attack...probably...
That phrasing is not reassuring when the potential consequences are catastrophic. It was a near miracle that the 9/11 attacks took "only" 3000 lives. Two 110 story office buildings collapsing during a weekday during rush hour in the heart of New York's financial district could have resulted in a death toll ten times that. It's what al-Queda was hoping for. Without a briefcase nuclear device or a "dirty bomb" on board the planes.

We have to consider what the jihadists are striving to accomplish, not just what they've accomplished so far. Risk analysis is only valid with relatively stable parameters - traffic fatalities, murder rates, lightning strikes are all fairly constant year to year. One "outlier" as Silver calls a 9/11 type attack, distorts the equation. One or more outliers invalidates it. And multiple outliers constitute war. Jihadists aspire to commit multiple outliers.

Whether they know it or not, and I think they know it, Campos and Silver, with their actuarial approach to the war, are one with the Obama approach. That is, the Obama approach before its absurdities were exposed by his recent response to the attempted bombing of flight 253. Then came a policy "adjustment". Sorry guys. There's votes involved. So now the president says we're at war with al-Queda. As with most of Obama's proclamations, these are empty words - Umar Farouk Abdulmutallab is still going to be tried as a civilian in a civilian court. But it sounds good.

Campos gets the rules of the game right - one score by the terrorists and they win. Only he misunderstands what constitutes a score and what the implications of the victory are.

No comments:

Post a Comment