Friday, September 23, 2005

Rita Washes Away Post-Katrina Rationalizations

After Katrina, one meme the administration tried to sell was that, okay, the flooding of New Orleans might have been predicted, but that the odds of it actually happening had been judged minuscule, and that reasonable cost-benefit analyses had led to the construction of defenses only against hurricanes no stronger than category 3.

Writing for Reason, Jonathan Rauch says:

"Remember, the odds of a Category 4 or 5 hurricane hitting New Orleans any given year were small. [Army Corps of Engineers Lt. Gen. Carl] Strock told reporters, 'We figured we had a 200- or 300-year level of protection. That means that an event that we were protecting from might be exceeded every 200 or 300 years. So we had an assurance that, 99.5 percent, this would be OK. We, unfortunately, have had that 0.5 percent activity here.'

"Remember, too, that reinforcing the levees was a multibillion-dollar project. An ancillary project to restore the protective marshes of the Mississippi Delta, which would have reduced the force of storm surges reaching the city, would cost something like $14 billion over three decades. For that kind of money, there are always competing priorities, some of them urgent.

"The question, then, is not whether the failure to improve New Orleans's flood protection was a mistake in hindsight—obviously, it was—but whether it was a reasonable choice in foresight, based on the probable odds and costs as they appeared at the time."
Rauch, and, for that matter, the commander of the Army Corps of Engineers, seem to have fallen prey to some common fallacies about probability. I'll explain in a second, but Rauch continues:

"Using the more cautious of Strock's figures, assume the odds are that a storm surge would overtop or breach the existing New Orleans levees once every 200
years. This seems, if anything, optimistic..."
Here's the thing about probability: Your deepest, strongest-held instincts about it often couldn't be more wrong.

Let's assume that Strock and Rauch are right, though, in saying that a storm surge would occur only once every 200 years. Therefore, it seems a little excessive, to Rauch, to prepare for a 1-in-200-year event.

Here are the problems with that reasoning. For one thing, just because something is expected only once in the next 200 or even 300 years, doesn't mean it comes at the end of that span. The whole point is that it could come at any time during that span. Thus, you can't claim there's only a 0.5% chance, let alone "assurance," that it'll happen now. Also, when considering what "now" means, there's no reason to think in terms of years. A sturdy levee system would, after all, last at least a decade, right? So, let's look at this in terms of decades. Suddenly, we're talking about a 1-in-20 event. And if the levee were to last (with appropriate maintenance) for 20 or even 25 years, now you're talking about a 1-in-10 event or even 1-in-8.

More to the point, as Rita has shown us, the randomization of nature truly is random. It's not random the way fate would be if fate existed, let alone were guided by a magic man in the sky; there is no sentient decision-maker spacing out low-probability events. If you toss heads a million times, the odds of it coming up heads the next time are still 50-50. In other words, once you win the lottery, at a million-to-one odds, the next time you play you still have just as good a chance of winning even if it's the very next day.

Or, in Rita's case, losing.


1 comment:

Anonymous said...

Hey, if I start smoking, the probability that I'll die of lung cancer in any given year is *really* small....

The Army Corp guy is a douche.

Newer Post Older Post Home