Friday, August 03, 2007
I use black swan in the Taleb-ian sense. Unlike Matt's otherwise very sensible post, I don't think trying to figure out why the bridge fell is going to somehow improve policy. I make this point repeatedly to students -- you can sometimes do everything right and get a bad outcome due to randomness. Black swans are simply outliers -- they are things that just happen that cannot be expected within the realm of our experience. I think that until investigation concludes, you should reserve the possibility that the bridge collapse is a black swan event. (Or perhaps a gray swan, for those who've read Taleb.)
Begin with the observation that no bridge can ever be perfectly safe. Rational allocation of scarce resources must involve marginal analysis: We allocate the money to make transportation safe up to the point where the marginal benefit of investing the dollar in project A is equal to that of project B is equal to that of project C, etc. And we invest only an amount where the increase of safety is worth the expense of providing it. I don't wait for the car 300 feet away that I judge to be traveling about 30 mph to pass before crossing the road; experience teaches that I can cross that road without incident. 99.999% of the time, it will be the case; every time I have done so before I have arrived to the other side safely. That doesn't mean the car will never hit me, just that the chance that it will is vanishingly small so that I do not allow it to disturb the decision that I'd rather be on the other side of the street now.
So it is with the I-35W bridge: We knew it was "structurally deficient", just like 73,533 other bridges in America. Repairing them takes a generation, we are told, and costs $188 billion. But, as Peter Gordon points out in another context, all applied economic analysis comes to ten words: At what cost? Compared to what? How do we know? We are told this is the cost, but what else can you do with the money? Would it have come out of light rail? Would it have come out of the DNR budget? Schools?
And the time consideration is vital. The argument over whether we could have prevented this particular event means either the Pawlenty administration would have been willing to impose the costs of detoured traffic on the possibility that this particular bridge would fail. It would have required you to build out over some period of time, probably longer than two years. What is the value of all the time that drivers in the Twin Cities would have lost? And that would have been certainly lost, compared to the possibility of failure. So the state made a decision, with full knowledge, knowing that "sometimes Things Fall Down" but investing the dollars into other, more likely or more beneficial projects. All scarce resources have to be allocated somehow. If you don't want to allocate based on marginal cost equals marginal benefit, tell us what your decision rule is.
Had the Pawlenty administration actually foreseen this black swan, it would have had to argue that this was the one to pay attention to, and damn all the other structurally deficient bridges in the state. Or it would have had to add billions more in taxes, plus all the detours of traffic, to repair all the deficient bridges; otherwise, it must prioritize and leave some other deficient bridges out there. Certainly the road construction firms would have been delighted, but they're the glaziers in the broken window fallacy.
Short of a report from the tragedy that says the engineers missed a flaw in the bridge, we will be left with a report that this was something that just happened, a random event, a black swan. (And not necessarily the only one; recall the Schoharie Bridge failure on the New York Thruway in 1987 after a "50-year flood".) And while engineers can determine whether something could be done to make bridges safer, I can give you the answer: There always is something more. But it's not free, and we don't know if it's worth it. That's a question for markets.
ADDENDUM: As one might have expected, the told-you-sos are now out. Engineers, again, are not paid to decide whether to invest in this project or that, just to tell you what it costs to get a certain result. Engineers never tell you "compared to what?" But engineers make great copy for an article, or witnesses for a legislative investigation. Neither journalists nor legislators will answer the compared-to-what question for you, either.