I knew better when I wrote it.
A couple of weeks ago I based KJR on a blistering, self-righteous critique of the various guilty parties responsible for the Katrina fiasco, deriving a number of lessons for IT along the way. While the guilty parties are just as guilty as before, it turns out the root cause of the flooding was much more complicated than my simplistic analysis (failure to invest despite a clear need to do so) suggested.
It’s almost always more complicated, which is why “self-righteous indignation” and “ignorance” are strongly correlated, if not synonymous. Simplistic analysis is for the simple, the intellectually lazy, and (to be fair) the excessively busy. None of the three are excuses for bad decision-making.
It turns out that while hurricanes have doubled in total destructive power over the past 30 years, and that global warming is probably a major cause, Trent Lott’s lost house exemplifies the bigger reason that hurricanes do more damage than they used to: Lots of Americans have chosen to build their homes along hurricane-prone shorelines.
It also turns out that Katrina’s storm surges were smaller than initially thought, well within the design parameters of New Orleans’ flood control system. Some newly renovated levees failed, apparently due to poor design (preliminary analysis suggests the use of flat plates instead of interlocking plates is a primary culprit).
Want to bet that at some point during the design process, the engineer who suggested interlocking plates was overruled due to cost considerations? And if not, that the contractor whose bid was based on interlocking plates lost out to the contractor whose design did not? The federal government does, after all, tend to award contracts to the low bidder (except, of course, when it awards no-compete contracts to Haliburton).
This new information reinforces the final point of the Katrina column — engineering does matter — but adds an entirely different perspective to it that provides another important, if prosaic reminder for IT: An absence of defects does not ensure the presence of quality. It’s a necessary condition, but not a sufficient one.
Depending, of course, on how you define “defect.” Too often, in IT, we equate bugs with defects, and as a result, testing with software quality assurance. It’s a basic point, but then, in IT as in most other disciplines, 80% of success comes from mastering the fundamentals. It’s still the case that in far too many IT shops, even testing receives too little time and attention.
In the case of New Orleans’ levees, it wouldn’t have mattered how good a job inspectors did after the installation was complete. Even if there were no construction flaws, the fundamental design was defective (assuming the preliminary analysis holds). The parallel for software development isn’t subtle: Software designs should be reviewed for conformance with software engineering standards prior to coding.
Not perfectly parallel is the need for stress testing, which would have been highly desirable for the levees were it not impossible, since there’s no practical way to set up Development and Test versions of New Orleans. There’s no equivalent reason to not stress test software.
But the larger lesson is the first one: Other than in Aesop’s Fables, it’s rare that challenging situations leave us with a few, simple, clear lessons. The world is a multivariate place, especially when those pesky human beings get involved.
Which leads to one final point, after which I’ll leave the hurricane post morti to the committees and experts. That’s the matter of what, for want of a better term, we can call “contingent responsibility.”
Contingent responsibility is what didn’t go wrong but just as easily might have. When you do something dopey, you sometimes get lucky and live to tell the tale … fortunately, or many more of us wouldn’t have survived our formative years. If you’re smart you recognize you got lucky, and avoid repeating your dopiness. If you aren’t, you figure it means your reasoning was sound.
People being what they are, I’m confident that when those who made the decisions to not invest enough in levee maintenance learned it was new construction that failed, they decided their allocation decisions were just fine after all. I’m equally confident that when those who discount global warming saw that the rise in coastal construction was the primary cost driver, they felt justified in continuing to fight emission reforms.
I’m confident, because I’ve seen too many business managers rely on luck as an alternative to investing in the mitigation of well-understood risks, disparaging those who warn of the consequences with this compelling logic:
“It hasn’t happened yet.”