Sunday, July 30, 2006

An Interesting Posting on Katrina


Ed Wenk, a long-time student of how technology can experience catastrophic failures, has an interesting op ed. on what went wrong with Katrina:

Summarizing the Berkeley report on the New Orleans levee failure, to which he contributed, he says

The Berkeley team found that all modes of failure were preventable at modest additional cost, that the safety margin of 1.3 was too low, that the storm intensity assumed in design was too low and driven by political, not engineering, processes. That piecemeal funding led to piecemeal construction. Looking back, the system was a failure waiting to happen.

His conclusion:

The Katrina disaster hatches an imperative we dare not ignore. We can learn from its lessons of human failure how to cope more successfully with extremes of weather. Otherwise, nature will find the flaw.

Or, to put it more succinctly: be afraid. Be very afraid.

2 Comments:

Anonymous Paul H. Edelman said...

OK, I'll bite. Incompetence, political influence, budgetary contstraints (and probably some level of graft) in public works strike me as very plausible explanations for catastrophic failure. Nowhere in this explanation does one see miscalculation because of a power tail ("the storm intensity assumed in design was too low and driven by political, not engineering, processes "), chaotic processes, adaptive behaviors, etc. Maybe Occam's Razor gives the right answer here: traditional public choice analysis gives a better explanation than any more elaborate mathematical model.

7/31/2006 11:20 AM  
Blogger Dan Farber said...

Well, maybe or maybe not. I suppose that we'd really need to see the full distribution of accidental harms to do a valid empirical test. But my impression is that there are more mega accidents than you would expect if (say) flood damage was a normal distribution. Of course, the particular story of each accident is different.

What Katrina has in trouble with other big failures of technology systems is that a lot of things had to go wrong interactively -- apart from the storm itself, you had poor soil testing, bad communication between the Corps' research division and its construction people. a federal disaster agency hampered by a government reorganization,etc. My impression, at least is that the curve of disaster damages seems to have a fat tail.

Putting aside the math question, there is also the question of whether the harm to New Orleans was purely a public choice problem. It seems to me that this is misleading at least to the extent that it implies that New Orleans (being in a corrupt political culture) is a unique situation. For example, the City of Sacramento could easily have larger levee failures with more loss of life. 9/11 is another interesting example -- quite a few things had to go wrong in order for the twin towers to fall, some in human systems and some in architectural design -- but many of the failures were interconnected.

Although I'm really interested in whether CAS theory has traction here and also interested in the shape of the distributgion, in the end it's more important for me that Katrina isn't just a fluke that can be dismissed as an outlier. It could easily happen again.

8/02/2006 1:41 AM  

Post a Comment

Links to this post:

Create a Link

<< Home

Google
 
Web Jurisdynamics