Jim Miller takes the counterintuitive position that 48 hours for an evacuation is actually better than 72:

If we put these points together, we see that we want to start our evacuations late enough so that people can be persuaded that the hurricane is likely to hit them, and that the costs they will suffer from the evacuation are worth the decrease in risk. My guess is that about the earliest that this can be done is 48 hours before a hurricane is predicted to hit. Those who think this too late are invited to look at Katrina on Friday morning, 72 hours before it hit New Orleans. At that time, Katrina was a Category 2 storm, hundreds of miles away from New Orleans.

And this is typical of hurricanes; predictions made 72 hours before they strike are not very accurate — and people in hurricane areas know this.

Many of you will already be wondering about this question: Would 48 hours be enough time to evacuate New Orleans? As far as I can tell from these statistics on commuters, the answer is yes. New Orleans Parish has more than 140 thousand workers commuting to work by car, motorcycle, or truck on an average work day. If we assume that those vehicles could carry an average of 4 people, then they could have evacuated the population of New Orleans (about 450 thousand) in a single trip, with the help, of course, of the "contraflow" system that makes the major roads out of New Orleans one-way. Even with slower traffic than usual, they could have gotten to safe areas in a few hours, assuming the state police managed the flow well. (You would want them, for instance, to preposition tow trucks and other emergency vehicles in order to keep the traffic moving.)

Now you would want to do a formal study to prove that I am right, but we should remember that 80 percent, by most estimates, of the population was evacuated. And, although they got warnings beginning on Friday, Mayor Nagin did not order an evacuation until Sunday morning, 23 hours before the storm hit.

He makes a pretty convincing case. As far as getting the commuters out, it's apparent that those that wanted to and were able to got out. That there weren't cars on the road when it hit seems to validate this.

But the bigger issue put forth by Hurricane Katrina is, in my view, those that didn't want to and more particularly those that weren't able to. There are a couple lessons I consider pretty concrete here (and by "concrete" I mean non-political, things both the left and right should agree on), one of which is that an 80% evacuation is not necessarily enough. 100,000 people in the midst of chaos leads to rather unfortunate things.

More lead time would allow a couple of things. First of all, it would allow for more time to mobilize busses and make arrangements for a place for those people to go. The officials dropped the ball on this so we don't really know if there really was enough time. If there was enough time for one run, we don't know about more than one run. That could require more time or it could not. The thing is that we don't really know. Secondly, we don't really know how many people both did not appreciate the severity of the danger and could have been convinced. A certain number (less than 20%, almost certainly) would have stayed behind anyway. But I would like to know the feasibility of some sort of alert system to separate the potentially horrendous from the potentially bad. I'd almost be willing to say that in the case of New Orleans, a level two hurricane that appears to be gaining strength is potentially horrendous. Easy to say now, of course, but I'm thinking of the future here and not looking to condemn what happened when Katrina came ashore.

Basically, Brendon Loy made the outstanding point that if there is a 1-in-10 chance that a nuclear bomb was going off in your city, you evacuate. Mandatory? Well a mandatory evacuation isn't exactly mandatory when lots of people didn't leave. But calling it that does lend a sense of urgency that cannot be defined any other way.

Maybe there is another definition. I don't know the laws in this particular area, so people are more than welcome to tell me that what I urge is already the case. The question I don't have the answer to is whether or not an employee can be fired for leaving town when a hurricane may be hitting. If so then I believe that we need to add a special protection to those that elect to leave early. I'd guess that to be the case in a mandatory evacuation (not an issue for Katrina since the city was destroyed, but it would have been had the hurricane not hit at expected), but I don't know it to be in the case of an "urged" evacuation. Maybe having a "strongly urged" evacuation with employment protection at 72 hours would be sufficient. Closing the city and schools and utilizing busses at that point might also be worthwhile.

Yes, something like this would lead to false alarms, but at least in the case of a city as vulnerable as New Orleans or simply as populated as Houston or Miami, false alarms are better than devestation. One thing we ought to be keeping in mind is that those that live in a coastal area need to factor hurricane scares in the same way that northern schools, business, and government factors in snow days. I don't believe this to be a completely unreasonable expectation.

Miller is concerned that the government may not value the costs of such declaration and declare them too often, presumably as a CYA measure. If anything, the opposite appears to be true. The New Orleans mayor was very aware of the financial repercussions of a mandatory evacuation and it appears that he may have hesitated on account of that. Leaving the decision up to local authorities who would be held accountable by business interests (which the mayor was clearly and understandably worried about) would serve as a balance for a mayor anxious to say that he or she did all that they could.