While it's still fresh, here are some lessons being learned in the middle of watching (and being a small part of) the net's response to Hurricane Gustav. I'm not putting a lot of links in this post.
Decision to evacuate: It was clear while following Twitter - both directly from the handful of NOLA folks I follow and generally from looking at the net - that people went through a huge amount of decision-making about when and whether to leave. Twitter looked like an actual help here for the folks I knew, both to help let others know where they were going (or whether they were staying) and to help make the evacuation process somewhat less isolating.
Evacuation path: Those who left relatively early had relatively easier driving; by the time evacuation was underway in earnest the traffic jams were huge, some gas stations were temporarily out of gas, and the contraflow caused a lot of congestion when it ended. Twitter again seemed to be useful to help people figure out which non-Interstate highways looked like reasonable exit paths, or at least if it wasn't useful at that moment it was something to file away for next time.
Lodging and companionship when you arrive: Here the contrasts associated with poverty and access to resources are the starkest. Those with means to leave early and drive far arrived to empty shelters, and people eager to help; those who went somewhat off the beaten path, and knew where they were going, found their choice of hotel rooms; those who evacuated late and didn't go very far found themselves in crowded shelters right in the hurricane's path. There were stories of Labor Day reunions with family that had left during Katrina for another city and never came back, and more than one account of people who evacuated to their previous Katrina evac destination.
Keeping track of things back home: New Orleans was out of the worst of the storm, and a cadre of well connected people stayed behind, and as a result there's been detailed reporting on status down to which bars are open and who has power (even though the Entergy site is only reporting aggregate numbers at the parish level). Other places had it much worse, with Baton Rouge the biggest loser, getting a substantial storm hit and a lot of NOLA refugees in shelters and really bad power and telephone recovery. I don't (personally) have as connected a Baton Rouge network as I do a New Orleans network, so it feels second hand and isolated.
Decision to return: This is a complicated set of decisions as of yet not completely played out, and really the next question. If I return home, will there be flooding on the way? Will the power be on? Will I have phone service, will the toilets work, will there be good drinking water, will I get stopped on my way home and then stuck? If I don't go home, can I afford the hotel stay?
The other question you have to ask, of course, is whether the Internet has been helpful for 2 million people, or just for 200 who are super well connected.
The OnStar service announced how much health and welfare traffic it took during Gustav, and how helpful it was to have trained professional staff on the other end of the phone line to help people during an evacuation - to find hotel rooms, to give driving directions, to be calm.
Various "social media projects" sprung up to coordinate information, but the real winner in the wild appears to be radio (in particular Mississippi Public Radio), which can reach people right close by even if their phone is out of battery or out of service.
CNN was broadcasting live (!1!1!) using Twitter to get real time feedback from people - how much of that is just the nature of broadcast news to seize on new technology, and how much is it some fundamental change in how reporting is done in the middle of a disaster.
The test, I'm sure, happens not this week or next but months or years from now as Houma, LA is rebuilt or when Pineville, LA and Alexandria, LA once again have clean running water. There's some disappointment in that the reaction has been "it missed New Orleans, no problems", when that's so clearly not the case.