The Katrina PeopleFinder project is still hard at work. There are a few hundreds sets of data still available for entry on the PeopleFinder wiki, and the PeopleFinder team has now started focusing on resolving shelter databases in a project called ShelterFinder.
I’ve gotten great feedback on my earlier post listing some lessons learned from my brief period of work on PeopleFinder. One of the best comments came from Steven Skoczen, who did a great deal of the heavy lifting to generate assignment pages for Katrina. His post offers additional pearls of wisdom: “Keep an Eye on the Top of the Tree” and “Entropy Takes Breaks Sometimes”. That we were able to cooperate on the project despite never having spoken in person or on the phone might or might not call into question my contentions about the importance of posses.
Also doing yeoman service on assigning data to be entered was Matt Hurst of Blogpulse. Matt’s excellent Data Mining blog features a post about “wrappers”, a piece of software that accesses a web-based data source and gives structured access to otherwise unstructured data. Matt was able to use some wrappers he’d already written to deal with bulletin board data for Blogpulse to rapidly chunk data for assignment. His post suggests that a wrapper strategy might help efforts like ours react more quickly in the future, and wonders whether the problem isn’t going to be wrapping the data, but finding all the relavent databases. Worth a read.
While Dina Mehta and I didn’t get to work together on Katrina Relief, I watched the project she was helping lead – a Skype-based virtual call center based around Katrinahelp.info. Dina writes on SkypeJournal about her experiences helping displaced people in the US gulf coast from her living room in Mumbai – it’s a wonderful reminder of how small the world can be when we choose to let it. The ability to build ad-hoc phonebanks like this in the future is guaranteed to be an important part of Recovery 2.0.