Since the Google 10^100 contest has descended into obvious bullshit and has turned out to be a waste of time (silly us, thinking they were serious about this effort), I hereby propose that we use up Slashdot's storage and bandwidth by posting our rejected ideas here. My idea isn't anything special, but the rules were that it had to be done with a modest grant (I loved the ones about using VTOL aircraft and passenger airships, yeah, that sounds cheap). Anyway here is my idea:
After a severe or prolonged disaster most people will not have access to phones or internet, especially in poor areas, so re-connecting with loved ones is nearly impossible (think about all the missing persons posters after 9/11 and Katrina!) I therefore propose that we develop free software for use by NGO's, the Red Cross, and other volunteers which will quickly scan handwritten information forms, and upload them to an OCR back-end index and portal, hosted by Google. The blank forms will have a bar code -- the refugee returns to any connected terminal and scans or types the code to receive any information the indexing technology can find about loved ones. Any form that cannot be OCR'd will be posted on the internet in a "Mechanical Turk" arrangement so that volunteers can help index the data. A single laptop and inexpensive scanner connected via HAM radio, SatPhone, or CellPhone can easily process thousands of requests per day, which would be impossible if each refugee had to use the keyboard and navigate through a search site.
Blank forms will have language-selection tick boxes which match the country, plus a few major languages like English. If forms are not available, the software should accept regular paper and issue a unique serial number to the refugee which can be written down. There can also be a few "standard responses" like the Red Cross "Safe and Well List" website has (note that this website is useless unless each refugee has unfettered access to the Internet which is why my idea will re-unite families much more effectively). We could also establish a phone bank connected to volunteers.
I think the first stage should be to create an extensible communications spec, then write the code for PC's, then perhaps later design a solar-powered ruggedized appliance.