Friday, January 13, 2012

Things They Should Invent: "people you may know who have died recently"

I've noticed social networks are pretty good at identifying people I might know or people I might be interested in. What if they could combine that with an obituary search and send you alerts of people you might know who might be dead or bereaved?

You could enter your employment and educational history, so the system can identify people who worked with or went to school with you. You could also enter the names of people you're interested in, either in that you'd want to know if they've died or you'd want to know if they're bereaved. You could customize extensively what kind of alerts you get. For example, if you grew up in a small town, you might want to be alerted every time someone in your age cohort is mentioned in a obituary, because it's probably someone you know. If you have an unusual surname, you could be alerted every time someone with the same surname is mentioned.

Unlike social networks, there's no need to be reciprocal. If you want to be alerted if your high school crush or your high school bully appears in an obituary, you can do so without anyone finding out. However, you could have the option of allowing the information you've entered about your age, hometown, employment and educational history to feed other people's alerts, so if you die or are bereaved, other people who probably know you can be alerted, even if only your name is mentioned in a list of survivors.

It does sound like it has the potential for false positives, but this could be partially mitigated with a small message saying why you're getting each alert, like on twitter recommendations. (e.g. "You are receiving this alert because you requested alerts about people who attended Beauxbatons between 1999 and 2003." "You are receiving this alert because it mentions John Smith, whose personal data indicates that he was born in Dog River, Saskatchewan in 1980.") Given the creepy accuracy of facebook, twitter and linkedin recommendations, I think it might work.

No comments: