The New York Times reports that Google engineers try to improve finding good job candidates by assigning an automated quality score to applications. Maybe that’s not a bad idea if you receive over 100,000 job applications every single month; the NYT reports that the past hiring process often took months to consider candidates, and judging grades and interview performance turn out to be a sub-optimal way of finding the best people.

What Google did now was to ask its existing employees to provide their “personality data” covering a wide array of topics – everything from “is your work space messy” to “what pets do you have.” Now they connected this “biodata” with the employee’s performance (aggregated using another 25 separate approaches on its own), and boom – if you believe in the magic of significant overlaps in statistics, you now know whether or not you ought to hire messy people with dogs (pictured above, sort of).

Specifically, having a pet didn’t turn out to significantly correlate with work qualities... other features turned out to be relevant, though. Still, the NYT says for Google it’s too early to tell if the system, in action since a couple of months, is really working... and some Googlers don’t even like this approach too much.

Oh well, I guess it’s better to automatically screen applicants for more than just their educational grades, e.g. to check if they started a couple of websites, a club, or if they published a book (disclosure: my own biodata definitely skews me towards believing that university’s not the only good measure for successful workers!). When I was judging tech applications at my old company – yeah, like judging 10 people a month is comparable to judging 100,000! – I’d always screen a candidate’s online record... did they create interesting websites? Are they active in web discussion boards? Did they ask lots of technical questions in usenet newsgroups?
I’m getting very suspicious if someone derives all their knowledge from often outdated traditional (German) learning courses or (German) tech books, and then doesn’t feel the need to try out this knowledge for hobby projects. In the kind of day-to-day web development I was involved in, it is often more successful to know the right strategies of how to find new information quickly than to rely on whatever you were taught years ago... and often, people acquire this knowledge when working on beloved projects of their own at home.