Three notes on this NYT article about Google’s hiring experience.
The headline says, “In Head-Hunting, Big Data May Not Be Such a Big Deal.” The article is all about the virtues of crunching data on one’s hiring practice. I’m confused by where they get the gloss that it’s not a big deal.
Does “Big Data” mean anything at all that “Data” does not? This is not crunching petabytes — or even terabytes — of data.
Those quibbles aside, this is a decent introduction to the fact that you probably suck at interviewing.
Asking people fuzzy questions about their life experiences? All available data suggests that it is completely unrelated to their future job performance.
Asking people to solve puzzles? All available data suggests that it is completely unrelated to their future job performance.
If you have the best interview techniques in the world (which involve formalized, standard interviews, and real problems culled from actual work, and general intelligence tests), then your interviewing is still not amazing. Think that you’re the exception? No, you’re self-deceiving.
And to be clear, “all available data” is quite a lot of data, not just Google. Here’s a decent overview courtesy of tokenadult at Hacker News.
The “interview” that I got at Pivotal Labs, and which I now use as my interview process, is simply to work with the candidate. For as long as possible — at least an hour, more preferably half a day. On my actual work. We pair-program and tackle whatever issues are actually at the top of the backlog. This still isn’t great: almost certainly the best “interview” technique is to hire lots of people and be fearless about firing those who don’t work out. But it at least minimizes the extent to which you’re testing the candidate’s interviewing skills rather than their job skills.