[wnissen]

Do you have a cite for this? I thought Google looked at interviewers and found that no particular interviewer did any better than any other, except, literally, one guy in a narrow sub-specialty. That's quite a lot different than concluding a 50% success rate is no better than chance, which I agree would be problematic at best, flat wrong at worst if the expected occurrence was low.

[jia_min]

Here's the NYT article where Laszlo mentions the "random mess" they found when they used the interviewer as the predictor and then subsequently mentions how they use structured interviews now instead: http://www.nytimes.com/2013/06/20/business/in-head-hunting-b...

The interview correlations I cited are from Schmidt and Hunter, 1998: http://mavweb.mnsu.edu/howard/Schmidt%20and%20Hunter%201998%...

My point is, it doesn't matter who does the interviewing - looking for people who are "good interviewers" is a unicorn chase because the vast majority of people don't have a magical ability to guess who's going to be a good performer pre-hire. Unfortunately, Amazon is making the same mistake with their "bar raisers": http://firstround.com/article/Mechanize-Your-Hiring-Process-...

Geez, you engineers don't seem to like to learn from each other.

[deciplex]

But Google doesn't do structured interviews. Not always, at any rate.

[jia_min]

The company line is that they use structured interviews. What actually happens in practice is often a different story.