On Job Interviews
There’s a riff over at Seth Godin’s Blog on interviewing—specifically, on how stupid it is and why it should be abolished. I’m glad someone influential is saying it. Interviews are a joke.
Microsoft’s interview system was, seemingly, more rational than many I’ve dealt with. For one thing, you were supposed to give the candidate a “hire” or “no hire” rating. In my group (legal) this morphed into a number of gradations, including “hire with reservation.” Lawyers are inclined to see many shades of grey, as you can imagine. But even with those additions, it was a fairly easy system to administer. We also stack-ranked candidates. We would meet and list on the whiteboard the candidates, then do pairwise comparisons until we sorted people into a rank.
But there were still problems with it. Here’s one example. We were interviewing an attorney and, as usual, the e-mails were flying between interviewers as the interviewee walked between offices. The interviewee was quite open and honest—traits which I tended to select for. Apparently, she was open and honest with everyone, but while some liked that quality, others didn’t. So, one e-mail would say things like, “She’s a straight-shooter,” another would say, “I question her discretion.” Different people see the same data and draw different conclusions. Since the data set is so small (20-30 minutes of interviewing), it’s hard to draw any reliable conclusions.
Another problem is one that Seth points out: the “professional” interviewer. I recall one candidate who was so good at interviewing that I gave her a “no hire” simply for that reason. There was no way to get any real answers out of her. It was like interviewing someone who spoke Martian—what’s she saying, exactly? Interestingly, I was the only “no hire” on her—though there was another interviewer who didn’t like her either, and he used my “no hire” for cover, in order to avoid expending political capital. Everyone else was—as I saw it—fooled. I don’t think we ended up hiring her, for some reason.
One example of her interview style will give you the picture. She was sitting in my office and asked me, “So what makes to excited to come into work here? What makes you passionate [her eyes widened] about your job?” Whoa, honey. Slow it way down. That question is just loaded with assumptions. Excited? “It’s a paycheck,” I said. “No more, no less.” It was fun to watch her game-show-contestant expression melt away. I stopped the interview early and punted her to the next person on the interview loop.
But is there any other way? Godin mentions an alternative: work with the person in a trial basis and see if they are any good. While I don’t know about the specifics of Godin’s proposal, this sort of system does, in my experience, work. Here are two examples.
When I worked at a law firm, the practice was to hire associates for a summer (this is the convention in the profession). The “summers” would do different kinds of work, and the firm could try before buying. This made the interview, in some ways, less important—or at least changed its focus. The interview was, mostly, a sniff test. Was the person annoying? The interviewers had grade and pedigree information, which is pretty useful. A high GPA means “disciplined and/or hard working.” A top law school means high GPA plus, probably, high IQ. Class rank can mean a few things and for me was never as important, though it probably should have been. But, ultimately, personality is the real differentiator, just like with pets. It’s no good having a beautiful cat that pees on your pillow. That’s what you test for in the interview: weed out the pillow-pee-ers.
At Microsoft we had an even better system: hiring outside attorneys with whom we had worked. Generally, we’d been working with these attorneys for a while, so we knew their strengths and weaknesses. While we had informal “no poaching” agreements with the outside firms, if the associate was “in play,” then we could hire them. This worked great. Because the way to really see if someone is a good hire is to work with them. Every other method is pretty lame in comparison—so lame, I think, as to be almost useless.