Saturday, September 22, 2007

How Would You Move Mount Fuji?

I think we all know that I've got a problem with cutesy, right-brained job interview questions, and I'm looking for employment someplace that thinks they're totally inane, but I have to admit the questions you get asked at a Microsoft interview are momentarily interesting. Still thoroughly uninformative, since they're basically verbal Rorschach tests — riddles you're not supposed to be able to solve on the fly, so not coming up with a correct, or even cogent, extemporaneous answer says more about the question than it does about you. I understand the excuse: it's not about the answer but about the reasoning process, but that's disingenuous. Example you've heard before: A bear starts in its den. It walks a mile south, then a mile east, then a mile north, and arrives back at its den. What color is the bear? I want to take a look at the cognitive process here...

There's something intuitively wrong with the story: walk a mile, turn ninety degrees, walk a mile, turn ninety degrees, walk a mile, arrive back at your starting point. You can't draw it on a map. Figuring out why exactly you can't draw it on a map isn't straightforward. I mean, that's the point, but it's expecting a lot of an interviewee under pressure to automatically switch into non-Euclidean geometry mode. Plus — of course you're aware, but how often do you think about the earth not being flat, other than Columbus Day? (Of course, if you're The View co-host Sherri Shephard, the answer is "never.") The question is really just a test of your ability to pull two barely-related facts randomly out of the ether, and that's probably not a skill you can put on your resume.

As a side note, the earth is elliptical (or whatever the three-dimensional equivalent is), not spherical, so if you want to be technical about it — and you do, because that's the sort of thinking the problem asks about in the first place — the brain teaser doesn't even work.

I didn't interview for Microsoft, but I did get asked one in my last interview. Question: How many windows are there in New York City? A lot. More than the number of protons in a carbon atom, fewer than the number of galaxies in the universe.

Could you be more specific?

Please, that eliminates practically all the whole numbers.

That's not what I said, because I was a little dumbfounded. "Do you want to know exactly?"

Another issue I have with these questions is that the guy posing them has never fully thought them out. Case in point: Whatever you think? Wrong answer. The question, "Exactly how many windows are there in New York City?" is most likely meaningless in any practical sense. (I explain below.) "Approximately how many windows are there in New York City?" seems ridiculously easy. Count the windows in a small sample of buildings that's representative of the whole city and multiply. I actually had to write this out on a white-board as if it weren't the simplest thing in the world. Plus side, I got to sniff some dry-erase marker.

As for the exact number... you'd have to count them (assuming no one's been keeping track of installed and broken windows since the days when counting windows was trivial). I can't think of a pigeonhole principle way of determining the window count. You could probably expedite the process by going down to City Hall and using the buildings' plans as your data source instead of the buildings themselves, or construction companies might keep records of how many windows they purchased and how many of those were to replace broken windows. The problem is that all of those methods take time, and unless the city is having a particularly lethargic day, the number of windows in the city is going to change while you're counting. "Exactly how many windows are (were) there in the city at a particular time in history?" has an answer, but until it's possible to count all the windows in the city in less than the time frame between installing or breaking a New York City window, the general question can't be answered.

The fact that Microsoft employees spend their time thinking up brain teasers like those rather than designing robust software probably explains why my computer crashes so much.

Oh yeah, hint: Why are manhole covers round? So they won't fall in the hole.