Writing

Feed Software, technology, sysadmin war stories, and more.

Sunday, April 15, 2012

Failing to recognize your own recruiting materials

When it comes to employment, there's the notion of "the place I quit was not the place I joined". This usually implies a change which was less than ideal. After all, the changes which improve things aren't the ones you complain about. It's when they pull the rug out from under you that you feel betrayed.

This story isn't about that. Instead, this is about a company which apparently changed well before I got there. I didn't realize it at the time, so it didn't raise any flags. Even now, I still can't bring myself to read too much into it. I can totally rationalize it away with any number of plausible reasons.

Set the time machine for the summer of 2004. Google was running crazy advertisements all over the place to entice people to join. They had some kind of math nerd billboard on 101 halfway up the peninsula. This was the whole "{ something about digits of pi } . com" thing. They also started canvassing magazines.

One of those magazines was Linux Journal. In the July 2004 issue, they included this little gem which covered two pages completely:

Google Labs vending machine

They give you a bunch of snacks with different item numbers, a bunch of weird buttons to use as "operands", and then a formula. If you can plug five values into that formula and get a match to one of the item numbers, you can buy that item. The catch is that there's one which can't be bought.

I looked at this and figured it would be a simple matter of brute-forcing it. I also figured there would be some "mathy" way of getting there, given the source and the news about their billboard. I decided that trying to find and then understand whatever the math joke might have been would have been far too annoying and painful. Instead of doing that, I could just write a series of nested loops to churn out all of the combinations.

A few minutes later, I had a small program which did exactly that. Running it through sort and uniq gave me a list of everything which was possible. Now I just had to find the one which wasn't possible. Given the small number of options, it was simple enough to do this by hand.

If it had been a bigger list, I probably would have gone to the trouble of typing in all of the options. Then I would have taken that and written some quick shell script gunk to grep for each option and yell when one didn't match. If all I got was one result, then that must be it. But, as you can see, it was small enough to where doing it by hand was probably faster than that.

The whole thing took under 15 minutes. I just filed it under "random late-night time wasters" and didn't think much more of it. After all, I had just started a new job in an interesting environment in a cool downtown office. I had no reason to bail out for the west coast.

Within a couple of years, that company changed. Things like Umbrellagate and Project Darkness happened. I wanted out, and so I applied to the big G. A couple of weeks later, I flew out here to Silly Valley for an on-site interview.

I mentioned the vending machine test to the recruiter who was showing me around that day. She had no idea what I was talking about. I also said something about it to one or two of the people who gave technical interviews... same thing. Years later, while actually at the job, I posted these same pictures to an internal Buzz post. Nobody really seemed to remember it.

My best rationalization for all of this is that the marketing folks who were working for the recruiting folks in 2004 were all contractors or similar, and they were long gone by the time I started interviewing in 2006. The people who I met probably had no idea about any of that stuff as a result. They had been busy doing other things.

That's my rationalization, at least. Who knows what actually happened.