Wednesday, March 28, 2012

Google interviews: would you get a job with the search giant?

In an extract from his new book, William Poundstone considers the logic puzzles, trick questions and mind-bending riddles that make Google interviews notoriously hard. Would you make it through to the next round.

We live in an age of desperation. Never in living memory has the competition for job openings been more intense. Never have job interviews been tougher.

For some job seekers, Google is the shining city on the hill. It's where the smartest people do the coolest things. In the US, Google regularly ranks at or near the top of Fortune magazine's list of 100 Best Companies To Work For. But unsexy firms also find themselves with multiple well-qualified applicants for each position. That is very good for the companies that are able to hire. Like Google, they get to cherry-pick the top talent in their fields. It's not so good for the applicants. They are confronting harder, ruder, more invasive vetting.

This is most evident in the interviews. There are, of course, many types of questions traditionally asked in job interviews. These include the "behavioural" questions that have almost become clich├ęs: "What is your biggest failure in life?" Questions relating to business: "How would you describe Holland & Barrett to a person visiting from another country?" And finally, there are open-ended mental challenges, such as how you would weigh an elephant without using a scale – something for which Google is particularly known, an attempt to measure mental flexibility and even entrepreneurial potential. The answer? Nudge the beast on to a barge. The elephant's weight will cause the barge to sink several inches in the water. Draw a line on the barge's hull to mark the water level. Then direct the elephant back on to land. Load the barge with 100lb bags of sand (or whatever is handy) until it sinks to the line marked on the hull. The elephant weighs as much as the sand.

The style of interviewing at Google is indebted to an older tradition of using logic puzzles to test job candidates at technology companies. Consider this one: the interviewer writes six numbers on the room's whiteboard – 10, 9, 60, 90, 70, 66. The question is, what number comes next in the series?
Most of the time, the job applicant stumbles around, gamely trying to make sense of a series that gives every indication of being completely senseless. Most candidates give up. A lucky few have a flash of insight.

Forget maths. Spell out the numbers in plain English, which gives you the following: ten nine sixty ninety seventy sixty-six. The numbers are in order of how many letters are in their names. Ten is not the only number you can spell with three letters. There's also one, two and six. Nine is not the only four-letter number; there's zero, four and five. This is a list of the largest numbers that can be spelled in a given number of letters.

Now for the payoff: what number comes next? Whatever number follows sixty-six should have nine letters in it (not counting a possible hyphen) and should be the largest nine-letter number. Play around with it and you'll probably come up with ninety-six. It doesn't look like you can get anything above 100, because that would start "one hundred" requiring 10 letters and upwards. You might wonder why the list doesn't have 100 (hundred) in place of 70 (seventy). "Million" and "billion" have seven letters, too. A reasonable guess is that they're using cardinal numbers spelled in correct stylebook English. The way you write out the number 100 is "one hundred".

At many of these companies, the one and only correct answer is 96. At Google, 96 is considered to be an acceptable answer. A better response is 10,000,000,000,000,000,000,000,000,000,000,000,000,000, 000,000,000,000,000,000,000, 000,000,000,000,000,000,000,000,000,000, 000,000,000. Aka "one googol".

That's not the best answer, though. The preferred response is 100,000,000,000,000,000,000,000,000,000,000,000,000,000, 000,000, 000,000,000,000,000,000,000,000,000,000, 000,000,000,000,000,000,000,000. Ten googol.

Puzzles such as this have drawbacks as interview questions. The answer here is a simple matter of insight: either you get it or you don't. There isn't a process of deduction to relate, and thus no way to distinguish someone who solves the problem from someone who already knew the answer. At Google, of all places, anyone applying for a job knows how to use a search engine. It's expected that candidates will Google for advice on Google interviews, including the questions asked. Consequently, Google encourages its interviewers to use a different type of question, more open-ended, with no definitive "right answer".

Read the complete Guardian article for more questions and answers. 

No comments:

Post a Comment