Join 3,564 readers in helping fund MetaFilter (Hide)


A knowlege of Java may also be required
December 26, 2011 7:15 PM   Subscribe

How to Ace a Google Interview
posted by Artw (146 comments total) 39 users marked this as a favorite

 
So... be smart and knowledgeable about a variety of subjects?
posted by fnerg at 7:16 PM on December 26, 2011 [5 favorites]


Be Atom Ant?
posted by maudlin at 7:20 PM on December 26, 2011


Disapointingly the time I interviewed for a FTE position at Microsoft, who are favous for this kind of crap, it was out of vogue and I just got asked about linked lists and big-O notation and other CS crap. Then the next few places after that were all about patterns because that was the thing of the time.

(When I interviewed as a v-dash they barely asked me anything, because v-dashes are not really considered human and are therefore not worth expending energy on.)
posted by Artw at 7:20 PM on December 26, 2011


Jim doesn't see where a word like "realistic" comes into play—unless Google has a shrinking ray.

Well they don't have one yet but what do you think Google Ads is funding hmm?
posted by The Whelk at 7:21 PM on December 26, 2011 [3 favorites]


These types of questions are really a roundabout way of not actually getting to know candidates. Rather than spend a reasonable amount of time learning how someone works, putting them at ease so you see a glimmer of their true personalities, asking them actual, real-life questions about the job itself, companies are more interested in personality tests and clever-clever "gotcha" questions like this.
posted by xingcat at 7:22 PM on December 26, 2011 [30 favorites]


I've interview at google, twice. I didn't get the blender question (or any question resembling it). All my questions were typical algorithm style questions.
posted by schwa at 7:24 PM on December 26, 2011 [1 favorite]


No wonder we got Wave and the new Reader interface from that.
posted by Threeway Handshake at 7:24 PM on December 26, 2011 [23 favorites]


For more practical advice on preparing for a Google software development interview, read Steve Yegge's magnum opus (note: "Yegge-" is a metric system prefix for "way too much to read") Get That Job At Google.
posted by Harvey Kilobit at 7:24 PM on December 26, 2011 [3 favorites]


2004 Reaction to Google's hiring practices: Oh how neat! So quirky and earnest!

2011 Reaction to Google's hiring practices: Oh fuuuck you!
posted by basicchannel at 7:25 PM on December 26, 2011 [73 favorites]


Come to think of it, current place of employment did ask me a lightbubble puzzle question that is straight out of the Microsoft playbook, but it was more in jest than anything else.
posted by Artw at 7:26 PM on December 26, 2011


I think we have a new idea, hire someone here at Metafilter LLC. to come up with and review questions to see if you're worthy enough to be a MeFite.
posted by wheelieman at 7:32 PM on December 26, 2011


How To Ace A Google Interview?

Well, for starters, don't be older than 40.
posted by hippybear at 7:33 PM on December 26, 2011 [25 favorites]


There seems a lot of selection bias in articles about Google interviews. I interviewed to be a site reliability engineer at google and it went on for hours and hours with multiple different interviewers. There was a few, shall we say, airy-fairy questions, like the ones in the article, but they were right at the beginning and were only a very small part of what we talked about over the day. For the vast majority of what we covered, you could draw a pretty direct line back to system engineering. On the other hand, I didn't get the job, so maybe those 20 minutes at the start were more important than I thought. ;)
posted by adamt at 7:33 PM on December 26, 2011


These types of questions are really a roundabout way of not actually getting to know candidates. Rather than spend a reasonable amount of time learning how someone works, putting them at ease so you see a glimmer of their true personalities, asking them actual, real-life questions about the job itself, companies are more interested in personality tests and clever-clever "gotcha" questions like this.

Eh. I've worked, interviewed, and interviewed others at top law firms, where the custom is to do what you describe -- try to have a conversation and ask easy, predictable questions to get a sense of personality and "fit." The result is that most candidates come in acting like they're on the campaign trail -- they know "what to say" and "what not to say," have rehearsed answers to questions, and you can end up with a pretty boring conversation despite your best efforts. It can still tell you something, but it's a frustrating process. I think I would learn more with questions like these, which at least have the virtue of being harder to prepare for, and might give you a sense of how the candidate thinks, rather than just how he interviews.
posted by eugenen at 7:34 PM on December 26, 2011 [1 favorite]


For the record if asked this question by Google I would talk at length about how I would use my new found superpowers for a career in supervillainy.
posted by Artw at 7:36 PM on December 26, 2011 [4 favorites]


2011 Reaction to Google's hiring practices: Oh fuuuck you!

A friend of mine endured their alternately inept and self important hiring process for six months. Toward the end, we agreed that the process itself was an answerless riddle: How could a company possibly want to hire the kind of personality she would have been required to have to put up with the chickenshit?
posted by mph at 7:38 PM on December 26, 2011 [25 favorites]


I do think it's stupid that the straightforward and correct answers to the blender question -- "jump to the side" and "duck beneath the blades" -- "don't get you a lot of points."

But I would be very impressed with anyone who came up with the right answer to the balloon question on the spot. That's a fun question.
posted by eugenen at 7:39 PM on December 26, 2011


My blender answer. "I would solve this problem like I solve every other problem I have, I google it."
posted by rakish_yet_centered at 7:41 PM on December 26, 2011 [32 favorites]


From everything I've heard, the crazy off-the-wall questions have been more or less banned at Google for quite some time now, in favor of straight up algorithmic, programming, and systems questions. The recruiter will actually ask you to rank yourself on a scale of 0-10 in various areas (algorithms, various languages, relational databases, etc.) and your own ratings are apparently used to guide what questions are asked of you. What you list on your resume rightfully influences the same.

For what it's worth, I'm currently in the process of interviewing with Google, and after making it through the initial phone screening, the recruiter was kind enough to provide me with a document enumerating a number of areas I should study in preparation for further interviews, along with references to useful white papers (GFS, BigTable, MapReduce, etc.) and textbooks (Corman's Introduction to Algorithms and Bentley's Programming Pearls, etc). All of it clearly ties into real-world engineering, and thus far, I've only been asked reasonable, if challenging, questions.

To wit, if you're going to build something at Google's scale, you have to be fluent in CS theory, and be able to reduce it to practice. You have to have a solid grounding in algorithms, Big-O analysis, data structures, etc., but also be able to understand the implications of how real-world systems work. It seems a bit arduous, but thus far, I'm having fun.
posted by SemiSophos at 7:42 PM on December 26, 2011 [15 favorites]


What does Stack Overflow say?
posted by Artw at 7:43 PM on December 26, 2011 [6 favorites]


I am old fashioned, I ask people to discuss the last couple projects they worked on, if they faced any unanticipated problems and how they dealt with them. How they have used any particular technology they have listed, sure you have SharePoint,MSMQ and like 20 other things on here but how did you use them to solve any specific problems. I want to see them get excited when they describe how they figured something out, or came up with a working solution out of fucking nowhere. Unfortunately half the time I end up with really awesome people who can barely program.

The last time I had an interview it was because a colleague asked me to come in and interview. He was managing a new project and wanted me to work on it. He said, it will be the CTO and two engineers in three separate interviews. One of the engineers teaches SQL, but not to worry about it since I wouldn't be doing SQL. First guy, the CTO asked me to diagram an architecture for some project, then asked me about various source control solutions. Second guy asked me standard C# shit. Third guy was the SQL guy, he asked me to rate my SQL skills from 1-10 and I said 2. He started asking me fucking execution plans for various types of joins and detailed questions about clustering. It was nothing but a dick measuring contest. At least they let me down easy by telling me they decided to hire a couple junior guys instead.
posted by Ad hominem at 7:43 PM on December 26, 2011 [1 favorite]


I talked to a couple of Googles at a SF party. Interesting conversation -- seriously. Two weeks later, I get a call - "we'd like to interview you. " I ask "what are the conditions of the job?" the answer "what do you mean?"

That was the day Google failed my interview.
posted by eriko at 7:43 PM on December 26, 2011 [17 favorites]


Just know what a Bloom filter is. That's how they tell if you actually read the three silly white papers the recruiter told you to read but nobody bothers reading because they sound so trivial.
posted by jeffburdges at 7:50 PM on December 26, 2011 [3 favorites]


The son of one of my friends worked at Google for a couple of years after getting his degree, then quit because it really wasn't the place he wanted to be. (He's saved enough money to hang out in Whistler for a year and is in no rush to look for another full time job, but will be going freelance instead.) I wish this post had come up a couple of days ago so I could have asked him about the interview process. It would have been nice to see our entire table of people try to answer the questions collectively or competitively.
posted by maudlin at 7:52 PM on December 26, 2011


These types of questions are really a roundabout way of not actually getting to know candidates. Rather than spend a reasonable amount of time learning how someone works, putting them at ease so you see a glimmer of their true personalities

For tech jobs, your personality isn't nearly as important as whether you know how to solve problems and whether you can think your way through unfamiliar situations.

It matters if you're easy to work with, if you're responsible, etc, but you're not going to figure that out in a job interview, people are good at faking that. You can't fake being a problem solver.
posted by empath at 7:56 PM on December 26, 2011


Take a shit in the blender, and force the gigantic smoothie-chef to clean it out. Float to the top in the washing process and hope for the best.

If I'm being honest with myself, I don't think I'd get the job.
posted by codacorolla at 7:59 PM on December 26, 2011 [19 favorites]


Empath: No, but you can, I suppose unfake it. I think my absolute best skill is to help other people talk through their problems. They can be technical problems, social problems, puzzles, whatever. When I'm working with someone else, we get things solved, fast. When I'm working on things by myself, even in subject areas where I should be knowledgeable, my problem solving skills are cut down a great deal. Can't really show what a good team worker I am by answering problems in an interview, even tho I'm great on the job.
posted by rebent at 8:00 PM on December 26, 2011 [1 favorite]


I have no idea why people submit to the entire interview process. It's dreamed up and run by "HR professionals", in my experience the most mentally unbalanced folks in the whole arcology.

As a rule of thumb, any place with an "interview process" in place is not worth bothering about.
posted by KokuRyu at 8:00 PM on December 26, 2011 [2 favorites]


I interviewed at Google. Six programming questions, one of which I tanked. In all that, nobody asked me what I was good at, what I wnted to work on, etc. It was surreal - nothing related to people, teamwork, collaboration. The good part was that I learned that I don't want to work at Google. Any place that suggests you can re-apply after studying for ridiculous questions like this ain't for me.
posted by parki at 8:01 PM on December 26, 2011 [4 favorites]


And Google doesn't even pay all that well
posted by KokuRyu at 8:03 PM on December 26, 2011 [1 favorite]


I've done this similar route with Microsoft. "Steve Ballmer asks you to invent an entirely new technology..."

My favorite was the task to redesign a 747. I said Boeing's customers are airlines, not Joe Schmo, so airlines like things like a lot of "mundane" things like fuel economy and rapid maintenance turnaround times. They apparently wanted 747s that could teleport.

Yeah, I'm done with them. I apparently can't do the secret handshake.

Moreover, since I meet a lot of Microsofties, and they all strike me as very drone-like and unrealistic. I fear they're unemployable outside of the campus. I see it on their resumes, and it's a strike against them.
posted by Cool Papa Bell at 8:04 PM on December 26, 2011 [5 favorites]


Most of these stupid interview questions aren't about the answer, they just want to know how you think and solve problems.
posted by hellojed at 8:12 PM on December 26, 2011


I interviewed at Google in 2007 and was essentially accepted for the position. I didn't get asked any weirdo questions. Then again, I was applying for a technical writer job. Maybe they save all the weirdo questions for techier folks.
posted by ErikaB at 8:15 PM on December 26, 2011


But I would be very impressed with anyone who came up with the right answer to the balloon question on the spot.

2. You're in a car with a helium balloon on a string that is tied to the floor. The windows are closed. When you step on the gas pedal, what happens to the balloon—does it move forward, move backward, or stay put?

That's one I thought I knew right away. If the balloon is just hanging in the air, it will appear to move backwards (at least until the string becomes taught or it is obstructed by some object in the car). Of course, it will appear to stay put to someone outside the car.

4. A book has N pages, numbered the usual way, from 1 to N. The total number of digits in the page numbers is 1,095. How many pages does the book have?

401?
posted by 3FLryan at 8:19 PM on December 26, 2011


My "answers":

1. Every finite sequence has an uncountable number of infinite extensions.

2. What gas or gases is/are filling the car (outside the balloon)?

3. Why, is my cell phone broken?

4. This is an absurd way to determine the page length of a book.

5. Why would anyone bet real money in a Monopoly game?


As for the blender, if I've been shrunk, I don't have 60 seconds, and it really doesn't matter anyway. I can't hold my breath that long, and my lungs wouldn't process oxygen given the change of scale.

I'm happy at my university, thanks.
posted by Jonathan Livengood at 8:21 PM on December 26, 2011 [7 favorites]


Pocket change? That is fucking turrible. I wouldn't hire that guy.
posted by nathancaswell at 8:26 PM on December 26, 2011


I don't think so, when you accelerate in a sealed car you don't feel wind on your face do you? A helium balloon has even less mass than air. Mass is fundamentally a measure of inertia is it not?
posted by Ad hominem at 8:27 PM on December 26, 2011


It's inertia. There's no part of you that is moving initially, yet you move forward. That's because you are in contact with the car and picking up its forward motion. That's why you feel pressed back into your seat when you are accelerating - your body is at rest and wants to stay at rest until the car acts upon you. The balloon is just hanging there, no contact with the forward energy. It has no energy until the string is taught and it picks up the energy from the car. Anything outside the car doesn't matter.
posted by 3FLryan at 8:31 PM on December 26, 2011


A long time ago I was a flower delivery guy. The answer to the balloon question is that as soon as you step on the gas pedal, the balloon attempts to occupy the space between your eyes and the windshield.
posted by Balonious Assault at 8:33 PM on December 26, 2011 [84 favorites]


I don't think so, when you accelerate in a sealed car you don't feel wind on your face do you?

No, but you do get pushed back into your seat. A pen on the dashboard will roll backwards. Coffee will slosh. Air is a gas that's filling the space that doesn't move around in the same way a very massive, non-gaseous object would due to the vacuum that would result. The balloon has a lot more mass than the air in the car.
posted by jimmythefish at 8:34 PM on December 26, 2011


The ballon has contact with the air, which is also being moved forward by the car. It has less mass, which is why it floats, it isn't you or a pen.

Oh well, here is an explanation of what happens. Mass and inertia do play a role.
posted by Ad hominem at 8:36 PM on December 26, 2011


But I would be very impressed with anyone who came up with the right answer to the balloon question on the spot.

I would have, but I spent a very happy half hour in a stick shift in a Toys R Us parking lot once.

"See how you think" questions can best be summed up as "I have to fill 45 minutes and some guy asked me this once."

As you get more experience in your field you start to have actual questions of your own to ask. I have yet to meet anyone over the age of thirty who thinks these things are a good idea.
posted by Tell Me No Lies at 8:37 PM on December 26, 2011 [1 favorite]


The air in the car is compressible. It has its own mass. As the car accelerates, the air stays stationary, on account of its inertia, compressing against the rear surfaces of the car, and flowing by whatever means back to the front. Suspended, as it is, in the air, the balloon moves with it, towards the rear of the car. Thus, the balloon moves slightly.
posted by klanawa at 8:38 PM on December 26, 2011


Note: I have no idea what the fuck I'm talking about.
posted by klanawa at 8:38 PM on December 26, 2011


The ballon has contact with the air, which is also being moved forward by the car. It has less mass, which is why it floats, it isn't you or a pen.

Oh well, here is an explanation of what happens. Mass and inertia do play a role.


Yep, would have blown that question for sure. Now I'm all prepared for my Microsoft interview, though, so thanks.
posted by 3FLryan at 8:39 PM on December 26, 2011


The balloon moves forward because it's essentially in a sea of denser balloons.

Start the 4 and 7 (no other choice), and restart the four when it runs out. When the seven runs out, restart it. When the four runs out, flip the seven for the extra one unit that's left in it.

digits:
the first 9 pages have 9 digits
the next 90 pages have 180 digits
1095-189 = 906
so, there are 203 more pages.
203 + 99 = 302
posted by esprit de l'escalier at 8:39 PM on December 26, 2011 [2 favorites]


I used to be a software engineer at Google. I left the company in 2006, so it's possible that crazy off-the-wall questions have been banned, but they certainly weren't while I was there.

While I was there, I was in charge of log analysis software. I liked to at least pretend that my questions were relevant to the job I was doing, so I usually spent at least half of any interview on the question, "Design a system to figure out the most popular queries Google gets on a given day."

That question is so vague that there really is no correct answer. There are certainly wrong answers; if someone outlined a system that I thought wouldn't work, then that clearly didn't help their standing. But typically everybody gave a reasonable answer to that question.

Then I started poking holes in their answer. The easy hole was typically: "Oh, I'm not sure you have enough memory to do that. How much memory does that take?" Did the interviewee use fixed lengths for their query strings? Did they make reasonable assumptions about the distribution of popularity of queries?

They generally arrived at the answer that this just wouldn't fit in the memory of a single computer. Even if they were overly optimistic about things, I could force that issue by saying that they were running on cheap computers without that much memory. So then they had to solve that problem somehow.

There were three good approaches people had to solve this problem.
1. Batch it up so that you don't need to fit it in memory.
2. Compute an answer that is probably close to correct but not guaranteed.
3. Use parallel computation so that even if a single computer can't store it, you still have enough memory to do it somehow.

The first of these I would dismiss immediately: "Oh, I want to compute this in real-time, so I can use it to monitor today's trends." (There were people who pushed back on this and batched up intervals of the query stream small enough to fit in memory, so this was still one of the "good" answers I heard.)

So, I just gave you the "answers" to my favorite Google interview question without actually telling you anything that would help score you well. Here's some real advice: the best answers always started with, "It depends, what exactly are you looking for?" Why am I looking for popular queries? What does it mean to be popular? What am I doing with them? What kind of resources do they have to solve the problem? If my interviewee got stuck, did they ask questions, or did they flounder?

Now we get to the crux of it: I wasn't asking to see if my interviewees could come up with "the right answer." My question was designed such that everyone came up with a right answer. I was more interested in seeing how they came up with their designs.

And clearly, one of the interviewers who interviewed me thought along the same lines. When I completely flubbed "write some code to reverse a linked list," he came back and asked me the same question at the beginning of the next interviewer's time. A bit nerve-wracking, sure (hadn't I just flailed at this for the last five minutes?) but I did manage to show that when I got stuck at something, I could take a step back and start over without wasting time fixing my old fundamentally flawed solution. So even though I was asked a pretty basic coding question (and to this day it's embarrassing to have cracked under pressure and gotten it wrong) that interview did result in illustrating some of the things I later looked for in my own interviews.
posted by grae at 8:40 PM on December 26, 2011 [10 favorites]


Crap I missed the helium part...
posted by jimmythefish at 8:41 PM on December 26, 2011


I suspect this thread is secretly a interview for a high tech startup.
posted by Ad hominem at 8:42 PM on December 26, 2011 [2 favorites]


The air in the car is compressible. It has its own mass. As the car accelerates, the air stays stationary, on account of its inertia, compressing against the rear surfaces of the car, and flowing by whatever means back to the front. Suspended, as it is, in the air, the balloon moves with it, towards the rear of the car. Thus, the balloon moves slightly.

But, everything inside the car can't move backwards or else a vacuum would form at the front of the interior. Yes, there's a hypothetical force (proportional to mass) on everything pushing everything back. But, there's a hypothetical force due to the vacuum pulling them forward, which is proportional to volume. Thus, denser things are pulled backwards, while less dense things are pulled forwards. Since the balloon floats inside the car, it must be less dense than the surrounding air. Thus, the balloon moves forwards.
posted by esprit de l'escalier at 8:45 PM on December 26, 2011 [1 favorite]


Helium balloon in a car.
posted by mullingitover at 8:45 PM on December 26, 2011 [11 favorites]


The correct answer to every question: "I can see that you're a man of taste. Perhaps I could introduce you to my friend, here, Mr. Abraham Lincoln."

Also, when they say "Big-O analysis," don't giggle.
posted by evidenceofabsence at 8:46 PM on December 26, 2011 [2 favorites]


But, everything inside the car can't move backwards or else a vacuum would form at the front of the interior

Aha! THAT is the "big picture" explanation I was craving. Thanks.
posted by 3FLryan at 8:46 PM on December 26, 2011


Another way to see that a helium balloon moves forward when you accelerate (assuming that it is ordinary air in your car and not hydrogen), is to note that being in an accelerating reference frame and being in a gravitational field are equivalent. Then ask yourself, what happens to a helium balloon in ordinary air near Earth's surface?
posted by Jonathan Livengood at 8:47 PM on December 26, 2011 [6 favorites]


I like to think I could have resolved the balloon issue on my own if I had not just googled it. Programmers these days, coding by google.
posted by Ad hominem at 8:47 PM on December 26, 2011


esprit de l'escalier: "But, everything inside the car can't move backwards or else a vacuum would form at the front of the interior. Yes, there's a hypothetical force (proportional to mass) on everything pushing everything back. But, there's a hypothetical force due to the vacuum pulling them forward, which is proportional to volume. Thus, denser things are pulled backwards, while less dense things are pulled forwards. Since the balloon floats inside the car, it must be less dense than the surrounding air. Thus, the balloon moves forwards."

Yeah, having seen the video, that makes perfect sense.
posted by klanawa at 8:52 PM on December 26, 2011


Hey asavage, I think you have a car, some helium, and a balloon to buy, and a myth to test.
posted by fnerg at 8:58 PM on December 26, 2011 [3 favorites]


digits:
the first 9 pages have 9 digits
the next 90 pages have 180 digits
1095-189 = 906
so, there are 203 more pages.
203 + 99 = 302


If there are 906 digits left, and all we have left are page numbers with 3 digits, wouldn't we need 906/3 numbers (i.e. 302 numbers)? And we used 1-99. So we need 100-402. So N=402.
posted by 3FLryan at 8:59 PM on December 26, 2011 [2 favorites]


err 401
posted by 3FLryan at 8:59 PM on December 26, 2011 [3 favorites]


I would ask God, who apparently just shrunk me to the size of a nickel and told me I have 60 seconds to live, to either save me from the situation he just put me in, or just press "liquify" immediately and remove me from this malevolently capricious universe
posted by crayz at 9:01 PM on December 26, 2011 [5 favorites]


I have no idea why people submit to the entire interview process.KokuRyu

I can only speak for myself, but I honestly find it fun. Someone is getting paid to ask me logic questions and quiz me about Linux arcana! Yay! If I do well enough, I get to visit friends in California, hang out with smart engineers for a day, and solve even more logic problems. Whee!

(Plus, having "A-ha!" moments and getting things right are good ways to release a little dopamine.)

I might get a job offer, but I don't yet know if I would accept it, as I still have quite a bit to learn about Google's culture, the role I'm being recruited for, and the campus I would be working at. Still, those concerns are immaterial in the absence of a specific offer, so I'm content to set them aside for now.

For the moment, I'm having fun with the quizzes and enjoying the motivation to finally read through Skiena's Algorithm Design Manual, a pursuit that will serve me well regardless of my employer. I'm also fortunate enough to be happy in my current job, relieving me of the pressure of wanting or needing to find a new gig.

In my case, I received an email from a recruiter inviting me to interview, decided that I could absolutely see myself being fulfilled working for Google, and took her up on the offer. Because hey, why not? It's like pub trivia: There's often a nice reward for winning, but even if you come in dead last, you can still have a lot of fun getting there.
posted by SemiSophos at 9:04 PM on December 26, 2011 [1 favorite]


I worked with Poundstone years ago, and I'm not convinced that he's got any inside dope on all these big secret things. But, evidently his publisher thinks he does.
posted by Ideefixe at 9:15 PM on December 26, 2011


I would like to repeat the advice above: if you're actually curious about what a technical interview at Google is like right now, read this blog post by Steve Yegge.
posted by troublesome at 9:25 PM on December 26, 2011 [1 favorite]


Well jeez, the blender question is obvious. You climb on top of the rotor and stand on the axis of rotation. It rotates under you harmlessly.
posted by charlie don't surf at 9:25 PM on December 26, 2011


As someone who has interviewed people for technical positions for years, I have to say that I think the whole process is fundamentally broken and really no one has a clue how to accurately gauge talent or fit in an interview. A friend recently took a new job where they have a month long period where new employees are contractors rather than FTEs, and even though I think this is kind of terrible, it's also probably the best option. This doesn't absolve you from interviewing, but I think it can cut down on the BS significantly.

Personally, I hate interviewing for jobs and have avoided doing so for my last few jobs. I find it so stressful that if I can get away with continuing this trend I fully intend to do so.
posted by feloniousmonk at 9:34 PM on December 26, 2011 [2 favorites]


I hate What we called Microsoft questions, which now apparently are Google questions, because they told me jack about the people I interviewed, and I walked out of an interview when the line turned to "you have a measure size X and size Y find Z" for the thousandth time.

And yet, someone I interviewed for a position last week I asked for them to design a dialog in a way similar to grae's question. And yes, the worst thing you can do is start solving the question without asking requirements. Hell, I give away the game by leaving most of the crucial requirements out of the original description.

You want to win as an interviewee? Be smart, be honest, ask questions. You want to win as an interviewer? Be honest, probe deeply, don't be tricky.
posted by dw at 9:35 PM on December 26, 2011 [3 favorites]


I've often thought of trying to make an FPP about the questions the government asks you when you try to secret clearance, but I haven't seen any articles to link to. Some examples of questions that people I know have been asked: Is your shit ever black and tarry? When was the last time you got a blowjob from the vacuum cleaner?
posted by 445supermag at 9:37 PM on December 26, 2011 [1 favorite]


Some examples of questions that people I know have been asked: Is your shit ever black and tarry? When was the last time you got a blowjob from the vacuum cleaner?

I would love to hear the reasoning behind this stuff. I'm imagining the worst Myers-Briggs quiz ever. It's like they put OkCupid in charge of the questionnaire.
posted by feloniousmonk at 9:40 PM on December 26, 2011 [1 favorite]


OH MY GOD WILLIAM POUNDSTONE IS STILL ALIVE

One of my friends got me an interview, mostly because he mentioned my name as a brainstorming "people google might want to talk to" thing.

See, the last thing I wanted to do at that moment was work in system administration. Anywhere. At all. Even google.

So, the HR contractor who called me asked me five technical questions, basically none of which I could answer. I think one was what the default file permissions were in Linux or something. I literally said "Pass" a few times.

I didn't wait for the Google helicopter to land in my backyard.
posted by jscott at 9:45 PM on December 26, 2011 [1 favorite]


I've been interviewing candidates for a software engineering position lately, and as a fairly inexperienced interviewer, have found the experience interesting.

After the basic "getting to know you" and "tell me about your last/current job" kind of stuff, I've been starting off with a pretty basic question to help me gauge where to go next: write a function, in the language of your choice, that takes in a string as input and returns a boolean indicating whether the string is a palindrome. If the candidate doesn't know what a palindrome is, I'll go over some examples and counterexamples. I make it clear that I don't, at least initially, care about efficiency.

The answer I'm really kind of hoping to hear is "return str.reverse().equals(str)" (or equivalent in another language). This solution is generally a horribly inefficient use of basically every resource except the programmer's time, which is often the most valuable resource we've got. I have had only one candidate give me this answer. Most candidates launch into looping over the characters in the string, and a good proportion do manage to stop the loop and return false when the characters don't match. But when I ask these candidates for another solution to the problem, I never get one, no matter what hints I give.

Personally, I do prefer the more open-ended design type questions like grae mentioned above. These types of questions, done right, mean I basically get to problem-solve along with the candidate. That experience gives me a far better sense of what it will be like to work with this person than making them regurgitate the contents of a data-structures textbook. Of course, given the aftermath of some of the tech company parties I've been to, some more attention to candidates' regurgitation proclivities might be in order (probably discriminatory however).
posted by zachlipton at 9:52 PM on December 26, 2011 [1 favorite]


Microsoft used to be notorious for asking goofy brain-teaser questions. Looking back, I bet that wasn't accurate either (I know I didn't get any when I interviewed there) -- these lists of reputed questions probably live forever and just get a new company name tacked on every ten years.

I can't swear nobody has ever been asked one of these in a Google interview, but in many hundreds of questions I have seen asked, none has been like this.

What do I see a lot of? Coding questions. When programmers do interviews, they love to ask coding questions. It's hard enough to get them to leave enough room in their asking of coding questions to get good coverage of design, systems, algorithms -- think they're going to blow interview time on blender hypotheticals? When they could be asking coding questions?
posted by away for regrooving at 9:57 PM on December 26, 2011 [2 favorites]


Meh. Not my google interview at all. 1) They called me. 2) All algorithm questions 3) They weren't that hard. 4) The one interesting question they did ask me, I came up with a better algorithm than the questioner believed was possible.

And then the next week I went in for an interview with a much smaller company that asked questions that made me think, and that I couldn't solve off-hand.

Meh.
posted by cytherea at 9:58 PM on December 26, 2011


What's the consensus on the effectiveness of Microsoft's historic goals in hiring? It seems that 20 minutes after Microsoft declared they were going to only hire the "smartest people" that it got pretty dull and predictable.

I've had the feeling that Google is kind of stagnating. I was under the impression that AdWords is the only technology they have that makes money, and they bought that.

Are these companies missing some sort of Loki factor?
posted by dglynn at 9:58 PM on December 26, 2011 [2 favorites]


I interviewed at Google about a month or two ago. It was nothing like the article implies, except that the interviewers spend the whole time taking notes, staring at their laptops, looking like they're ignoring you, which seems pretty rude to me.

Parki said:
I interviewed at Google. Six programming questions, one of which I tanked. In all that, nobody asked me what I was good at, what I wnted to work on, etc. It was surreal - nothing related to people, teamwork, collaboration. The good part was that I learned that I don't want to work at Google. Any place that suggests you can re-apply after studying for ridiculous questions like this ain't for me.

This was largely my experience. None of the questions were stupid trick/gimmick questions like the blender question, but the whole experience was basically like taking one giant final exam for my entire Comp. Sci. curriculum over again. Several of my interviewers seemed smug and mildly put out to be talking to me.

I expected it to be technical, and I guess missing one question badly enough DQ's you. Ok, I can live with that. But the lack of attention to anything else was off-putting. One guy actually looked at my website (the URL is on my website) and asked me about a project I have listed there (I build surfboards as a hobby). He asked me a couple technical questions before that to be sure, but at least afterward he seemed to take an interest in me in general. He seemed like a cool guy and I think he would have been fun to work with, but he was one of seven different people I talked to that day, and mostly they seemed to be in generally the same state of mind that House is when he has to work the ER (if you've ever watched House, M.D.).

They didn't offer me the job. I will be working down the street for Evernote instead. I am OK with that.
posted by tylerkaraszewski at 9:59 PM on December 26, 2011 [1 favorite]


I don't know how many data points the writer of that article had, but I interviewed twice at Google and the questions asked (by about 10 total interviewers) were straightforward technical analysis. No weird logic problems or blender questions.
posted by CheeseDigestsAll at 10:03 PM on December 26, 2011


Assuming each of the candidates has the bare minimum qualifications for the position (you did fact check their resume and called references/previous employers), is there any evidence that these involved interviewing techniques produced better results than picking people at random?

Beyond that, supposing you did hire sub-optimal people, doesn't the company want to maximize the return on their investment by offering training or otherwise apprenticing for needed roles (which also builds loyalty)? If you can't maintain retention, the costs of rehiring and re-training far exceeds the benefit of holding out for the perfect candidate.

And for these companies that supposedly value creativity and problem-solving, is the company structured to take advantage of the geniuses in their midst?

Given that the best way to find exemplary employees is to make your business attractive to them (and they will find you), I suspect most of these companies fail at rather rudimentary business practices.
posted by quintessencesluglord at 10:11 PM on December 26, 2011 [2 favorites]


Bend up the blades so they can't hit you?
posted by Mei's lost sandal at 10:12 PM on December 26, 2011


I interviewed at Google about a month or two ago. It was nothing like the article implies, except that the interviewers spend the whole time taking notes, staring at their laptops, looking like they're ignoring you, which seems pretty rude to me.

My interviews at Google this year were also nothing like the silly brain-teasers the article discusses. The interviewers did need to take notes about the interview, but for the most part they scribbled down notes about my whiteboard-work on paper, or took periodic photos of the whiteboards to help them write up the interview later.
posted by JiBB at 10:16 PM on December 26, 2011


My answer to question number 4 would be: "I am not authorized to response" and would hope for some extra geek points.
posted by Henrik at 10:19 PM on December 26, 2011


Dunno man, Microsoft is doing a lot of things right in the last couple years.

Recently they contributed code to Redis. Granted that is to protect windows server marketshare, but the Microsoft of yesteryear wouldn't have done that.

Contrary to what people think, I think Blamer has been good for them. Since he lacks a technical background he is less likely to reject ideas not invented at Microsoft and has made Microsoft less insular.

Hell, they recently announced they are going to expose the office API to Javascript so people can move off VBA. Fucking JavaScript.

This is just an outsider's perspective.
posted by Ad hominem at 10:21 PM on December 26, 2011 [1 favorite]


I worked at Google for five years and gave hundreds of interviews there for engineers.

This article is basically bogus.

Most of the comments here seem pretty bang on the money, though. :-D
posted by lupus_yonderboy at 10:21 PM on December 26, 2011 [4 favorites]


You are shrunk to the height of a nickel and thrown into a blender. Your mass is reduced so that your density is the same as usual. The blades start moving in 60 seconds. What do you do?

Rough plan: 1 disable device < 30 seconds. 2. escape blender 3. return to normal size. 4. Revenge upon those who did this to me.

Key questions: blender contents and fullness? actual height (are we talking diameter of nickel or height as measured when nickel lies flat? Relative density to contents? Blender model and design?

Based on these answers I Macgyver my way through steps 1 and 2. Step 3 requires research and investigation. Step 4 requires lawyers, guns and money.
posted by humanfont at 10:27 PM on December 26, 2011 [3 favorites]


I aced both questions immediately. The balloon question is an old physics brain teaser from decades ago, and the blender question instantly triggered my bug nerd persona to answer with "I'd jump like a flea of course."

Now accepting job offers.
posted by Aquaman at 10:39 PM on December 26, 2011 [1 favorite]


Y'know, if I was Google, and I was going to hire either Steve Yegge, or William Poundstone based on the contents of the FPP article and the aforelinked blog post, I'd probably hire Steve Yegge.
posted by fnerg at 10:55 PM on December 26, 2011


Wait, isn't William Poundstone basically retreading his book from 2004, replacing Microsoft with Google?

I guess that shouldn't be surprising.
posted by curse at 11:18 PM on December 26, 2011 [1 favorite]


You are shrunk to the height of a nickel and thrown into a blender. Your mass is reduced so that your density is the same as usual. The blades start moving in 60 seconds. What do you do?

Activate the remote rescue drones put in place two nights ago, and use Grow Ray, developed some years previously for just such an eventuality, to return to normal size. Don night-vision goggles and cool new armored gauntlets. Locate goons, kick them in the jaw and use hoarsely whispered threats to get the location of the Sous-Chef de Sous-Sanity from them. Break into Sous-Chef's abandoned-cafeteria hide-out and beat him into submission while delivering moral lessons and philosophical observations about nature of evil (again, in hoarse whisper). Tie Sous-Chef up with own apron and dump him on the steps of City Hall for Commissioner Gordon to take into custody. Stitch wounds and research kitchen-themed crimes during day, then return for rooftop meeting with Gordon that evening and share findings. Disappear mysteriously in the few seconds it takes Gordon to turn from the wind and light a cigarette.

Easy. Now ask me about my plan to deal with Superman gone rogue.
posted by No-sword at 11:24 PM on December 26, 2011 [2 favorites]


Wait, isn't that blender thing a plot point of one of the Honey I Shrunk the Kids sequels?
posted by feloniousmonk at 11:37 PM on December 26, 2011


If it's a Blendtec, I'd stand in the corner and hope that the airflow didn't draw me towards the spinning blades.
posted by SillyShepherd at 12:19 AM on December 27, 2011


William Poundstone tangent, though: he wrote The Recursive Universe. I thought that book was awesome. It had Conway's Life as a framework, intertwingled with information theory, Maxwell's Demon, Borges' Library of Babel (for movies, I forget why movies), self-description, self-reproducing Life patterns. I was in high school and it's possible the book just hit me at the right age, but it would make me sad to find that out.
posted by away for regrooving at 12:47 AM on December 27, 2011 [2 favorites]


If I'm the size of a nickel but my mass is reduced so my density stays unchanged, I should be able to easily jump out or climb out using the surface tension of my spit on my hands.
posted by Megafly at 1:23 AM on December 27, 2011


hipppybear nailed it about not being over 40, I sat through 5 interviews with the last two being earnestly questioned about whether I thought I could 'fit in' with less 'experienced' folk. I forget exactly what was said as it was several years ago, but he last guy even made like 3 comments about my age during the interview. Chuckleheads.

I came home after being dinged and CNBC had a lengthy segment on how/why it was the best place in the world to work...
posted by sfts2 at 1:47 AM on December 27, 2011 [2 favorites]


Surely the Google way of replying would be to copy the answer of the most successful other applicant and then sell all the information you can glean about he interviewer to the other candidates.
posted by Grangousier at 1:51 AM on December 27, 2011 [4 favorites]


You are shrunk to the height of a nickel and thrown into a blender. Your mass is reduced so that your density is the same as usual. The blades start moving in 60 seconds. What do you do?

Well, I have a belt made of white dwarf star material for exactly this kind of situation... *shrinks, jumps into ear of interviewer, begins kicking holy hell out of eardrum*
posted by benzenedream at 2:41 AM on December 27, 2011 [3 favorites]


Assuming each of the candidates has the bare minimum qualifications for the position (you did fact check their resume and called references/previous employers), is there any evidence that these involved interviewing techniques produced better results than picking people at random?

I studied industrial psychology, with a lot of focus on the validity of different recruitment methods. It's interesting that you mention calling previous employers - that has basically zero validity in terms of predicting future performance.

That said, there are methods that actually work better than chance. And one of them is mentioned in the article: "There is significant evidence that "work sampling," the use of tests similar to the work being performed, is a better predictor of future performance than the usual job-interview chit-chat. Google does a lot of work sampling, such as requiring coders to write code in the interview.". The research I've read agrees with this.

Situational interveiewing, done properly, can have predictive validity too - that's asking people to describe particular problems, and how they dealt with them. [Though I will say I studied this a while ago, and given that everyone is aware of situational interviewing now, and probably has some set answers to the most common questions, it may not be so effective now].

So, these particular questions? They don't look great to me - you're not hiring someone to drive around with a helium balloon in a car, you're hiring them to programme - so see how they programme. On the other hand, the article suggests that this is meant to be a work sample test for innovation. I don't know if there's any literature on the effectiveness of this approach.

TL;DR: there are recruitment methods that are better than chance. Unfortunately, HR and managers aren't necessarily aware of them, and don't necessarily use them. A combination of structured interviewing and work sample tests is likely to be the most effective approach. Psychometric tests (which could be intelligence or even personality) also have some effectiveness.
posted by Infinite Jest at 2:43 AM on December 27, 2011 [1 favorite]


Personally, I hate interviewing for jobs and have avoided doing so for my last few jobs. I find it so stressful that if I can get away with continuing this trend I fully intend to do so.

I'd love to hear how that works.

My last job interviews were two and three hours. I asked more questions and learned more about my interviewers than they did about me. I got the job.
posted by SteveInMaine at 3:28 AM on December 27, 2011 [1 favorite]


Actually, for the blender thing, I'm pretty sure you could just jump out. Remember, gravity isn't scale invariant. A very small object accelerates at the same rate we do, and beyond that your mass would scale down at a cube compared to your height.

Actually, they mentioned that at the end of the article. But whatever, still figured it out before getting there.
posted by delmoi at 3:34 AM on December 27, 2011


There's no secret to it, SteveinMaine, it's just luck and networking. The more people to whom you are a known and reliable quantity the easier it becomes to bypass the hiring game.
posted by feloniousmonk at 3:45 AM on December 27, 2011 [2 favorites]


Although I understand the reasons, really guys some questions I can not stand overly absurd for a job interview :-) :-(
posted by techno blogger at 4:17 AM on December 27, 2011


That helium balloon demonstration is odd. I know the answer. The answer makes sense to me. I can explain why it's the answer. But when I see the balloon doing exactly what I know it will do in a car, it looks weird and wrong.
posted by jeather at 5:24 AM on December 27, 2011


Growing up and going on car trips we had a book of these brain teasers sitting in the back with us. I didn't realize it was our parents preparing us for google training.
posted by raccoon409 at 5:44 AM on December 27, 2011


Manhole covers are round because manholes are round.
posted by thelonius at 5:51 AM on December 27, 2011 [3 favorites]


I interviewed at msft in early 1996, when riddle/puzzle questions were popular at least among some of my interviewers. One of them asked the question where you have a 3 gallon bucket and a 5 gallon bucket and you need to get 4 gallons of water. I managed to figure out the answer, then realized it seemed familiar and realized I had seen that exact puzzle in "Die Hard 3" several months before. I noted this fact to the interviewer.

The next engineer who got interviewed by this guy, the guy asked the same question but first asked if he'd seen "Die Hard 3".
posted by rmd1023 at 5:52 AM on December 27, 2011 [7 favorites]


All this, and no one picked up on what bugged me in the intro: the interviewer showed up late and sweaty.

That right away, unless I'm unemployed and any job is a good job, would have me walking. Disrespectful and jerky. (yeah, yeah, it's an amalgamation, not a single story, but c'mon, really ? )
posted by k5.user at 6:45 AM on December 27, 2011


The balloon question seems simply solved based on observations of a helium balloon in a car when i was a kid - but this was side-to-side motion, not front to back. With a balloon anchored in the center of the passenger space we observed that a balloon appeared to "lean into" turns, like it knew which way we were turning. This turns out to be purely an affect of body roll, and has nothing to do with gasses moving around in the passenger compartment. During a turn, the car body's normal upright vector instead points to the outside of the turn, but the balloon stays upright to earth's gravity (not being influenced by the car's body). For an observer in the car the divergence between the balloon's earth-upright vector and the car body's tilted-to-the-outside-of-the-turn vector _makes it look_ like the balloon is leaning into the turn, but really it's the car that is leaning out.

Extended to the front-and-back example, I would think that the initial body squat (front end of the car up and back end down) caused by the acceleration would make it look, to an observer in the car, like the balloon would move slightly forward. This seems to be the essence of the bubble-level explanation in the link.

So the proper answer seems to me to be: If the transmission is in a forward gear (an unstated assumption) to an observer in the car it will look initially like it is moving forwards WRT the car's interior, but to an observer outside the car the balloon will move forward with the car.
posted by achrise at 6:48 AM on December 27, 2011


When I interviewed at Google, they had an Apple employee and a Microsoft employee tied to two chairs. They handed me a loaded pistol and asked me which one I would shoot.

I immediately shrunk down to the size of a nickel and escaped through a vent.
posted by orme at 7:07 AM on December 27, 2011 [15 favorites]


That's one I thought I knew right away.

Fundamentally, the answer is relativity. Gravity is merely just another force, producing acceleration. If you are in a sealed box with no information coming in or out, it is impossible for you to tell if the 1G acceleration your are experiencing is the result of Earth's gravity (because the box is sitting on the ground) or not (because the box is in a spaceship accelerating at 1G.)

So, you know what a balloon would do under a 1G acceleration in the direction of the Earth's center of gravity. From that, it's easy to figure out, as a general case, what the balloon will do under other acceleration vectors.
posted by eriko at 7:13 AM on December 27, 2011


Glad I'm a musician and never have to deal with this bullshit. If I suck, I don't get hired. If I play well, I'm known. Networking helps but we're still only as good as our last performance. American job interviews are so preposterous!
posted by ReeMonster at 8:18 AM on December 27, 2011


Situational interveiewing, done properly, can have predictive validity too - that's asking people to describe particular problems, and how they dealt with them. [Though I will say I studied this a while ago, and given that everyone is aware of situational interviewing now, and probably has some set answers to the most common questions, it may not be so effective now].

I've been reading Hire With Your Head, and his method is situational interviewing focused on prior work experience. The key difference is that the questions aren't focused on "describe hypothetical situation" but on "describe something you actually did."

I know this has worked well for people hiring creatives, tho I've had trouble adapting this for coders.
posted by dw at 8:29 AM on December 27, 2011 [1 favorite]


This turns out to be purely an affect of body roll, and has nothing to do with gasses moving around in the passenger compartment.

Body roll gives you a few degrees of lean at most. The balloons in the linked videos lean past forty-five degrees even at low acceleration. Think about how fast a helium balloon accelerates upwards given one g of downward force on the surrounding air, and how easy it is to produce a large fraction of a g of acceleration with a car.
posted by the atomic kung fu panda bandit inquisition at 9:21 AM on December 27, 2011


When I interviewed at Google, they had an Apple employee and a Microsoft employee tied to two chairs. They handed me a loaded pistol and asked me which one I would shoot.

You failed to demonstrate how you think outside the box. You shoot the Google employee who handed you the pistol.
posted by charlie don't surf at 9:47 AM on December 27, 2011 [2 favorites]


Job opening!
posted by maryr at 10:21 AM on December 27, 2011 [3 favorites]


The only winning move is not to apply.
posted by mendel at 10:22 AM on December 27, 2011 [4 favorites]


Huh, I did the hourglass problem differently. Put the 4 minute glass on its side and shake it to distribute the sand equally on either side. Sticklers can fine tune the split by suspending the middle of it on a string. Start the 7 minute glass, and when it runs out, tip the 4 minute (now 2 minute) glass up to start it running. It will finish at 9 minutes.

Not what they want to hear for a programming job, though.
posted by maniabug at 10:36 AM on December 27, 2011


When I interviewed at Google, they had an Apple employee and a Microsoft employee tied to two chairs. They handed me a loaded pistol and asked me which one I would shoot.

They videotaped my interview at Google.
posted by Tell Me No Lies at 10:41 AM on December 27, 2011 [2 favorites]


Yikes, it's been a while since I've interviewed anyone for a software position. But even twenty years ago, the best idea was to

a) show me your code - bring me some code that you've written and let me look at it. In the book Peopleware (DeMarco & Lister), they make the point that in any job that requires creatives, e.g. architecture, advertising etc, they usually ask the applicant to bring in a portfolio, samples of their most recent best work. Why don't software firms do this?

b) give me a 10 minute presentation on a design project you were involved with. It's not all about the tech stuff. The best software people are those who can communicate effectively, can handle questions on the fly, can see the big picture as well as the nitty-gritty details.
posted by storybored at 10:51 AM on December 27, 2011 [1 favorite]


I've connected an USB microcontroller to a complex set of door locks on the door of the interview room along with a linux laptop and the URL to the subversion repository with an undocumented project containing. The API for the microcontroller. In one hour you need to meet me for the f2f part of the interview across the hall. If you can't be on time don't bother to show up. You can pick up the phone and ask to leave at anytime should you like to give up.
posted by humanfont at 12:38 PM on December 27, 2011 [4 favorites]


I despise these sort of interview tactics, but I also find the classic "getting to know you" interview format tired and useless. I was hiring software developers a few years ago and I did something novel. After a bit of pleasantries I asked the person to look over a pertinent web development project problem we were facing.

The problem itself wasn't that interesting or hard, but by pair programming the thing together I was able to gauge if the candidate asked good questions and seemed level headed. A few bombed the exercise and I felt bad for them but it was instructive all around. I thought I could predict who would do well in a working interview but the results were surprising to me. Some people were exposed as having very little depth who were very well spoken and others really came out of their shell when given something concrete to do.

Since then I've become a big fan of doing a working interview. I can't imagine garnering much success asking people arcane questions.
posted by dgran at 1:05 PM on December 27, 2011


Since then I've become a big fan of doing a working interview.

I usually plan this for half of each interview. At least twice I've gotten solutions to my problem from the interviewee :-)
posted by Tell Me No Lies at 1:20 PM on December 27, 2011


storybored - most folks don't own the code the write. I do a little hobby dev work, but any professional code isn't mine to take with me or show around. Trade secrets and the likes ..
posted by k5.user at 1:23 PM on December 27, 2011


One of them asked the question where you have a 3 gallon bucket and a 5 gallon bucket and you need to get 4 gallons of water. I managed to figure out the answer, then realized it seemed familiar and realized I had seen that exact puzzle in "Die Hard 3" several months before. -- I watched Die Hard 3 last night...and when I read this article, that's the first thing I thought of: all those riddles!
posted by epersonae at 3:06 PM on December 27, 2011


Huh, I did the hourglass problem differently. Put the 4 minute glass on its side and shake it to distribute the sand equally on either side. Sticklers can fine tune the split by suspending the middle of it on a string. Start the 7 minute glass, and when it runs out, tip the 4 minute (now 2 minute) glass up to start it running. It will finish at 9 minutes.

That works if you start the 7 minute glass while you're fiddling around with the 4 minute glass, otherwise it's going to take more than 9 minutes.
posted by empath at 3:16 PM on December 27, 2011


This article is full of bullshit. Google gives interview training addressing most of the objections in the thread.
posted by Ayn Rand and God at 6:20 PM on December 27, 2011


I'm the head software architect at my current company. When I interview people, my goal is to figure out if the candidate is passionate and "gets it." I ask a number of "simple" questions and for the most part, I don't have any questions that I care about the face value answer. Really. Almost every question has a specific ulterior motive which has more to do with how they answer rather than what they answer.

I have every candidate write code. This part of the interview looks simple but brief but ends up taking an hour or more. So for more junior candidates, they get an "easy" question. After they code it up, I will usually say something like, "I see x bugs in your code, would you fix them for me?" If I'm feeling evil, I will say "I see at least x bugs in your code, how many do you see?" After fixes, then comes the part where I look at their solution and declare that it works now but I wouldn't ever release code like that - why (and this is usually because while the solution appears to be simple, most people write it badly - even if it works).

For more experienced engineers, they get a simpler question on the surface. I give them a code sample, they have to tell me what it does. Very few experienced people get it right. Then I make them hand compile it and run it. Then we have a long discussion about language semantics, compiler design, optimizer design, and so on.

Finally, I let the candidate interview me. Anything is fair game. I've gotten some interesting questions.

I should also mention that I start by telling them exactly what I'm going to do.

And I do this without any gimmicky questions. No puzzles. Puzzles divide the crowd into maybe 5 bins: heard it before, never heard it but can think on your feet, never heard it before but a creative thinker, never heard it before and doesn't solve problems under pressure well, never heard it before and not a good problem solver.

And there's the problem - unless there is a good ulterior in the question the bin sorting is near useless because not solving problems under pressure is indistinguishable from being not a good problem solver. Further, if I've heard the question before I can *totally* act like I've never heard it before and make it look like it I've never heard it and can think on my feet.

See the problem? The filter doesn't work.

Here's a blog I wrote about the importance of passion in your job and interviewing for it.
posted by plinth at 7:18 PM on December 27, 2011 [3 favorites]


not solving problems under pressure is indistinguishable from being not a good problem solver.

Unfortunately that applies to all sorts of interview questions, in particular code writing and bug spotting. I'm certain I've missed out on some deep thinkers in the interviews I've done. There's just no way to tell.

With people who clearly know what they're doing my interviews tend to run short, so we usually spend the remaining time trading evil problems. Normally I don't count that time towards the interview at all, although I have on occasion given bonus points for being able to explain evil clearly.

Last year I interviewed someone who got...

int a = {0,1,2,3};
int b = 2;
char c = 2;
printf("%d %d\n", b[a], c[a]);

,,, instantly and explained exactly what the compiler was doing. Very impressive for anyone who doesn't do compilers for a living.
posted by Tell Me No Lies at 9:36 PM on December 27, 2011


When faced with a difficult question, I always find it best to answer with another question...

"How about I just whip up a batch of gorgieres instead? Do you have any chopped ham? A dash of brandy would be perfectly delightful in this Morney sauce!"

Okay, that last sentence wasn't exactly a question. But can anyone argue that point? I think not!
posted by slogger at 10:23 PM on December 27, 2011


Crap, that should have been a[4] in the declaration.
posted by Tell Me No Lies at 12:46 AM on December 28, 2011


I had interview questions like this at a company in England. I won't get into specifics, but a co-worker was asked something like this:

A young man has seen the Texas Chainsaw Massacre movie. He then decides to get a chainsaw. He then goes out and kills a woman walking down the street.

Who is to blame for this?

A. The film studio and producers
B. The movie theater/cinema
C. The woman walking down the street
D. The man himself
E. The writers of the movie

You then have to state WHY the person or persons are to blame for this.
posted by Ms. Moonlight at 5:08 AM on December 28, 2011


Original or remake?
posted by Artw at 7:42 AM on December 28, 2011


Yeah, I actually killed a woman with a chainsaw after I saw the remake. It was THAT BAD.
posted by empath at 8:21 AM on December 28, 2011


"5. A man pushed his car to a hotel and lost his fortune. What happened?"

Are you kidding me? There's no way that was an actual Google question. It is an extremely culturally loaded question that depends on specific experiences that not all suitable candidates might have. If you know what it references, you pretty much have the answer without any problem solving work at all. And if you don't know, no amount of problem solving or creative thinking will get you to the correct answer.
posted by Deathalicious at 12:41 PM on December 28, 2011 [1 favorite]


"5. A man pushed his car to a hotel and lost his fortune. What happened?"
[...] if you don't know, no amount of problem solving or creative thinking will get you to the correct answer.


Sure it will. The trick is to give those questions the answer they deserve.

For example: All of the man's belongs (including his money) were simultaneously being transported by a Cessna that was being guided by the bonfire on the thirty foot tall metal tower that was welded to the top of his car. When he pushed his car into the parking garage the tower fell over, setting off a series of explosions in the outside parking lot that disoriented the Cessna pilot and caused her to become lost for a few seconds. Fortunately she was able to fairly quickly orient herself and land the plane on the front driveway. Later, the man fell in love with the desk clerk and they were married in a lavish wedding that included 12 live penguins.

Then you challenge the interviewer to prove this answer false using the information given in the question.
posted by Tell Me No Lies at 12:56 PM on December 28, 2011 [6 favorites]


For example: All of the man's belongs (including his money) were simultaneously being transported by a Cessna that was being guided by the bonfire on the thirty foot tall metal tower that was welded to the top of his car. When he pushed his car into the parking garage the tower fell over, setting off a series of explosions in the outside parking lot that disoriented the Cessna pilot and caused her to become lost for a few seconds. Fortunately she was able to fairly quickly orient herself and land the plane on the front driveway. Later, the man fell in love with the desk clerk and they were married in a lavish wedding that included 12 live penguins.

False. It was 11 penguins, and they were dead because of a gas leak in the armored zoo semi-truck that transported them transcontinentally from Cape Horn.

Enjoy working at a second-rate tech company, loser.
posted by 3FLryan at 4:04 PM on December 28, 2011 [2 favorites]


Crap, that should have been a[4] in the declaration.

Which explains why I was thinking Is that not C? It looks almost like C, but not quite?
posted by wierdo at 4:07 PM on December 28, 2011


Enjoy working at a second-rate tech company, loser.

Like I'd work for a company that asks people about dead penguins in interviews.


Which explains why I was thinking Is that not C? It looks almost like C, but not quite?

Probably. The stuff in the printf makes people say the same thing though, and it *is* correct.
posted by Tell Me No Lies at 4:09 PM on December 28, 2011


[it was a joke]
posted by 3FLryan at 5:39 PM on December 28, 2011


[and if you were just joking back, well, shoot]
posted by 3FLryan at 5:41 PM on December 28, 2011


Actually, that seems like a silly restriction. What if the company was a zoo, and you were a biologist specializing in penguin mortality?
posted by 3FLryan at 5:43 PM on December 28, 2011


I think the question is supposed to be hypothetical, not the job :-)

[I also am engaging in conversation of a jocular nature]
posted by Tell Me No Lies at 6:18 PM on December 28, 2011


I need a simple string parser that extracts numbers from a plain text document. The parser should create a list of found numbers along with their start and end position in the string. Strings will be in utf-8 and IS0-8859-p1 format. Please write a parser class, a unit test and documentation. You have 30 minutes. Here is some sample data:

16 people were killed and 14 people were saved in the fire.
The zip code is 80210

15 minutes into the exercise later. To help you out here is some test data from our customers. Please include these in your unit tests:

"two thousand four hundred people entered the contest. There were 16 winners"
"one 19 year old was involved in the accident."
"a hundredth of an inch"
"the seventieth time"
"3 radians"
"π is the radio of the diameter of the circle to its circumference"

Do you need additional time?
posted by humanfont at 6:57 PM on December 28, 2011 [2 favorites]


I've got an email screen question : Do you use stackoverflow.org or similar? If so, send me urls for several of your answers which you feel are particularly good.
posted by jeffburdges at 10:04 PM on December 28, 2011 [2 favorites]


> What do I see a lot of? Coding questions. When programmers do interviews, they love to ask coding questions.

I am disturbed and appalled that any so-called programmer would apply for a job without being able to write the simplest of programs.

> no one has a clue how to accurately gauge talent or fit in an interview

How do we hire when we can’t tell who’s right for the job?
posted by morganw at 11:42 PM on December 28, 2011


Why we don't hire programmers based on puzzles, API quizzes, math riddles, or other parlor tricks
posted by Artw at 3:22 PM on January 5, 2012


Interview Programming Problems Done Right
posted by Artw at 10:12 AM on January 6, 2012


It needs to be relatively simple. If it takes more than about 20-30 lines of Python to solve it's probably too complex;

20-30 lines in an interview? Pascal's Triangle and coefficients used for binomial probability?

Please tell me that was a joke.
posted by Tell Me No Lies at 10:20 AM on January 6, 2012


Agreed, Pascal's triangle should only require one line :
pascal = iterate (\row -> zipWith (+) ([0] ++ row) (row ++ [0])) [1]
Just don't let its clarity and simplicity tempt you into point-free form :
pascal = iterate (ap (zipWith (+) . (0 :)) (++ [0])) [1]
Via the Haskell wiki's Blow your mind. See also The Evolution of a Haskell Programmer.
(we're using memoization here by defining an infinite array, meaning you'll compute each value exactly once by calling pascal !! n)
posted by jeffburdges at 11:21 AM on January 6, 2012 [1 favorite]


Agreed, Pascal's triangle should only require one line :

perl -le '$_=1;print,s/(\d+) ?/+$1 $1/g,s/\S+/$&/eeg until$a++>15'

Just don't let its clarity and simplicity tempt you.
posted by benzenedream at 2:58 PM on January 7, 2012 [3 favorites]


« Older The New Dealers...  |  So if I'm thinking about this ... Newer »


This thread has been archived and is closed to new comments