Hacker News Reader

Comment by sdfx | Hacker News Reader

"How many gas stations are in the US? [...] the best response is to compliment the interviewer on offering such an interesting and intriguing question, then say 'I don’t know.' Offer to research the answer and get back to her or him."

Actually, this is a fermi problem. [1] The purpose of this question is to figure out if someone can break a question down into parts and estimate reasonable values. But even more so, it is supposed to test whether you are enthusiastic about strange and puzzling problems. [2]

Whether you think these questions are stupid or not, the right answer isn't "I give up, just google it", but rather "let me make some assumptions and calculate a value based on some estimates"

Having said that, I think the number is arround 100,000 [3]

[1] http://en.wikipedia.org/wiki/Fermi_problem [2] http://www.joelonsoftware.com/articles/fog0000000073.html [3] http://www.uh.edu/engines/epi975.htm

View on HN

My answer would be to give a completely random number and then justify it by quoting Dilbert:

Dilbert: Studies have shown that accurate numbers are not that much more useful than the ones you make up.

PHB: How many studies showed that?

Dilbert: 87

I probably wouldn't give this answer in an interview, im more of the "I don't know" group of people.

p.s. Dilbert strip: http://www.dilbert.com/strips/comic/2008-05-08/

Interesting. My first guess was 300,000.

The town I grew up in is about 1,000 people, and has a gas station. Most of the small towns around that one of comparable size also have a gas station. So I figured if the population is about 300,000,000 and people per gas-station is about 1,000, then there are about 300,000 gas stations in the U.S.

This is the way I would answered in an interview as well. I guess it heavily relies on the type of town you are from. Even smaller towns in remote areas tend to have a gas station. But I think whats important is not your answer, but how you came to the number, and 3x off is actually not bad for something you don't know anything about.

Funny, I came to the same estimate using my town of 30,000 with approximately 30 gas stations.

I used the same method but of course had different numbers to plug in: 4 gas stations (I am probably forgetting a few in parts of town I didn't frequent) in a town of roughly 10,000 people.

Maybe you'd get a better figure going by number of pumps per person and then <strike>making up</strike> coming up with an average number of pumps per station. For example, (I'm guessing that) the lone gas station in a 200 person down doesn't have a lot of pumps.

I came to 200,000 by guessing how many were in an area served by one high school, and trying to remember how many high schools are in the US. Odd, but I figured a high school grouping would be some indicator of economic activity and thus civilian gas needs.

If I was an interviewer and a job applicant told me that he would "google" the answer to a question, I would quickly show him the door and put his resume in the circular file.

Why? It's often a good idea to google for an answer. See the second result for "number of gas stations in the us", for example:

http://www.infoplease.com/askeds/number-gas-stations-us-1995...

If you want the candidate to show his reasoning abilities by making good guesses based on reasonable assumptions, you should tell him to do so beforehand.

I would be surprised by the amount of people who have the intuition to actually go to Google in the first place. It's a lot better than a "Hmm, I don't know--how many?"

It shows that someone is willing to go and seek out the answer on their own.

What do you do when you come across a simple fact that you need to know in order to continue?

I dunno, I know people who's problem-solving skills amount to "paste the error message into Google and try the first solution". I want people who can derive a solution from first principles because often that's what you need to do.

Pasting error messages in Google saved my ass a few times.

It is not necessarily the wrong thing to do, but it can't be all you got.

Of course not. Its just fast and in many(mostly trivial and boring) cases the problem is already solved. Lets just say that sometimes its a better idea to just google for something instead of trying to figure it all out and in some cases-reinvent the wheel.

If I were interviewing a candidate, I wouldn't ask pointless questions about "simple facts." I would ask about design methods and give small design problems.

I'm a hardware engineer. As shocking as this may be to some, much of what I need to know cannot be easily found on the internet. I have to crack open a book to find good high level information. With the rising popularity of open college classes, this will probably change, but we aren't there yet.

For programming, the internet can be a very good resource. When I was a full-time programmer, google served as a very good reference. I would open a browser before I took a look at my bookshelf.

I'm just the opposite.

I want practical, get the job done type people.

I don't want cowboys who are more interested in clever solutions and/or impressing others with their knowledge, "obvious" intelligence, or skillz.

I certainly don't want some douche sitting in his chair theorizing from first principals when they could have googled the answer or asked team mate for help.

Finally the era when what one knows was self-contained in their heads is over. My memory and knowledge are huge and they are kept in google and wikipedia.

I guess this makes sense, if you're hiring for Yahoo! Search.

Reply on news.ycombinator.com