Fog Creek Software
Discussion Board

How many gas stations in USA?

I know questions like "how many gas stations are there in USA" are asked in interviews, but I have no clue on how to arrive at an answer, or even to show thought processes.

Heck, I have no idea how many cities there are in the USA, how many of them have gas stations, or even whether there the number of gas stations in a given area remains constant throughout the country (I bet it doesn't.)

So how on earth am I supposed to arrive at anything close to meaningful?

And what do such nonsensical questions help prove in the first place? Shouldn't the ability to do your job matter more than "thinking outside the box" or whatever it's called?

Kenneth Branch
Friday, November 22, 2002

Let me have a crack at it.

USA population = 280M ? (I'm not sure about this any American should know the answer to this)

I'd say that a small town of say 5000 people would have 2.5 gas stations (gut feel)

Therefore a ratio 2.5 per 5000 people

Gas Stations = 280M* 2.5/5000 = 140,000

? Anybody Else?

Friday, November 22, 2002

Look at no 3. Impossible question.

Prakash S
Friday, November 22, 2002

Dear Kenneth,
                      Presumably you are asked these kind of questions because the ability to think round a problem of a type you have never seen before is part of "doing your job well."

                      One way of solving it you haven't thought of would be to use the internet to find out the number of employees working in the sector (hidden in government statistics somewhere), make a guesstimate of the average number of employees in one gas station, and do the math.

                      Another would be to work out the break even point in gasoline sales of a gas station, add on 30% and then divide the totlal sales of gas in the USA by that figure.

                        We have a load of these puzzles in the English reading book we use at my tech college; there are normally only two or three students that show any interest in them out of a class of forty, but they are always among the best half-dozen students.

Stephen Jones
Friday, November 22, 2002

Another way you could find out. Go to the web sites of the major comapnies and see how many gas stations they have that let you use their fleet card (Exxon Mobil have 16,000 +)

Then do a drive around and work out what proportion of gas stations in your area belong accept the cards you have info on. Phone up a few ex-lovers and ask them to do the same in their part of the States (tell them you're thinking of coming to see them but don't want to run out of gas!)

Equally you could simply tell the interviewer that you always mail these kind of questions to Steve Jones in Saudi Arabia together with a cheque and he sends the answer back asap.

And while we are on questions nobody can know the answer to why does may name have the long underline after it, and what has caused it?

Stephen Jones
Friday, November 22, 2002

Just read the Joel article... he says "Of course it's OK if they are radically wrong. "

I totally disagree with this. The very point of asking a question like this is to see how valid an answer the candidate will get.

This tests not only their ability to attack problems with limited information, but also on their ability to draw on knowledge not related to their field. The assumptions they make have to be fairly valid.

If for example, a candidate does not know that the population of the US is in the 250 million region, then I probably do not want to employ them. The candidate needs to demonstrate that they do have an interest in the world out there... that they know more than the latest gee whizz computer language (insert arcane topic in specialist area here).

At the same time, the candidate needs to be able to state whether their guesstimate is in the correct ballpark.

If you arrive at a figure of 5 million for the number of gas stations in the US, I would say that you really are way off the mark, and you should be able to do so yourself, and possibly revisit the problem, checking your assumptions and arithmetic.

Without this sense of realism, the entire excercise is pointless.

Friday, November 22, 2002

The National Geographic have just published the latest in the periodic surveys they do of the geographical knowledge of 18 -24 year olds in varying countries.

The survey is at

The first question was "What is the Population fo the USA?"

25% of the respondents in the US gave the correct answer (betwen 150 and 350 million), whilst 33% gave the answer "over one billion". The interesting thing is that 18 -24 year olds from every other country in the survey got the answer for the US population higher than the Americans themselves.

But statistics and gography are both fields in which the average US citizen is notoriiously ignorant.

Stephen Jones
Friday, November 22, 2002

Depends on what kind of answer you are looking for when asking the question.

If you are looking for out-of-the-box thinking then you do not worry about the answer, but if you want core competency + knowledge about the world in a candidate then yeah.

Prakash S
Friday, November 22, 2002

"that 18 -24 year olds from every other country in the survey got the answer for the US population higher than the Americans themselves."

Whoops! I mean the number of 18-24 year who got the right answer to the qeustion about the US population was higher in all other countries in the survey than in the USA

Stephen Jones
Friday, November 22, 2002

The reason Joel said it doesn't matter whether its the right answer or not is that no one cares what the answer is.  The point of the question is not the question itself, but the way the  interviewee would set about finding a way to get the answer.

Simon Lucy
Friday, November 22, 2002

That type of question is BS, pure and simple.  I've never asked one like it, nor have I ever been asked one like it.

Why is it BS?  Because neither the interviewer nor interviewee has any context with which to know whether the answer is close or not.

I'm not interviewing someone to teach "new math" nor to learn it.  Correct answers count, and that's what we interview to find out.  Everything about the interview should be screaming this:  We expect the highest quality of work from you, not some half-baked BS conjecture with respect to something you have no basis to know.  And, when you don't know something, we expect you to research a good and correct answer, not some POS, off by one, "almost" answer, but the bug free answer.  One that will endure the pounding of a million users with zero defects.

There is only one good answer to that question:  "I don't know, but if you need me to know, I can go find out."

Finally, there is a "type" of person this question attracts: L.A.W.:  Loudest Asshole Wins.  These are the same people that like to prefix a sentence with "Bill said".  Enough.

Nat Ersoz
Friday, November 22, 2002

My answer would something like:

"I used to work in a gas station, and we did around $6,000 in business in a day.  That's uh ... $42,000 a week.

Now, I'm pretty average, and I spend maybe $25 in gas a week.  And there's probably 1 car for every 1.5 people in america - i'd be more like 1 for 1.8 + corporate vehicles.

Now, at 250,000,000 people in america, that's 250 mil * 25 / 1.5 = $4,166 Million in gas per week.

Now, if we think as Darwinists, throw in Adam Smith's invisible hand of free markets, and mark-up for the increased cost of gas, I'd guess the average station brings in around $50,000 a week now.

(Writes on scratch paper) - That's the same as, um ...4.166 divided by 50, or 0.083, which is, uh, 83,000.  So, 83,000 gas stations in the US.

Notice the other guy said 140,000, so I'm off by a factor of < 2, which is pretty dang good for an impossible question. :-)

The other guy's answer was more straightforward than mine.  Point being, it's possible to approximate the answer to a problem with little data if you know how to think analytically. 

Now, what coder would you hire who either can't think analytically or isn't willing to work on problems outside his domain of expertise? :-)

just my $0.02 ....

Matt H.
Friday, November 22, 2002

Jeez, Nat. I thought the question was kind of BS myself, but that job sounds more fun than the marine corps, or wherever you happen to work.

Also, your use of "new math" is incorrect. New Math was about teaching 1st graders arithmetic using set theory, not about "reasoned guesstimates."

Just my a-hole comments of the morning, enjoy. ;-)

Friday, November 22, 2002

210,000 Gas Stations ( in 1993 )

Why derive with guesses when you can research with Google.

"In 1992 the Violence Policy Center released More Gun Dealers Than Gas Stations...210,000"

You'd want to double check their sources....

The point of the question isnt to get the right answer, but to show you can make reasonable guesses without getting all red in the face.

Empty Tank
Friday, November 22, 2002

I’ve been asked this question once in an interview. They told me I had the best answer they had heard. My answer:

“I have no earthly idea. But it would be easy to find out. I’d pop over to and look up the NAICS code and data for Gasoline Stations. Then I could tell you how many there were.

Of course, I would then make sure I forgot that information as quickly as possible. That is not the type of information one should know of the top of their head. It’s too weird.”

Some notes:
NAICS replaced the old SIC codes used for tracking industry.
The Gasoline Stations NAICS code is 447.
There were 126,000 stations as of 1997.
They employed 922,062 people
They had sales of $198,165,786,000 (ok, we are in the wrong business guys).
If for some sick reason you care, this information is available at

Friday, November 22, 2002

Cool post Marc,

Now just for the record, MTV is heavily closed captioned (for the hearing impaired) while infomercials are not.

Why is that?  I'm using this as the basis for my next interview question.  If the interviewee doesn't kick my ass so hard that my feet are liberated from the ground, then its a "no hire".  Sorry.

Working on Geode captioning this week.  Been subjected to rap music all week.  Does it show?

Some cracker named Bella on MTV right now, waggin' his finger, Yo.

Nat Ersoz
Friday, November 22, 2002

I was asked this in an interview recently. I said I'd call BP and ask how many petrol stations they have. Multiply by four (they have about 25% market share).

Then I gave them a whole bundle of "book" answers to it and mentioned that it's not a good question to ask.

I still got offered the job...

Katie Lucas
Friday, November 22, 2002

One of my Physics professors was notorious for giving
his students impossible problems and seeing if they came up with good approximations. The story runs that this conversation once took place:

Prof: What is the inductance of a wedding ring?

Student: 4.18317, sir

Prof: (after a slight pause) And what units is that in?

Student: Arbitrary units, sir

David Clayworth
Friday, November 22, 2002

"Then I gave them a whole bundle of "book" answers to it and mentioned that it's not a good question to ask."

I've had a few points in interviews where I've politely critiqued 'broken' questions and I think got more respect than i would have from scrambling for an answer to the impossible.

Robert Moir
Friday, November 22, 2002

Back in engineering school I took a required course whose name I've forgotten, but which I think was something outrageously imaginative, like "Engineering I."  About 80% of the course consisted of making informed estimates that were pretty much along the lines of the gas station problem.

I hated that course.  I complained about it to my advisor, who was also the course organizer.  We're just making guesses based on too little information, I told him.  What's the point?

Well, guess what?  I found out that a good deal of the up-front work I do in any engineering project amounts to the kind of informed guessing we did in that class.  How fast will the system be?  How much hardware will it require?  How many people to staff the design effort?  Sure, you can come up with fairly precise information eventually, but at the outset you have to know whether it's going to take 10 people or 1000, ten thousand dollars or ten million, a hardware platform the size of a PC or a battleship, etc.  And if you think all of these distinctions are obvious and easy to make, you've never worked on a project with huge cost or schedule overruns, or one in which the hardware or software underperformed by a factor of 10.  Happens every day.

Some engineers are unwilling to do this kind of guessing, claiming that the answer is unknowable.  Well, a precise answer may not be available, but an approximate one often is.  And that approximate answer may be enough to tell you whether the path you're on is reasonable or hopeless.

The gas station question is NOT an "aha" question.  You don't need some magical, trick insight to get to the answer, just reasonable smarts and the willingness to make intelligent guesses in the absence of information.  A precise answer is, of course, unnecessary, but if you estimate that there are 2000 gas stations or 20 million, I'd be concerned.

And, of course, if you start to reason along the lines of, "Well, most people don't drink more than a teaspoonful of gasoline a day...," I'd refer you to the executive research committee.

Hardware Guy
Friday, November 22, 2002

Nat ol' buddy,

You'd make a fine accountant and I'd hire you in a second. But a poor engineer and a poor businessman. Successful engineers, analysts and businessmen have to make estimates every day based on limited data. And they understand that it is not always necessary to get an exact answer but just in the right range.


C is not a perfect designed language, nor is UNIX a perfect OS. Multics was designed to be a perfect OS. Shoot for perfection and you'll never get to market. "Did he say never?" Yes I did.

So how many people are going to be using this web site at its peak? 10 hits a day or 100,000 hits a day? Obviously I need the general answer and not the exact figure. Getting the exact figure wouldi be such a huge waste of time I don't even know where to begin - the exact figure would indeed be impossible, but getting a rough estimate is incredible valuable information! How could you not know this!

Sorry, Nat, but that was like the lamest post I've seen. Are you just some low level coder? Or do you even do engineering at all?

Ed the Millwright
Friday, November 22, 2002

Well, OK.  Alberto's very accurate estimation of gas stations changed my mind.  Very interesting.

Nat Ersoz
Friday, November 22, 2002

Ed, mah man...  my mouth ran away from me on that one.  Suffice it to say, yes it was lame.

Never mind...

Nat Ersoz
Friday, November 22, 2002

There's one of these questions that I've heard a couple of places, but never heard or figured out a convincing answer:

Why do bathroom sinks usually have overflow holes, but kitchen sinks do not?

Friday, November 22, 2002

K I'll take a bash

Kitchen sinks, aside from being larger (more time) are almost always in pairs.  The divider between them is not as high as the outside edge.  Both sinks are virtually never plugged at the same time as they are only distributed with one plug!?!

Brad Siemens
Friday, November 22, 2002

ALaura Roenitz, senior market analyst for Kohler, provided the following response and it probably applies to other brands of kitchen sinks.

Kohler kitchen sinks do not have overflow drains for a couple of reasons:

• The overflow is not necessary, because the saddle (the part between the sinks) usually is lower than the rim. In most cases of double basin kitchen sinks, the second drain acts as an overflow of sorts.

• An overflow drain creates an area that isn't rinsed and sanitized often and can harbor germs. In kitchens those germs can contaminate food.

I had to know! <g>

Brad Siemens
Friday, November 22, 2002

How come I haven't got a job?!!!

Friday, November 22, 2002


Brad Siemens
Friday, November 22, 2002


It's no problem - I've been listening to a lot of gangsta rap lately too which I've found to be very inspiring.

Ed the Millwright
Friday, November 22, 2002

> Why do bathroom sinks usually have overflow holes, but kitchen sinks do not?

This is one of these false premise questions like "Why are most manhole covers round?" when actually most (I estimate over 90%) at least in London, UK are actually square or rectangular.

Similarly, I've never seen a kitchen sink in the UK or the other 2 countries where I've had a kitchen (Switzerland and the Netherlands) which did not have an overflow hole.

Neil Butterworth
Friday, November 22, 2002

I follow you, but still, It's not a false premise in the US.

Translation to other cultures is easy: "Why *should* most man hole covers be circular?"

(Since they abviously should be! We Yanks did it right for once!)

Sarain H.
Friday, November 22, 2002

Translation to other cultures is easy: "Why *should* most man hole covers be circular?"

(Since they abviously should be! We Yanks did it right for once!)

Please remember that the UK is the country that is (or was) home to people that invented:

- the steam engine
- the computer (twice!)
- the jet engine
- antibiotics

and quite a lot of other useful stuff.

If such an inventive people decide that manhole covers should in general be rectangular then perhaps there is a good reason for them being thet shape. Perhaps a manhole designer might like to comment here?

Neil Butterworth
Friday, November 22, 2002

"Perhaps a manhole designer might like to comment here?"

Thank you for the introduction. As the world's most preminent manhole designer, I must say that circular is definitely the way to go. We did exhaustive testing of square, rectangular, regular polygons, irregular polygons, stellated polygons, ovalloids, trapezoids and other shapes and round was always best.

But if you like, we can have others rehash all the reasons why here. I have to take the Steam Train to Philadelphia this evening and regret I can not participate, but I wish you best of luck with your thread hijacking.

Phileas T. Farnsworth, inventor of the manhole
Friday, November 22, 2002

I asked a candidate, "How long would it take a cannon ball to drop to the deepest point in the ocean?"  He looked at me dumbfounded.  I gave hints.  He was still baffled.  I walked through the whole thing.  He said, "Oh."  And his reaction was like he was not sure I was all together upstairs.  He was a strong candidate in all other ways.  He was hired by my boss, despite my misgivings based on this one question.

I was wrong.  The guy turned out to be one of our best programmers, and could crank out code faster and better than most.

I don't think I'll ask such questions again.  They gave me the wrong opinion of that guy.

Glade Warner
Friday, November 22, 2002

"Why *should* most man hole covers be circular?"

That's too easy, the answer (in USA) is "because GW's electioneering  is funded by a company that only makes round manhole covers"

Similar situations (round/square/oblong) exist in other parts of the world.

Alberto, world famous petrol station estimator
Friday, November 22, 2002

"when actually most (I estimate over 90%) at least in London, UK are actually square or rectangular"

Yea, but we know the Brits are a load of squares anyway!

Cool dude!

Stephen Jones
Saturday, November 23, 2002

This has been a fascinating thread.  Not, of course, because I learned how many gas stations there are in the USA, but because of the various attitudes by posters concerning this question.

My own attitude agrees with that of posters such as Ed the Millwright and Hardware Guy.  This is not an impossible nor even difficult problem, although it would be bit unfair to assume everyone has the information needed even if most adult Americans drive cars.  Nevertheless, I would be concerned about the problem solving capabilities of anyone who just threw up their hands and said "I can't do that. I never took a gas station estimating class".

Sunday, November 24, 2002

I find it interesting that the only factor involved in most the calculation was cars/.

Brad Siemens
Sunday, November 24, 2002

Here is my solution to the problem:

The primary function performed at a gas station is the fueling operation (FOP).

I fill up once or twice a week.  Most adult Americans are drivers.  If my exeperience is not too far from typical then we have 2E+8 drivers times 1.5 FOP/driver/week for a total of 3E+8 FOP/week.

A FOP takes a few minutes.  Let's say 5 minutes per FOP and a station might have 6 pumps.  A single station could produce about 1700 FOP/day.  But a lot of the time pumps are idle, especially at 3AM.  At this point I take a bit of a shot in the dark and guess that a typical station performs 500 FOP/day.

Take the ratio 3E+8 FOP/week /  500 FOP/day/station gives a result of about 86000 stations.  Off by less than a factor of 2.  Not bad for a SWAG.

Sunday, November 24, 2002

6 or 42 depending on whether you believe George Carlin or Douglas Adams

Brad Siemens
Sunday, November 24, 2002

What would Richard Feynman say when asked "why are manholes round?" at an MS interview?

Absolutely hilarious!

Feynman fan
Monday, November 25, 2002

Uhh, I think the best way is to load the RV with cases of Paps Blue Ribbon and drive from Maine to Florida and around to CALIFORNIA and back around and stop at every station to pick up bags of Cheetos.  Then count the number of bags of Cheetos minus the bags u ate.

Ken Jennings
Wednesday, July 21, 2004

*  Recent Topics

*  Fog Creek Home