# number of giraffes in the US: on an IQ test?

I recall being told once that there is an IQ test, or maybe just reasoning test, that asks the subject to reason out loud about how many giraffes are in the US. The point is to see if they can…well I’m not sure what the point is, but presumably they would reason like this:

1. There are 50 states in the US.
2. Each state averages one or two large metro areas that might have amenities like zoo parks.
3. Each major city averages one zoo.
4. Each zoo averages 2 to 4 giraffes since they are kind of popular things to see.
5. Ergo, there are maybe 50 to 100 major zoos in the U.S. and about 100 to 400 giraffes. Hold a gun (or IQ measure instrument) to my head and I will guess 200 giraffes in these here United States.

Anyway, here’s my question: is it true there is a test or tool that uses this kind of question, if not this question? What is it called?

While we’re at it, how many giraffes are there in the US, in case I need to crib for this exam someday.

These type of questions are called Fermi problems, where one is called upon to perform heuristics by employing general factual or implicit knowledge, to guess at an answer.

That type of question is quite common in job interviews. For one example: http://joelonsoftware.com/articles/fog0000000073.html (search on “impossible question”).

I don’t think you’ll find it on an actual IQ test, however, because it is too subjective. How close the test subject comes to the correct answer (if that can even be accurately determined) is not as important as how they go about making their educated guess. You need to be there while they reason it out; just looking at the answer will tell you (almost) nothing. So it’s mostly an informal, unscientific (but possibly quite accurate) way to get a feel of how someone will react in a situation where they have to make a best guess based on incomplete information.

I don’t know that I’d put this on an IQ test. If you’ve never been exposed to this sort of thing, you’ll be hopelessly floundering, no matter how much potential knowledge you’ve got. Whereas, once you’ve seen it done once, you’ve got a handle on how to solve it. I’d put it on a mathematical aptituide test, though.

One of my ex-bosses used to do things like this on lectures about education (":How many gas stations are there in the US?"), and I think John Allen Paulos talks about it in his book Innumeracy.
as for how many giraffes really are in the US, you’re on your own. Start doing a survey on the internet, or contact the zookeeper’s association. Don’t forget about private zoos, like York’s Wild Kingdom in Maine, which has giraffes but isn’t a publicly-supported institution.

Well, that’s true for a lot of IQ test questions anyway. Practicing a couple of tests beforehand, to learn the standard patterns, definitely helps. Especially since, with a lot of such tests, most of the questions are not really difficult but there’s a very strict time limit, so that you won’t get a good score unless you can answer most of them almost without thinking consciously.

I think it’s a good test to see how comfortable someone is at pulling numbers out of their ass. Personally, I’d hire the ones who say ‘I don’t know’, but maybe they are looking for someone who can shoot the BS with a straight face.

For example, your logic falls apart at step two, since you just pull that information out of thin air, and build on it, and each successive step does the same thing.

Fortunately, I have written a computer program to solve this. The correct answer is 1,497 giraffes. That must be true, since it came from a computer.

I think it’s a reasonable test of one’s ability to extrapolate an educated approximation in the absence of clear data. If someone wants to know if it’s more likely 400 or 4000, then the method works well. Sometimes to answer the question with a reasonable deduction is better than expending the effort to collect data.

As a side note, I think giraffes are cool. I’d like to have a zoo with nothing but giraffes- I’d call it Giraffic Park.

I had an engineering professor do something like this to illustrate a point. One was “How may piano tuners are there in San Diego?”

X people, Y% of people have pianos, Z% of people with pianos get them tuned in a given year, W piano tuners needed to cover this amount of work or something.

The answers always varied wildly but the average of the guesses of the class was nearly always very close to the real answer.

Yeah, but what happens when the Giraffes get lose and eat the zookeepers?

What everyone else said.

I don’t really like these things, but it’s a fair thing to test that people can do, when you look more at the process than the answer. Everyone should be able to come up with at least a rough guess.

However, if I wanted to be anal, rather than looking it up, or saying I don’t know, I’d give an error estimation: we could be a factor of two out on lots of steps, which can easily lead to couple of orders of magnitude on the whole thing, which is fine as long as everyone KNOWS this.

I seem to recall reading that a giraffe can decapitate a man with a kick, so the real question is what happens when they learn to play soccer with * Human Heads * ???

We promote them to moderator.

Heck, it’s a lot easier than that to figure out how many giraffes there are in the country. Just count all the giraffe legs and divide by four.

Yes, but you have to divide by a very tall four, or you’ll come up short.

Douglas Hofstadter also gave a rather in-depth treatment of this type of question in one of his books, and he made the point that in practice, the errors tend to be self-correcting: if you guess too low in one step, you are likely to guess too high on another one – the chance that you’ll err in the same direction each time is not very great. Which is why a smart person, with a decent amount of common sense, basic mental calculation abilities and a reasonable amount of real-world knowledge (e.g. having some idea of the number of zoos per state and the number of giraffes in a typical zoo) will usually stay within an order of magnitude of the correct answer.

Of course, I wouldn’t want someone to apply this kind of seat-of-the-pants reasoning to designing a nuclear power plant. But if someone came up with an answer of 4 or 40000 to the OP’s question, or claimed that the answer was impossible to estimate, I’d have my doubts about their ability to handle real-world issues in which the correct answer cannot always be easily determined.

This is obviously his variation on the original Fermi problem, posed by Fermi himself. He was at the University of Chicago at the time, so the city was Chicago, but the rest is the same. The problem has migrated into the lore of the recruiting biz in high tech, particularly since nobody understands why someone would ask the question?