4 ms·

I've asked out-of-town job candidates "How many gas stations do you think are in this town?" but I've not had much success with that. I think for engineers it's

I've asked out-of-town job candidates "How many gas stations do you think are in this town?" but I've not had much success with that. I think for engineers it's vital to be able to do order-of-magnitude estimations and I'm trying to get my kids to learn that but it doesn't seem to come naturally to people.

tinco 14y ago
I think it has to do with how you phrase the question. People go to job interviews knowing that they're going to be tested for their knowledge, and phrasing the question like that triggers a "how am I supposed to know that?" response. But you're not actually asking them to know it, you're asking them to reason about it. So perhaps instead of asking "How many gas stations do you think are in this town?" try asking "Could you give me a reasonable estimate of the amount of gas stations in this town?". I think a lot of people just lock up when they're asked something they don't know in a synthetic situation even when in a real situation they would actually make very good judgements since in any real situation the context is often so much more clear, unless you like to keep your employees in the dark..
- Terretta 14y ago
  > I think a lot of people just lock up when they're asked something they don't know in a synthetic situation even when in a real situation they would actually make very good judgements since in any real situation the context is often so much more clear, unless you like to keep your employees in the dark. I've had that mental block happen, not when asked to estimate something like balloons in a gym, but when asked to estimate something I knew for a fact was a single search query away. I was asked, "So how many travelers or trips in the US each year?" Googling "how many travelers in us" gives http://www.ustravel.org/news/press-kit/travel-facts-and-statistics http://www.ustravel.org/news/press-kit/travel-facts-and-stat... as the top result. Knowing this is that kind of data, I was unable to put myself in a frame of mind to make believe reasoning through it. The interviewer told me to just pretend we were white boarding a marketing plan for a travel tool. I pointed out that in such white boarding or brainstorming conversations, someone would look that data point up because it's obvious a lookup would be more quickly available and more accurate than running through a Fermi style reasoning chain. Partly because the best developers are "lazy", morally opposed to reinvention of wheels, your point about performance in a synthetic situation is dead on.
Joe-Z 14y ago
How often do you really need these kinds of caluclations? And if you use them wouldn't it be faster / more precise in most cases to just google it? Honest question. The problem I see is that for most of these you at least need SOME information (e.g. the number of inhabitants of NYC) to start from. To get that initial information you'll most likely use a search engine anyway, so why not just invest a little more effort to find a exact number?
- DanBC 14y ago
  These questions are never about the answer, but about how someone would go about getting the answer. Some people freeze. Some of those people stay frozen even after a bit of discussion. Some people just cannot start the process of answering a question like this. Luckily, most places have stopped asking questions like this because they'r ejust not very good for whatever those people are recruiting for.
leoedin 14y ago
I think the problem with a question like "How many gas stations do you think are in this town" is it is too simple to reach the answer, and the approximations are too vague. The solution for everyone is to google it. The approximation method requires some big assumptions about how many gas stations there are per-capita, or per-car, or per-area. I don't think anyone would be able to easily approach that off the top of their head and expect to get a reasonable answer. Just as approximation is a useful tool, knowing when it doesn't work is also useful. What would your approach be to work out how many gas stations are in your town? Have you done that calculation, and come out with a reasonable answer? What assumptions did you have to make?

tinco 14y ago

I think it has to do with how you phrase the question. People go to job interviews knowing that they're going to be tested for their knowledge, and phrasing the question like that triggers a "how am I supposed to know that?" response. But you're not actually asking them to know it, you're asking them to reason about it. So perhaps instead of asking "How many gas stations do you think are in this town?" try asking "Could you give me a reasonable estimate of the amount of gas stations in this town?". I think a lot of people just lock up when they're asked something they don't know in a synthetic situation even when in a real situation they would actually make very good judgements since in any real situation the context is often so much more clear, unless you like to keep your employees in the dark..

Terretta 14y ago
> I think a lot of people just lock up when they're asked something they don't know in a synthetic situation even when in a real situation they would actually make very good judgements since in any real situation the context is often so much more clear, unless you like to keep your employees in the dark. I've had that mental block happen, not when asked to estimate something like balloons in a gym, but when asked to estimate something I knew for a fact was a single search query away. I was asked, "So how many travelers or trips in the US each year?" Googling "how many travelers in us" gives http://www.ustravel.org/news/press-kit/travel-facts-and-statistics http://www.ustravel.org/news/press-kit/travel-facts-and-stat... as the top result. Knowing this is that kind of data, I was unable to put myself in a frame of mind to make believe reasoning through it. The interviewer told me to just pretend we were white boarding a marketing plan for a travel tool. I pointed out that in such white boarding or brainstorming conversations, someone would look that data point up because it's obvious a lookup would be more quickly available and more accurate than running through a Fermi style reasoning chain. Partly because the best developers are "lazy", morally opposed to reinvention of wheels, your point about performance in a synthetic situation is dead on.

Joe-Z 14y ago

How often do you really need these kinds of caluclations? And if you use them wouldn't it be faster / more precise in most cases to just google it? Honest question. The problem I see is that for most of these you at least need SOME information (e.g. the number of inhabitants of NYC) to start from. To get that initial information you'll most likely use a search engine anyway, so why not just invest a little more effort to find a exact number?

DanBC 14y ago
These questions are never about the answer, but about how someone would go about getting the answer. Some people freeze. Some of those people stay frozen even after a bit of discussion. Some people just cannot start the process of answering a question like this. Luckily, most places have stopped asking questions like this because they'r ejust not very good for whatever those people are recruiting for.

leoedin 14y ago

I think the problem with a question like "How many gas stations do you think are in this town" is it is too simple to reach the answer, and the approximations are too vague. The solution for everyone is to google it. The approximation method requires some big assumptions about how many gas stations there are per-capita, or per-car, or per-area. I don't think anyone would be able to easily approach that off the top of their head and expect to get a reasonable answer. Just as approximation is a useful tool, knowing when it doesn't work is also useful. What would your approach be to work out how many gas stations are in your town? Have you done that calculation, and come out with a reasonable answer? What assumptions did you have to make?