Hi. I did an interview recently and failed. I could not answer this question properly.

Estimate the number of doctors needed in a town of 4million people.

I think it's more of a critical thinking problem. I dont know.

Whow knows how to explain this?

Yeah they didn't want the WHO recommended number. They needed me to walk them through how I would estimate how many doctors are needed.

Ask them what's the budget then they will know how many they can get 😂

Subjective answer.
Red flag.
Leave.

It's a Fermi estimation, don't let it bother you. I think they're a waste of time but some companies do them because reasons.

I’m a little wasted but at my dumbest I still see holes! Number of doctors needed in a town of four million people..they could be young and healthy or old and frail on average. Anyway my pathetic inability to produce a solution acceptable for the standards of the intellectual overlords who posed this question makes me consider other roles suitable for my pleb brain. They telepathically suggested a junior toilet cleaning role

Do you guys know some cool companies that need a remote dev?

I'm fullstack. React js, React native, node js, java and native Android.

I would pass a codility test if need be.
But this doctors question is just insane.

One place I interviewed asked if I would jump off a building if my manager told me too. I told them no, but I wish I had asked: Is there a risk analysis and a safety plan? Or maybe: Is this considered overtime and is there hazard pay?

Are the 4 million people in the town dead or alive?

All of these questions require the same kind of analysis strategy to come up with rough estimations.

Like, assuming everyone goes to the doctor let's say once per month, and a treatment on average takes let's say 30 minutes (mostly faster, but also some lengthy surgeries), and a doctor works 8 hours per day.

4 million people times 0.5 hours is 2 million doctor hours per month, or dividing by 30, 67k doctor hours per day. Divided by 8 hours per day is 8333.

But since it's only a rough estimation of the order of magnitude, we round that up to 10k because doctors also need sick leave and holidays.

What matters is that you can show basic analytic thinking. How's that relevant?

Well you should be able to tell whether you can host your blog on a PC, or whether you need to rent half of Amazon.

Wow. I wish I had seen this prior.

@Root Yeah... I once had as imilar question and when I answered with an integer, because you know, half a person doesn't work that much, I got wrong.

You could also add in that in a real case, you would refine your model and your numbers through domain related research.

For example, the model above assumes that the visits can be scheduled. But you also have emergencies and accidents that require immediate treatment. So you would look how often that happens, then make some statistical model and decide how much safety buffer against overloaded EMRs you'd want to have.

But that would take more time than available during an interview, especially because you also have some questions left that you'd like to ask them.

This is so subjective. The term doctor can mean a lot. You may need a lot of GP's for 4 milion peeps, bur then what about special medical employees like surgeons. Also, the age group of the population matters. I honestly hope it's a trick question because it makes no sense to me.

@Tonnoman0909 It's not about delivering a correct and precise answer. It's about how you approach problems to get a first impression what order of magnitude this might be about - and that's fundamental engineering logic.

It's like, are we talking grams, kilograms, or tons here? Would I need a lab scale, a person scale, or a truck scale as equipment?

@Fast-Nop is right.
I think you could even answer the question in a 'parametric' form.
Because you don’t know the values of the parameters but the formula is important.

Something like:
Population divided by number of visits per year, divided by average visit time in hours, times working hours per year, and so on...
(Not the correct formula here)

Then you can make an assumption for the parameter values and calculate an estimation result value.

You don't.

You say: "I don't think this will be a fit if this answering this kind of bogus questions is a requirement. I'd rather answers question related to my profession."

@Lensflare That's also helpful for plausibility tests, which in turn are useful for error detection.

For example, there may be a formula involved, and there's a division by zero - only that because of rounding errors, it's not really zero so that the program doesn't crash, but gives an implausible result.

An engineer is expected to raise eyebrows here and investigate what's going on instead of taking every result at face value.

@k0pernikus If a job applicant thinks that back-of-the-envelope estimations are unrelated to a tech job, I'd simply not hire him.

How many people in a town of 4 million eat capn crunch

@Nanos probably the reason why they want to hire a developer instead 😁

@Fast-Nop gauging an applicant's thought process is essential. it would just have been easier to understand if the question is actually relevant to the job. Like would he likely have to design a chat bot for the medical industry which would at least help the applicant get a starting point of questions to ask also. I'd take the interviewer's question as a requirement and that means exploring with inquiries of my own to fully understand what they really want to achieve. In the end though, budget is always the typical constraint.

@JustThat exactly this. It's my main issue with these kind of questions. At most I'd just keep asking them a million things about it before even giving them an answer.

@Nanos probably a design improvement which they could consider. I had an experience with a bot that didn't even answer the expected results from their job aid. Since there was a report button, I reported it broke. After a day someone actually replied and explained the issue and they even knew the things I typed and what broke it so it would be possible to submit up. At least the human asking the same set of questions again can be taken maybe as being comprehensive and can be a good thing unless one is busy. There are people who try to bypass the bots and just go straight to talking to a person. Talking to a human is still so much better.

@Fast-Nop The chances that an applicant will answer these questions because they happen to google bullshit interview questions the day before is higher than the chances they will be able to answer them in a reasonable way.

Worse they make people that would be an absolute fit for the position fail that interview which completely moves the goalpost.

Questions like these (and whiteboard coding) are the very reason we have this anomaly it tech that passing an interview is often *harder* than actually doing the daily job.

OP case is a perfect example for this insanity.

As it turns out, I won't ever need to get an estimate for the doctors in a town of 4 million people. And for any other estimation, I rather do a data-driven analysis before.

Back of the envelope may be an helpful tool, but too often it's an excuse for management to pressure technical employees into "just give me a number, it doesn't have to be exact" only for them to wonder: "Why the heck did you fail your estimation?"

@k0pernikus I don't see what was "hard" about doing some sort of estimation - see my previous comment where I wrote a possible sample answer. Even without pocket calculator app, you'd just round much more generously, or drop the number calculation entirely and just outline the logic.

If an applicant cannot come up with any sort of useful answer, it's not because it's "hard", it's because the applicant doesn't have that kind of thinking skill.

If the company values people who can do rough plausibility checks on their data instead of blindly believing anything a computer spits out, then candidates who don't even have any basic idea how do that are a bad fit for that position.

That's different from complicated algo whiteboard crap because the point of estimating orders of magnitude is that you neither are expected to give any precise result nor to do big research.

@Fast-Nop People in this very thread asked if it was a trick question. That fact alone should sound an alarm that it is in fact a silly question.

It's not necessarily hard but I rather find them pointless even with your framing. It's more like a big lipped alligator moment if you aren't expecting it. (And if you are expecting it, what are you checking? People then just game the system.)

My main point is that an interview is already a stressful situation to be in in the first place. Yet there's this idea of making people jump through hoops in there.

I'm sure there's a reasonable way to come up with answers to questions like: How many balloons fit in this room? What is the storage space required to host all images on Google Street View? How many golf balls can you fit into a Boeing 777?

I truly can't be bothered. I'd leave.

Even if your developer position truly (I doubt it) requires that kind of estimation approach, the interview is not the right place to test for it. Find another.

@k0pernikus That is already the "easy" sort of question. If the applicant fails at that, he's just not fit to pass any job interview. Or probably any sort of talk (the days of the lone hacker in the basement are over and won't come again).

Because the grill questions come afterwards. Not useless "trick" questions, but tough ones where you will get at least on the right track if you have worked with that before, but will fail if you havn't.

@Fast-Nop I don't understand why there has to be "grill questions" in a god damn job interview for a coding gig.

It's all about: Is the person a good fit for the company and vice vera. (They still can fail a job interview but that should never depend on such a question.)

Your comments, especially your last one, lack empathy.

I've had my fair share of job interviews on both sides of the table. Never had I ever to answer one of these nor did I ask them myself.

@k0pernikus The grill questions are necessary to tell the wheat from the chaff. We're not talking friendship here, but business.

Here's a classic, nice set for embedded: https://rmbconsulting.us/publicatio...

Of course one can google that beforehand, but an applicant who is aware enough to even prepare for that is likely also competent.

@Fast-Nop It would be funny if it wasn't sad. The author admits looking the stuff up while writing it but shifts the goalpost for the candidate. I just can't take it seriously and having people like this as gatekeeper for developer jobs is part of the problem.

@k0pernikus You overread the keyword "most". I'd fail on the last sub-question, and on a bad day, maybe also on the two before, but that still leaves 5/8 that I'd get right no matter what. It's also one question among a lot of questions, and the overall picture will become clear.

@Fast-Nop I got back to the recruiter with your answer and they said it's a very strong answer. Its what they expected.

Unfortunately they still won't proceed with me. I guess I should have asked the question earlier. :-(

@kivisi Now that you know how to tackle this type of question, the next ones will be easy. Even if you don't really know, you can just replace that by assumptions while explaining that you would of course validate them if you had more time.

@fast-nop thank you so much. I've learnt something.

in the meantime, do you have links to places I can look for work?

stack -> java, kotlin, react js

@kivisi Unfortunately not, even if I knew which country because this tech stack isn't my domain. My wild guess would be looking for something with finance and banks (Java, backend).

Thanks