There was a time when I was hiring vendors for my team (at a big company) rather than FTEs and I had a bit of leeway to diverge from the company's interviewing guidance. What I would do was sit them in front of a computer with Visual Studio and internet access for an hour and ask them to do something like write a program that restarts a service given its name, and they could use whatever (e.g. google, Wikipedia, and/or stack overflow) short of copying and pasting the code to do it. I figured if they couldn't do that, then they couldn't do it in real life either, and if they could it was good enough.

I think this is fair. I prefer take-home assignments, but your idea is a refreshing take on the on-site interview.

I think I prefer the on-site version, because that way I know they aren't BSing me when they tell me "this should only take 2 hours".

Well, if I was a manager and there are about 200 developers like you with a GitHub and experience and open source contributions applying for about 3 positions, I am going to make a decision to filter candidates out.

Most managers would reach for the algorithms/hackerrank/leetcode because it's the least risk to them.

> Most managers would reach for the algorithms/hackerrank/leetcode because it's the least risk to them.

You mean it's least effort. Least risk would probably be hiring a private investigator to dig up everything about the candidate for the past 10 years.

It’s least risk in the same sense as “Nobody ever got fired for buying IBM [/Microsoft/Oracle].”

In other words, there is actually substantial long-term risk involved in simply electing for the default solution, but it feels safe on the short term and won’t make you look bad to your boss.

That carries the significant risk of getting chewed out for being completely ridiculous.

Is it? It solves the problem of somebody lying on their resume. It's probably not that much more expensive than chewing through a dozen candidates over the course of a year without finding a single suitable match.

I think that's because HR tends toward avoiding false positives rather than false negatives, as the latter tends to be more expensive. Therefore, they would rather miss an ideal candidate in order to avoid the risk of hiring a sub-standard one.

In a perfect world, HR would understand the position well enough to properly vet it but in practice they tend to revert to easier measures like the ones described.

No, they do it because most don’t know how to interview. So, they just do what everyone else is doing.

I thought software engineers with even a hint of talent were in short supply? Have I read the market totally incorrectly?

The problem is that those are not the only ones that apply.

The problem is that there is no reliable metric with which to measure talent. There is no interview process at the moment that currently does this accurately.

I'd say the problem is lacking accreditation. It's bonkers a person given a CS degree is not understood to know CS well enough to render whiteboarding unnecessary.

For a medical doctor it suffices to gets one doctorate. I don't think hospitals will drill him on the core fundamentals of his craft similar to whiteboarding in cs. A patient arrives with a slight fever and a headache. Quick! Which lab tests would you assign him!

It's ironic that for a doctor that actually would make sense since it's something practical that is needed on the job. E.g. "Patient has those symptoms, what could it be?"

How do other fields actually deal with that? Electrical Engineering should be close to our software field... Are they asked to draw common circuits or so?

> Well, if I was a manager and there are about 200 developers like you with a GitHub and experience and open source contributions applying for about 3 positions

When is the last time a hiring manager had 200 good resumes for 3 positions?????

For most companies it is the opposite experience - many open positions, few good resumes, which makes me wonder why they work so hard on keeping people out

Big "celebrity" companies like Google do get flooded by huge numbers of resumes, which is one reason they have the hiring process they have.

I have a mixed view on this. I don't have a traditional CS background, so when I started focusing on datastructures and algorithms a couple years ago, I found that it helped my programming. Recently, I was in a position where I had a chance to create an interview process for a smaller firm. I included some algorithm questions, but it was roughly 25% of what mattered. We looked more at communication skills and past work. Judging someone on a random algorithm question seems a bit too harsh for me. Even the best developers probably have some holes in their algorithm knowledge and would miss a few leetcode "easy" questions!

And the hiring managers buy that reasoning?

It's a decent reasoning and would only be able to get away with it if you have some decent years of experience OR some very good open source contributions. Or Both.

Some do, some don't. Some women don't want to date me, either, but I don't stay up worrying about it :P

I refuse to give these kinds of interviews. They don't really tell me much other than they know algorithms and data structures. And with thousands of articles and even whole books dedicated to passing those sorts of interviews, that's not saying much at all. What I really want to know is:

- can the candidate get the work we need to do done?

- what do they prioritise when doing the work? What do they value?

- can they do it well? In reasonable time?

- how well do they fit in with my team? Are they somebody my people want to work with? (Probably one of the most important things)

- how much will they grow? Are they open to feedback? Are they focused on getting better?

A whiteboarding session doesn't really tell me much here. But if all the answers to the above are positive, I'm sure the candidate has the flexibility to find and learn what they need to solve a problem.

But this is all negative. What do you do in an interview to predict how effective they'll really be?

I don't think anyone is especially fond of judging candidates by using riddles, trivia questions, and unrealistic tasks, but the constraints of the interview process rule out a lengthy and thorough assessment.

There are hundreds of companies that don't use algorithmic interviews, they are neither necessary nor particularly useful.

http://they.whiteboarded.me/companies-that-dont-whiteboard.h...

What's the positive answer to "What do they value?" for you?

Join the resistance, refuse these pointless interviews!!

http://they.whiteboarded.me/companies-that-dont-whiteboard.h...

I don't mind those questions in an interview as long as the recruiter is not just reading them from a paper sheet, and actually know this stuff. This can be a way to find a candidate with a strong theorical CS background. Which can be relevant for some positions.

Now, I've seen my fair share of recruiters that did not really understand the topic at hand when asking theorical CS questions, let alone the desired answer. I do so always get really suspicious when they occur.

> It's kind of how in dating people have deal-breakers (e.g. no smokers, no kids, and so on) -- when looking for a job, this is one of mine.

Not everybody can afford to be as picky as you regarding job choice (this is not meant derogatorily).

Are you able to get good jobs that way? Not criticizing, just genuinely curious.

Yep! I think the biggest boon in job searches (for me) has been Stack Overflow.

Thanks for sharing your experience