Comment by blackflame7000 | Hacker News Reader

Comment by blackflame7000 | Hacker News Reader

Parent Full thread

I totally understand your philosophy and you make some great suggestions, but overall there's one big flaw with whiteboard interviews: Artificial Pressure and its impact on people's ability to think clearly on the spot. I am a very methodical thinker which is why I tend to prefer work in large scale ore high-performance systems where you consider the implications of thread contention and so forth. I've designed and built multiple large scale production systems and i've also simultaneously forgotten whether it should be += or =+ during a whiteboard interview because of a phenomenon I can't quite describe. Obviously I was laughed at for someone who's actually had 10 years experience with c++ to forget something so basic but it happens when you are on the spot.

The biggest problem with whiteboards is that they reward exactly what they say they try to weed out, rote memorization. It's no secret you could spend weeks practicing on leetcode for an interview but those skills simply don't translate to real-world programming. Algorithm design is maybe 2% of software engineering. The vast majority is knowing what pieces of technology are out there in order to avoid reinventing the wheel all the time. Whiteboards don't test that skill. It tests a marathon runner on their 100 meter dash time.

> Obviously I was laughed at for someone who's actually had 10 years experience with c++ to forget something so basic but it happens when you are on the spot.

This 100% happened to me (apart from being laughed at) only it was even more of a meltdown. Also 10 years experience and a problem I could do in my sleep. It was my first interview after a long time and it triggered performance anxiety and nervousness.

The solution is you need to practice live interviews, not just leetcode solo.

If the system is so poorly evaluating people that they have to train an irrelevant set of skills to succeed in it, what does that say about the system?

I am a big fan of the take-home problem. For example, Symantec once asked applicants to design a simple virus detector with wildcards (similar to grep) and then they gave interview spots to people with the best performance. During the actual interview, however, there wasn't much whiteboard coding only designing or "how would you do X" sort of questions.

There's a small list of companies I'd do a, "Take home test," for. In my experience, most companies that ask for it are looking for free labor.

I've never done a take home test that produced anything that could have been vaguely useful for the interviewing company. Usually the problems are pretty artificial.

They absolutely should be artificial problems. If they're not carefully designed to test what the hiring company cares about, it's a bad test

Well if they are good enough for you to put their code into your program then shouldn't they be good enough to be offered a job? Seems like it works exactly as intended in terms of trying to find someone who can do the job

I've had multiple take home tests result in changes to the products that company offers. I've never received a job offer from those companies. Maybe your experience differs, but I've had a sour taste left by such practice.

I wouldn't say coding algorithms is "irrelevant". It's sort of like saying the SAT is irrelevant. It's a general aptitude test. It's not about the "relevancy" of trigonometry to your day to day work.

You could argue it's stupid and colleges should only look at personal essays and GPAs and extracurriculars. But then there's problems with that too (lack of standardization for one).

You could argue that high schools shouldn't teach trig or calc because they're largely irrelevant to most tasks in most fields and "what does that say about the system". But actually they do come into play once and awhile and lay a conceptual foundation, so they're still sort of useful to learn. And thus they form a good common foundation to test aptitude inside an hour or so, or at least no one's thought of a better one.

I knew when the interviewer pulled out his phone to snap a picture of only that part of the whiteboard that I didn’t get the job lol

Whiteboard interviews exist because without them companies wouldn't have a metric to artificially restrict hiring. Passing a whiteboard interview makes both the company and interviewee feel special and maintains the bubble of software engineering salaries.

Companies don't care about artificially restricting hiring: they want developers to be as cheap as possible.

The truth is, software engineering really quite hard, but that whiteboard interviews just happen to be a rather poor way of evaluating candidates.

A work sample test, structured interview, or IQ test (alone or in combination) would be a much better filtering process.

> A work sample test, structured interview, or IQ test (alone or in combination) would be a much better filtering process.

I'm not sure what any of these mean in concrete terms. Is an IQ test like those "why are manhole covers round" Microsoft questions of yore? Yuck. Is a work sample a homework test? That takes more time for both parties. It's not a good substitute for a 1 hour first round screen. And what's a "structured" interview??

You should probably look up what an IQ test involves instead of just guessingly blindly, getting it wrong and then writing off the suggestion based on a wrong assumption.

Is stratifying income based off of IQ any more than we already have something that would benefit society?

in sofar as IQ helps predict actual job performance, I'd say yes.

Now I'm not saying the situation isn't complicated. In the mid-20th century, IQ testing was a great driver of social mobility. However, as was pointed out in _The Bell Curve_, it's becoming the opposite. From what I understand, the heritability of IQ and assortative mating (i.e. people marrying other people of similar IQ/educational attainment) are the main drivers of this.

I don't think the formation of an entrenched class system is necessarily a good thing, but I think in the local context of our industry there's still a lot of good that could be done by improving our hiring process. I mean, would you rather hire people who only sound smart, or who are smart?

You have questions, and I have answers!

https://www.wired.com/2015/04/hire-like-google/

If you have more questions after reading that, then ask away.

Reply on news.ycombinator.com