Hacker News Reader

Comment by dyingkneepad | Hacker News Reader

It can work in favor of but also against you :)

Where I work, once most of the candidates have been filtered out and there are only 1-5 remaining, managers get help from the senior engineers to assess the candidate's technical skills before interviewing. In my specific team we try to find code samples from github and contributions to open source communities (we love ohloh.net).

But we also do a "technical interview" with some fizzbuzz stuff. More than once we found candidates with good-looking open source patches but that couldn't implement fizzbuzz-like stuff on a whiteboard.

View on HN

> But we also do a "technical interview" with some fizzbuzz stuff

Why does anyone still think this is a good interview technique that provides any valuable metrics?

Because in the past we hired some programmers that were really good at BS-ing during interview and really crappy at programming.

So now we make sure we hire people who know what "a &= b" does in C and what a #define statement achieves.

I'm not saying you don't screen them at all with a technical task, but it should be something representative of what your organisation does, rather than some overdone toy problem

When I said 'fizzbuzz-like' I meant a simple program that is representative of what we do in the organization. Fizzbuzz is supposed to be a simple challenge as far as I'm aware. Stuff like flipping bits with & and |, or knowing how to use preprocessor macros (since we abuse macros a lot). Or awareness of function pointers.

I guess I missed that fizzbuzz now means things like fizzbuzz, rather than actual challenge

Because there are still people who can look good on paper, and even talk their way through interviews, but can't actually code. You do not need the multiple-hour-whiteboard overkill that some places do, but you do need at least one coding gate to prove their resume is not a sham.

I look at FizzBuzz as a "assert" test before the thorough evaluation.

It should be logically impossible for an experienced developer to not pass a simple test like that. Those tests are not for metrics- they're just for prematurely weeding out corner cases.

Agreed!

Whiteboarding is a skill and should be practiced prior to onsite interviews.

Reply on news.ycombinator.com