[tchalla]

> My least favourite interview technique has to be "talk me through a complex problem on a whiteboard".

Let's say an interviewer reads your resume and is interested in a project. The interviewer comes to the interview prepared with some questions. They ask you to start with a high level summary of the project. Then, ask you - "What choices did you have to make during the project?". Then, ask you some specific questions about these choices. In each of these questions, the interviewer is not supposed to "evaluate your answer" but rather gain an understanding of your project and choices to be made. At the end of the interview, the interviewer would walk away with two things - (1) a clear picture of the project almost as good as the interviewee and (2) a clear picture of the design decisions the interviewee made (options they had and why they chose one over the other).

The above interview is a conversation and you are allowed to be silent. The process will be communicated to you upfront. In fact, silence is encouraged because the interviewer would much rather give you time to think over a spur of the moment answer.

[majormajor]

I would love, as an interviewer, to be able to do this style of interview every time.

It's astonishing how often you get nothing useful as an answer. Things like "my boss suggested it" as the reason for choices being made, or "not really" when asked about alternatives considered, or problems they ran into...

The hypothetical project whiteboard interview gives people who haven't done sufficiently interesting past projects and opportunity to show that they can think about them, at least...

[shepardrtc]

> Let's say an interviewer reads your resume and is interested in a project.

That's not what they were talking about. The complex problem is supposed to be novel to the interviewee.

> In fact, silence is encouraged because the interviewer would much rather give you time to think over a spur of the moment answer.

Yes, they certainly say that. But for a normal person trying to figure out a novel problem, ignoring everything that's going on and sitting there silent for 30 seconds or longer while the person who single-handedly can determine whether you pass or fail - often depending on their mood - stares at you is incredibly difficult. And if their mood isn't good, they absolutely see the silence as not knowing the answer and push for you to move on. I had it happen to me on a Facebook interview. I personally can't concentrate enough to think through a difficult, multi-layered problem in that situation. It is in no way reflective of what I can do while working.

Online coding tests are more than sufficient. And a discussion afterwards about said tests is sufficient to make sure there was no cheating.

[omginternets]

This is of course a great way to interview, when you're only concerned about assessing competence. It seems to me, however, that assessing competence is one of several concerns that the interviewer is expected to balance.

In particular, HR is expected to structure the hiring process in such a way as to avoid accusations of bias. In order to achieve this, they are expected to come up with precise criteria for hiring, which is at odds with anything vaguely qualitative or holistic, let alone ecological.

The result is that we strictly rank-order candidates using hard measures. In doing so we sacrifice the ecological validity of the test.

[louisstow]

Those are great, I'm more talking about a complex problem you're given and have to whiteboard the solution while making sure not to be silent for too long or you'll interrupted or given a hint.

[tchalla]

I was offering an alternative to the novel problem on the spot. I’m glad you like this procedure better : )