> And my personal pet peeve, asking the candidate to solve a problem that you just solved a couple of days ago (perhaps after several days of collaboration and research).
I think these types of questions lead to some of the best discussions between interviewer and interviewee. The interviewee should realize that they aren't expected to just pull a perfect solution right out of thin air -- instead the point is to have a dialogue with the interviewer so that the interviewer can see how you think and communicate. And having a couple of big hard problems that the interviewee doesn't have enough information for an immediate solution is a good approach: you can see what kinds of questions the interviewee asks, how he tries to attack it, and the interviewer can provide information and nudges along the way, having all the background information fresh in his head.
In my reading of interview feedback, too many interviewers don't understand that the candidate shouldn't be expected to pull a perfrect solution right out of the air.
Often too much criticism is feedback is placed on the fact that the candidate went through three or so iterations and too much emphasis is placed upon the fact that the candidate's initial solution was quadratic or worse. This really should reflect strong analytical thinking.
I suppose that could happen in the best circumstance.
But in my experience, there are two things wrong with the "here's a problem I solved last week" question:
- the scenario is way too specific. Say you have some unusually pathological database that was developed in house, and you need to query it using a proprietary wire format that only your company uses. And you discover some kind of trick that sort of works. This is a TERRIBLE question to ask a candidate, but I've seen it done. The candidate is fighting in the dark, trying to grasp the outlines of the problem, for the full ten minutes.
- asking about stuff you did last week leads to the interviewer comparing themselves with the interviewee. A fair-minded person can see there might be other approaches. Most people are proud of their own solutions though, and won't accept alternative answers. It's better, psychologically, to use a problem you aren't personally invested in.
I think these types of questions lead to some of the best discussions between interviewer and interviewee. The interviewee should realize that they aren't expected to just pull a perfect solution right out of thin air -- instead the point is to have a dialogue with the interviewer so that the interviewer can see how you think and communicate. And having a couple of big hard problems that the interviewee doesn't have enough information for an immediate solution is a good approach: you can see what kinds of questions the interviewee asks, how he tries to attack it, and the interviewer can provide information and nudges along the way, having all the background information fresh in his head.