> You can take deliberate steps to lower the stakes in your interview process by changing the kinds of transactions that happen in the interview (for instance,

> You can take deliberate steps to lower the stakes in your interview process by changing the kinds of transactions that happen in the interview (for instance, by constructing interviews that involve collaborating on a problem instead of challenging the candidate). If you want to evaluate all candidates on a level playing field, this can be really hard to pull off. I work in the defense sector. We have some pretty strict guidelines on how we're supposed to evaluate people. Because of those guidelines, we often try to give each batch of candidates the same general set of questions/challenges, etc. We do try to be collaborative during our interviews, but it's really tough to collaborate without leading when you've worked a particular problem 1,000 times. Further, even when you do this well, the candidate often interprets your help as their failure. "Oh crap, he pointed me in a different direction, he must think I'm an idiot." > You can push more of the recruiting process into situations where the candidate controls their environment (for instance, by doing stuff at home, and on a fixed but flexible time scale) and less on in-person timeboxed interviews. I think this is really key to our success. Our interview focuses on two programming problems which require you to interpret imperfect requirements, ask questions if you deem it necessary, provide an estimate of when you think you'll complete your work, code/document/test two full applications (one per problem). We ask the candidate to do this well before the interview, on their own time. Typically we don't discount for how long it takes a candidate to complete their work. I've had awful submissions which took someone 40 hours, and excellent submissions which took someone 4 hours, and vice versa.