I'm torn on homework assignments. On one hand, Gayle's point is a good one: they're asking for a big time commitment from candidates. On the other hand, they

I'm torn on homework assignments. On one hand, Gayle's point is a good one: they're asking for a big time commitment from candidates. On the other hand, they seem to be the challenge that maps most directly onto actual work performance. Our process here is: 1. Non-tech phone screen 2. Tech phone screen 3. Homework 4. Onsite The onsite is, to many candidates, surprisingly light. This is because the homework makes up most of how we're evaluating your tech skills. The onsite is just to make sure that you wrote the code you submitted, and then to test your non-coding skills (architecture, communication, working with non-engineers, etc). But we do lose some number of candidates who just don't get around to doing the homework (or possibly, as you do, refuse to do it). This worries me, since I suspect the people least willing to do homework will be the most senior and/or those with the most competing offers. We've recently started experimenting with allowing candidates to chose between homework and a code sample. The idea is that a sufficiently complex and interesting coding sample should tell us about as much as coding homework. It involves more time on our part to evaluate it, but less for the candidate to prepare it. The early results have been mixed (a lot of unsuitable coding samples), but we're iterating on how we ask for coding samples to try to get better quality ones.