I've long felt the key is to get the interviewee to do some real work for a day. Pair programming seems like it might work well. You'd have to come up with som

I've long felt the key is to get the interviewee to do some real work for a day. Pair programming seems like it might work well. You'd have to come up with some suitable toy problems that you could make significant progress on in a day.