[rwmj]

I'm not going to defend this style of interviewing, but I think it's useful to have tests which filter out people who cannot program at all. If you do any hiring you'll know that a surprising proportion of interviewees who make it to an on-site interview for software positions simply know nothing about the subject. When I did interviews at my start-up we'd have a Linux machine connected to a projector for interviewees to use (this is for Linux programmer and sysadmin positions), and a small but significant proportion of the candidates had no idea what to type into the terminal to list files, or what an absolutely trivial program (like loop to 10 printing numbers) did.

[cdent]

Sure, can't dispute what you say. There has to be some amount of "you must be this high to ride this ride" but you don't need to measure 7 times or whatever before thinking the first measure was accurate.

I've been on both sides of the interviewing tables many times and the processes that worked the best were the ones that involved telling stories to another another and seeing how the other reacted. You know, actually interviewing, humans having a conversation.