Some people think a technical interview is for asking tricky pseudo mathematical questions ("can you fit your hand under a string tied tight around the circumference of the world if I increased it by 5 inches?") or making a person draw UML on the whiteboard ("diagram how you'd make a class model for a '3d rectangle'").

It's not. It's not for softballs either, but my preferred method is to see their level of professional committment to and excitement about their craft. I can figure out how you think just by talking to you about something you like. Ref. Malcom Gladwell's Blink.

I ran across this the other day and one other data point justifies my interview style as fact. From notgartner blog:

As with all discussion with technical interviewing it seems to come back to build lists of questions, but answering questions is not what a technical interview is about. I believe it is all about:
  • Finding out if the candidate is passionate about technology.
  • Finding out the depth of the candidate by asking questions.

For what it's worth, it's probably best to have an actual introverted ex-math team nerd (fie, scratch that, since that was me) in the room asking those other types of questions, since a 'get up on the board and write' scenario forces the candidate to expose their presentation and interaction skills. Yes, I know, I just undermined the 'technical' part by saying it was good for psychological assessments. Again.

Sue me.