Interesting paper about testing non-programmers for programming ability. How do you judge if someone would be good at programming before they try?