There were a couple of things I learned from these pages:
Ask your candidates some generic but vital software engineering questions, and don't bother with specific technologies. Because technologies come and go, but it takes good software engineering skills to cope with ever-changing technological environments.
Ask open questions, so that you can determine a person's professional views and talent from the discussion that might follow.
The experts don't agree whether or not you should let candidates write code for some exercises on-the-fly. Joel Spolsky and Steve Yegge think this is essential. Scott Meyers and Kevlin Henney think it is a waste of time.
Apeirophobia is "the fear of infinity" and medorthophobia is "the fear of an erect penis". Though I'm still struggling to put this factoid into good use.
Armed with these bits of knowledge (except for the last one) I created a big list of questions. I made liberal use of the Guide to the Software Engineering Body of Knowledge. I copied its classification of knowledge areas, and changed it a bit so that it would better fit my purposes. I then used the SWEBOK guide and the abovementioned articles to come up with 100 questions in total.
I am not going to reveal all one hundred questions, as this could possibly lead to me hiring more kangaroos with perfect scores, dyslectic or not. This test will take about an hour when I make some random selection of between fifty and seventy questions. The end result will be a nice diagram showing the scores over the entire range of ten knowledge areas. Yesterday I tried the test on my good friend Niels. His overall scores were quite high. And the nice thing is that I now know I'm never going to let him contribute to any testing activities on my projects.
I realize that people who've seen these ten example questions will enjoy an advantage over other candidates. But, any candidate who can prove being a regular reader of my blog will get ten bonus points anyways. So this will surely not hurt the statistics. All in all, I'm quite happy with the results, as I think the new test is quite comprehensive. The only thing I haven't been able to test yet is cyberphobia and logizomechanophobia: "the fear of computers or working on a computer".
Note: Greetings to Johanna Rothman and Mike Griffiths, who will bringing the total number of readers of this blog to the staggering amount of 19.