To this end I set a small challenge: write a routine to perform a binary search, but don't test it.
I thought people might be prompted to think differently about their code and to take it as a real challenge to get it right. However, people seemed to dash off their code, and then when it failed somehow say that the test was unfair. Clearly I was wrong in my belief that people would see that the whole point of the challenge was that inadequate (in this case non-existent) testing should force them to change the way they work.
What really bothered me was that so many simply didn't seem to care that their code was wrong.
I am in the process of starting up a second challenge, with any luck having learned something about programmers from the first. As of 2002/01/17 it is being checked by a few hardy souls before it goes live. If you're interested in helping this process, or if you're interested in trying the challenge when it's ready, let me know at mailto:email@example.com
In the meantime, here is a link to the first challenge:
ContentsThere were no headings
in the main text so there
is no table of contents.
Links on this page