[lhnn]

To me, it's a check against a teacher making an obscenely hard test that even the smartest students can't pass. The logic would be that if the best and brightest can only get a 60, then the exam should be curved.

This assumes that the best and brightest are actually really smart and knowledgeable.

[scott_s]

It's not obvious until you try to create a test yourself that creating a fair test is hard. Particularly for an intro course, since it's material that you know so well that you barely think about it consciously anymore. I took the same approach, and I was cognizant of the fact that some of my students came in with prior programming experience.

[JoshTriplett]

I had a professor once (for modern physics, i.e. relativity and quantum mechanics) who said that if anyone could get 100% on his test, they'd saturate the "sensor" and make the measurement less accurate; he wanted to see the bell curve centered low enough that it didn't get clipped by the maximum score. He calibrated his tests so the highest scores would typically fall in the 50s or 60s, and then applied a curve. Actually getting 100 would require knowing all the material for that course in advance and then some.

An interesting, if nerve-wracking, philosophy. :) It's the only time I've ever left a test and not known for sure what grade I'd get.