[Calvin Locke]

If a student could demonstrate proficiency with a certain set of ideas and processes, why not let him or her move on to more advanced ones?

Because it's not fair to those that don't (and especially to the subset that never will) demonstrate proficiency...

[Myrddin]

I agree with Khan's approach. I used a similar approach when teaching embedded systems from 1980 to 1983. The assignments and grade earned for completing them was published the first night of class. Instruction is paced to present all the material over the semester. Individuals are free to progress as fast as they are able, but must complete each assignment in order and perfectly. The more skilled students helped me by tutoring those having difficulty and were rewarded with new challenges beyond the requirement to achieve an 'A' grade.

My approach was a hybrid between the self-paced physics classes that I tutored (taught) at UCSD and the curriculum of the electronics department for the embedded systems class.

The class was sponsored by Regional Occupational Program (ROP) and offered as the ROP class Electonics 91 or on the AS Electronics path as Electronics 51. The result over 3 years was conveyed to me by ROP. The people who completed the course were hired by DEC and IBM...91% of the class got hired! I stopped teaching the course in June 1983 as I had just accepted a full time systems analyst job inside Pacific Telephone (later Pacific Telesis). The job demands were too significant to risk missing teaching time due to emergency demands at the office.

At UCSD, I had experience with self-paced courses in physics, chemistry and psychology. I like the approach because I was capable of blasting through the self-paced course, earning the units and clearing study time for the rest of my non-self-paced courses. Most quarters were 18 to 22 units, so putting the wraps on it was important.