A Proposal for Graduate Student A/B Tiering

It’s that time of year again — graduate student enrollment.

Many people say: first-class undergraduates, second-class master’s students, third-class doctoral students. I won’t comment on whether that’s right, but the uneven quality of graduate students is certain. Although high school in China is cram-style education, the benefit of cramming is that everyone comes in the same, on the same level. But graduate students vary wildly — some are interns at Microsoft Research Asia, while others haven’t even passed CET-4, can’t use Windows properly, or studied classical Chinese as their undergraduate major… (You ask why such people are admitted? I don’t know either — the graduate entrance exam even has专业课 as a national standardized test, and they managed to pass!)

How do you create a unified training plan for such a diverse group? If you repeat undergraduate courses, many complain it’s too easy. But others can’t even understand undergraduate courses.

So I’m thinking of introducing a promotion/relegation system in graduate education, like in soccer and basketball. Regardless of your entrance exam scores or undergraduate grades, after enrollment, we have another round of assessment based on the Software School’s own standards. Then divide into A and B tiers. Tier A takes advanced courses — data mining, advanced graphics, formal theory, etc. Tier B should have no complaints — they go back to undergraduate courses, striving to reach the average level of our software school graduates. This might benefit them more. The tiering can be adjusted each semester. If someone in Tier B performs well, they can be promoted to Tier A; if someone in Tier A performs poorly, they must be relegated to Tier B.

Graduate students need to drop their pretensions. Thinking that as grad students they must do research, forcing ducks onto perches, won’t work. A more pragmatic approach is better.