A proposal for improving doctoral program ranking strategy.
Has anyone proven that the doctoral ranking of a university CS dept is actually predictive of the quality of an individual student or graduate? For "top" programs the quality variance may be less, yet this means that rankings are a crutch for analysis and are poor predictors of student quality outside of the top 10. Is the Pareto principle is at work here, ie the top 20% of CS programs are producing 80% of the research? A simple examination of top CS conference proceedings says no.
The following letter was published in the Letters to the Editor of the April 2012 CACM (http://cacm.acm.org/magazines/2012/4/147353).
The idea of establishing a universal index to rank university programs, as discussed by Andrew Bernat and Eric Grimson in their Viewpoint "Doctoral Program Rankings for U.S. Computing Programs: The National Research Council Strikes Out" (Dec. 2011), was apparently first proposed more than 50 years ago in the story "The Great Gray Plague" in sci-fi magazine Analog (Feb. 1962, http://www.gutenberg.org/files/28118/28118-h/28118-h.htm). I hope anyone proposing such an index today would first read that story and carefully consider the implications of limiting alternative sources of research. The index algorithm would likely miss the control variables, and different measuring variables would likely be used in studies in different institutions. Over time, the index algorithm would likely focus on the institutions with the highest-ranked indexes and their particular ways of viewing research results regrettably away from other lines of research that might otherwise yield potential breakthroughs through new theories.
Randall J. Dyck
Lethbridge, Alberta, Canada
Displaying all 2 comments