US News recently released its latest beauty context results rankings of computer science graduate programs, including specialized rankings in subdisciplines. UIUC came in 5th overall (tied with Cornell and behind the Big Four—CMU, MIT, Stanford, and Berkeley—who all tied for first), 8th in artificial intelligence, 8th in programming languages (tied with my undergrad alma mater Rice), 5th in systems (after the Big Four), and 13th in theory. Here is the full theory list.
- MIT
- Stanford
- Berkeley
- Cornell
- Princeton
- CMU
- Harvard
- Columbia
- Georgia Tech
- Washington
- Texas
- Chicago
- UIUC
- UCSD
- Yale
- Brown
- Wisconsin
- Caltech
- Duke, Penn (tie)
- Rutgers
- Maryland, Michgan (tie)
UIUC also ranked 5th in computer engineering (behind the Big Four) and 5th in engineering overall (behind MIT, Stanford, Berkeley, and Georgia Tech, and just ahead of Purdue).
In mathematics, UIUC was ranked 17th overall (tied with Minnesota and UPenn), 10th in the bizarre category of algebra/number theory/algebraic geometry, 9th in discrete math and combinations (sic), 3rd in logic, 15th in topology. UIUC was not listed in the rankings for statistics, applied math, or geometry.
What do all these numbers mean?
Almost nothing.
Sure, the US News theory ranking is a small-constant-factor approximation of reality, where "reality" is defined as "my opinion". (In particular, UIUC's #13 ranking in theory is pretty accurate.) But they should be taken with a heaping pile of salt. Some excellent students would be much happier and more successful at Harvard instead of MIT, or at Columbia instead of Cornell, or Caltech instead of Georgia Tech, despite the disparity in rankings. The most obvious factor is the compatibility of research strengths and interests. For example, UIUC is one of the best places in the country to study computational geometry, but it's not a great place for complexity theory; Chicago is exactly the reverse. Other reasons include departmental culture, course options and requirements, availability of good (effective, productive, and/or well-known) advisors, funding, success at faculty placement, cost/ease of living, off-campus social/cultural life, family issues, etc., etc. Students ain't robots, nor brains on sticks, neither. One size does not fit all.
All else being equal, MIT and Berkeley are the best places in the country to study theoretical computer science. Alas, ceteris is never paribus.
A discussion on Lance's blog hits all the major points.
Recent Comments