Eigenmumbles

Last year, I was on the admissions committees for both computer science and applied mathematics. Consequently, I spent an inordinate amount of time in January reading folders. I agreed to serve on both committees again this year. The end of this process is in sight now, but before I put it behind me and get on with the semester, I thought I would share what I’ve observed about the folders that stand out.

I look for two things: background and research potential. The CS Ph.D. program at Cornell has qualifying exam requirements in theory, systems, AI, and PL that should be passed in the first couple semesters. A student who will struggle with those requirements is a risk. On the other hand, a student with great coursework who has never worked on an undergraduate research project is also a risk. There are exceptional students who are clear stars despite a lack of research experience or a hole in their background. But those students are the exceptions, and not the rule.

Strong applicants have strong transcripts. Though I adjust my criteria some depending on what I know of grade inflation at different schools (including what I can learn from the letter writers and from the other people reading folders), I look for mostly A or A- grades at most schools. I don’t mind a handful of B grades. But if a student who wants to study AI has B- grades in AI classes, I wonder. I care about the quality of the courses, too. A student who wants to study systems should have more than one good systems courses in her transcript.

Strong applicants come with letters that say they are special. A letter that says “Sally came to my office hours frequently, and did very well in class” does not lower my opinion of Sally, but it does not have nearly the impact of a letter that says “Sally made an important contribution to the research done by my group, and will be the third author on a major journal publication.”

Strong applicants have experiences with strong places. That does not mean that every applicant must attend a major research university; there are some outstanding small liberal arts schools that graduate highly talented students. It does mean that I place much more weight on an applicant from the University of Eastern West Northlands if she has a glowing letter written from a summer research project mentor at MIT.

I care about the statement of purpose. I want to read statements that tell me in clear terms why the applicant would like to attend Cornell, in what areas she might want to work, and which faculty she finds particularly exciting. I want to see that the student understands what she did in her research, and did not just follow directions without ever learning about the broader context.

I understand that beginning graduate students generally don’t know the topic of their dissertation. But I want to see that they have a feasible idea of where they might start. I am not impressed by a student who says she wants to do research on proving P != NP, and I am worried by a student who says she is fascinated by the work of one professor and would not dream of working with someone else. Murphy’s Law does not go away in graduate school, and the most famous professor in the department cannot advise an infinite number of students. By all means, dream big, but have a backup.