Advice for Graduate Students

UofT Grad Skill Seminars 2007

I recommend this book to every graduate student: Getting What You Came For by Robert Peters. (Much of the advice is not relevant to Canadian computer science students, but the overall message is, and it's a good read.)

Richard Hamming: "You And Your Research."

Randy Pausch's Time Management Talk: PowerPoint (better), HTML.

Ten Lessons I Wish I Had Been Taught, by Gian-Carlo Rota.

Last, and least, I recommend the book Academic Duty by Donald Kennedy to potential and future faculty. It's philosophical, rather than being a "how-to" book.

Reviewing Style: Against The Second Person

(This is a suggestion for everybody, not just students.)

I frequently see paper reviews that refer directly to the authors in one way or another, for example:

"The authors left out a reference to the work of X."
"Your method gives poor results in the second case discussed."
"You say that you use algorithm X. Why? You need to explain why you chose this algorithm."
"The authors need to run their algorithm on dataset Z."

In my opinion, this is very poor style for a paper review. A paper review should be about the paper, not the authors. In other words, all of the text of the review should discuss the strengths and the weaknesses of the paper that is being reviewed. The authors—who they are and what they did—are irrelevant and should not normally be discussed at all.

Referring to the authors directly makes the review personal. Directly addressing the authors often comes off as rude, or as an attack. Compare:

"This paper does a bad job of comparing its method to previous work." to:

"The authors do a bad job of comparing their method to previous work." It's like telling the authors that they, personally, messed up, rather than that the paper has problems. This can be quite offensive. In many cases, authors are aware of shortcomings in their work (e.g., they didn't get to everything before the paper deadline), and the problem lies not with the authors (which they may object to being told) but with the paper (with which they must reluctantly agree). Even if the comments are relatively mild, referring to the authors clouds their emotional response, which in turn can affect their understanding of the review, and even their judgment of the review process (e.g., "the reviewers for this conference are mean-spirited.") Furthermore, in a rebuttal phase, when the authors have very little time to respond to a review, inflammatory comments make it much more difficult for the authors to construct a reasoned, useful rebuttal.

It is very easy to remove direct references to the authors from a review. In many cases, simply replacing "The authors" or "You" with "The paper" (and adjusting the grammar accordingly) works pretty well. This works for the first two examples above. The second could become: "The paper uses algorithm X. Why? It should be explained why this method was chosen." The fourth can be "The paper should include results on dataset Y" or "Dataset Z should have been tested." This change does create more passive voice; in my opinion, this is preferable to being rude to the authors. It is still possible to express the paper's flaws in very strong terms while remaining polite.

I think there are a few cases where it is acceptable to directly address authors, e.g., to congratulate them on stellar work, or to chastise them for submitting the same paper over and over despite terrible reviews. But, in every case where you wish to refer to the authors, ask yourself: could I say the same thing without referring to the authors at all? Nowadays, nearly all reviews I write make no mention whatsoever of the authors.

If the review were addressed toward anybody, it would be addressed to the Papers Committee or Associate Editor, not to the authors. The review is making a recommendation to the committee/AE— (even when the review is written by a committee member). When I've been a committee member/AE reading a review, reading a review that refers to "you" is always a bit strange.