How to initiate a paper discussion

Course Page

Many graduate seminars have a common theme: students read papers prior to class and then discuss them in class. Often, one student is asked to initiate the discussion by describing a paper and posing some initial discussion questions. Below are some suggestions for how to initiate discussions.

Be succinct -- Class time is limited. Your goal is to begin discussion, not to substitute for it. In general, you shouldn't take more than 3-5 minutes. Spend about 75% of your time describing the paper, then spend the remaining time on two or three questions.

Don't repeat what the author says -- Everyone has read the article, and there is no reason to repeat what the author has already said. Particularly if your grade depends on how well you initiate discussion, you may have a strong desire to be expansive, summarizing all the major points of each section in the authors own words. Resist this impulse.

Provide an "outside perspective" -- Instead of recapitulating the author's points or organization, provide an alternative way of looking at the paper. This will help other members of the class see the paper in a different way and improve the discussion. Try to move beyond the author's perspective, and cast the topics of the paper in a new light based on ideas and information not necessarily contained in the paper.

Compare and contrast with other papers -- Use the context provided by other papers discussed in prior classes. Consider how the work in the current paper is similar or different, better or worse.

Prepare ahead of time -- First, read the paper and note the major themes that you think best characterize the paper. Second, review the paper again and try to move beyond the paper. What doesn't the author consider? Are the experiments or arguments conclusive? What are the limits of the paper's conclusions? What lies beyond those limits? Third, reinterpret and reorganize the themes of the paper in the light of your review.

Hit the high points -- In your description and questions, hit the most comprehensive themes. Save minor points for later in the discussion.

Be provocative -- The goal of your comments and questions is to stimulate discussion, not necessarily to provide a balanced summary. If you disagree with the thesis of paper or if you think the authors missed something, say so. Consider asking questions that identify two or more extreme positions and then asks where the paper lies in that space. For example, you might ask: "Smith's algorithm can be viewed as either a trivial recombination of known statistical principles or an innovative and elegant approach that supercedes all previous work. Which is it?" Finally, consider taking an opposing view (even if you don't agree with it) as a method of encouraging discussion.

Make notes -- Make some notes about what you want to say in class. Don't write out all of what you will say, but do note the important points and your questions so you won't forget them when you present in class.

Practice -- After you make your notes, practice your comments once or twice. This will only take 10 minutes, it will improve your delivery, and it is the best way to check if you will stay within your target time (3-5 minutes).