I would like to present some classic computer science papers in a class I am teaching. I know teachers usually assign papers before the lecture, but I can't figure out how to present material they've already read and likely understood. In the case of a highly technical paper, I'd go over the more technical portions, perhaps letting them know beforehand not to struggle with the proofs, programming examples, etc., but I'm thinking of less technical papers, such as:

While I don't need to use pure lecture format, I am not sure that discussion would work well, since some of the papers (such as "Reflections on Trusting Trust") aren't really discussable, and my students lack the experience to discuss others (such as "No Silver Bullet").

EDIT: I was hoping not to task the students with extra prep, since they already have a lot of assignments to do, but, as some of you suggest, that may be the best way.


4 Answers 4


I recommend the discussion format. The two downsides of it you list can be overcome by using a guided discussion. When you assign the paper, also include a list of questions that you expect the students to be able to answer after having read the paper. When the class meets:

  1. Select students at random (or through a random permutation).
  2. Have the student answer an assigned question as a discussion seed.
  3. Open up the floor to other students to make comments and pitch in their ideas (thus starting a discussion).
  4. When the topic dies down, move to the next question on the assigned list and repeat from step 1.

This format has several advantages:

  • It ensures that the students have read the paper (since they might get randomly selected)
  • It encourages them to think of their own answers and participate in discussions.
  • It eliminates the intimidating factor of the professor having tight control of the discussion (since you only moderate who speaks)
  • It lets you address the key points of the paper (through your initial selection of questions) without having to go over the whole paper in class.

I have experienced this format in upper level undergrad/grad courses during my undergrad and enjoyed it greatly. It really encouraged me to read papers and reflect on them carefully.


One possibility is to ask the students to present.

  1. It gives an opportunity for the students to practice presenting something.

  2. Sitting from the audience point-of-view with (I expect) expert knowledge, you will be at a great vantage point for judging whether the students fully comprehend the paper and whether they are focusing on the "correct things". Sometimes students spend too much time tracking down irrelevant details and miss the big picture.

  3. Sometimes (depend on local culture etc.) students are more likely to engage in discussion and ask detailed questions if it is their peer doing the presentation. (The "intimidated by the professor effect".)

  • 2
    One problem with this approach is that unless the professor is very disciplined and keeps their mouth shut (oh the horrors!), the presentations degenerates into a dialogue between prof and presenter.
    – Suresh
    Apr 6, 2012 at 15:46
  • @Suresh: very good point indeed! Though from my experience (as a student in those kind of situations), usually the degeneration correlates with no one else having done the reading :) which I hope is already preempted in the context of this question. Apr 9, 2012 at 15:32

Usually I start by asking the students some simple questions about the paper. After a few questions I am so confident that most of them understand almost nothing about the paper, that I feel comfortable presenting it as if they never read it...


I propose the presentation format as well but rather than have the students present from anything, maybe have a pool they can choose from or assign topics? With regards to the discussion format, sometimes it can devolve into only a few of the students talking spiritedly while the rest are confused or bored.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .