Discussions
In the spirit of the course content, we will crowdsource how we read and learn the material. To cover more breath, we will divide and conquer a set of research papers among the students in class. This “jigsaw technique” means that individuals only need to fully read 1 out of every 4 or 5 papers. Your peers will read, present, and facilitate discussion around the other papers, so you only need to skim these. That way, you become an expert on some papers while still getting exposure to many more! We expect active participation during all discussions and activities.
Each week, students will select and read ONE paper listed on the course Reading List. Then collaborate with 1-2 peers to deliver a short, high-quality presentation of your selected paper. We will sign up for paper slots together as an in-class activity on the first day. Then you can reach out to your peers to meet up during the week before you present.
Course dashboard (with reading sign-up sheet)
Keep your presentation short (5 minutes or so, definitely no more than 10 minutes) and prepare a set of key questions to lead a discussion. Maybe one or two of you presents slides and then the other peer facilitates the discussion. Presenters may prerecord their talks. Work with your partners and make sure to post your talk video to your section dashboard before class. Each paper will be allocated about 15-20 minutes of class time for presentation plus discussion.
How should I READ research papers in class?
As a presenter, you should read the paper in-depth. You will be teaching others who have only skimmed the paper. You might even print out the paper and take notes with a highlighter and pen. Do not just try to consume the information, but critically reflect on the assumptions, methods, and results. One way to start thinking critically is by questioning the paper. Here are questions to consider when reading:
What are the motivations and rationale for doing this work? (see the paper's Introduction)
What is the proposed solution? (see the paper's brief description in the Introduction and a detailed version in the body of the paper)
How was the core contribution evaluated? (see if the paper has an Evaluation or Results section, also read the discussion)
What are the contributions? (typically these are outlined in the Introduction, discussed in detail in the Discussion section)
What are future directions for this research? (see the Discussion section)
Thinking critically, what are the good ideas in this paper? Do these ideas have other applications or extensions that the authors might not have thought of? Can the results be generalized further? Are there possible improvements that might make important practical differences?
How does this paper relate to other papers and books that you have been reading? Sometimes you find that other researchers might have a different problem and approach, define the problem differently, or approach the same problem uniquely. How does this paper connect with other research?
Here are some additional resources to check out:
Pieter Jan Stappers created this Map of Reading, Reviewing and Writing and this nice video resource.
Bill Griswold wrote excellent advice on how to read and take effective notes on research papers.
Michael Mitzenmacher on How to Read a Research Paper.
S. Keshav offers a "3-pass method" for How to Read a Paper.
How should I PRESENT papers? How should I lead the class discussion?
Your goal is to succinctly teach students what you learned by reading the paper. You don't have much time, so try to hit the key points. In preparing the presentation, try to make your slides visually interesting. Include relevant figures from the paper. Try to reference at least one other paper related to your selected paper. For example, you might search for a paper that cited your paper, and then report back on how the newer paper builds on the prior work. Make sure to grab a screenshot or a figure from the referenced paper. Use words on your slides at a minimum. Whenever possible, show don't tell.
The last slide in your presentation should include the paper's title/authors and 2-3 questions for discussion. For discussion questions, it is usually helpful for the question to not have a clear answer, but rather be one that people might approach from different perspectives. Avoid questions that are either too easy (e.g., “Was it good that this happened?”) or too general (e.g., “What do you think about X?”). Take some time to create insightful questions that will fuel generative discussion.
Julia Baum's guidelines for how to lead a discussion of scientific articles.
Scott Klemmer's guidelines on thinking critically about research papers
Some example presentations with discussion questions: Future of Work, Planning with Crowdsourced Data
What should I do on papers I am not presenting?
For the papers you are not presenting each week, your job is to skim each paper before class (read the abstract and intro, skim the other sections, and review the figures) and actively listen during peer presentations and contribute to discussion. Be respectful but also inquisitive during class presentations.