Anant Jain

How to Lead a Technical Reading Group

Paper Review

Introduction

I recently started a Technical Paper Reading Group at Brex. In order to facilitate the group better, I read this paper that explains the role of a technical reading group leader, and suggests strategies for facilitating a discussion. Here are some of my notes/highlights from the paper:

  • Your primary objective is to delve deep, learn, blabber on about your technical topic, and get to know others with this interest.
  • Your role as a group leader/facilitator:
    • Find appropriate reading material for your topic.
    • Facilitate discussions
    • Make sure that your reading group happens
  • Finding technical papers:
    • Ask professors/graduate students for suggestions — not too relevant for a workplace reading group like ours, but I reached out to some of the senior most engineers in our organization for paper suggestions.
    • Canonical papers, tutorial papers in the field
    • Important patents, theses, excerpts from technical books
    • Literature survey
    • Personal library
  • General tips for reading materials
    • Choose readings that will be academically accessible to students with the prerequisites you set for the group.
    • Aim for a length between 8 and 40 pages (for a 1.5 hour meeting — this doesn't apply much to us, since we intend to meet only for 30 minutes. I think of our reading group as more of an accountability mechanism than anything else.)
    • Make sure to select papers that provide good overall coverage of your subject area. In general, breadth is better than depth.
    • Try to make sure that most of the readings either discuss a particularly important result in your field, or are survey papers or review papers that have broad emphases.
    • Make sure that your readings thoroughly explain their methods, if applicable.
    • Ask your group participants for feedback on the papers selected and if they have any suggestions for future papers.
  • Facilitating discussion:
    • In any reading group, you can expect a maximum of 50% to participate actively, 20% to talk sometimes, and 30% to never talk.
    • At the start of the meeting, ask/make/force everyone ask 2 questions at the beginning and write them on a white board if possible. With these questions, all the participants now have a stake in the discussion and wants their questions answered. Importantly, this approach communicates to participants that they are allowed to not understand parts of the paper, but that they must read carefully enough to have questions. This is the secret to making reading groups work.
    • The first question you discuss can simply ask for the big picture and summary of the reading material.
    • Look out for and cut off discussions that seem to be wandering or not gaining much progress.
    • Make sure your group is able to discuss the entire reading within the scheduled meeting time if possible and respect that people have other things they need to do when the time is up, but definitely allow further discussion if there is interest.
    • Encourage everyone to talk!
    • Ensure that no particular group member dominates the discussion. This includes you, of course.
    • Always promote questions, especially the kind that lead to further discussion and insights.
    • Generally maintain lexical order in the discussion. Go from the beginning to the end of the reading.
    • Awkward silence is an indication to go onto the next question on the board!
    • Have a laptop available for reference when papers refer to things your group is not familiar with.
    • Invite experts if you believe a discussion could benefit from their presence.
    • A good closing question is what the group thought of the paper.
  • Some limitations
    • Lack of direction for the reading group. Mitigate by seeking out a goal for the discussion based on the reading material. Participants need not reach the goal while reading the paper, but the goal will give the entire discussion a general focus.
    • Student participation issues in discussions. Start with a slightly easier (less technical) paper to get everyone talking and more comfortable, and provide additional readings and outside resources for getting the members of your group up to speed to the level with which you would like to run your group.
  • Find a co-leader and a professional mentor for the group. Get the group to give lightning talks.

PDF

Over the next few Saturdays, I'll be going through some of the foundational papers in Computer Science, and publishing my notes here. This is #15 in this series.