Programming Languages, Etc. Reading Group (PLERG)
» Format
The following is stolen from inspired by CMU's concertrg.
The purpose of PLERG is to expose students to more of the culture of our research area, and to read and discuss papers that will broaden our knowledge of the field. It also gives us an opportunity to interact with other students who share similar interests but may lie outside our day-to-day realm of activity. We try to cover a variety of papers of the following (non-exhaustive) types sorts categories varieties:
- Classic papers, papers which are part of the programming languages "canon"
- Current research papers, papers from the last several years that represent the breadth of current research in programming languages
- Related research papers, papers in slightly different but fundamentally related fields (i.e. theorem proving, formal methods, software engineering) that can foster interaction with other students at the institute
The format is described in three parts: how we select papers, what occurs before we meet, and what occurs during the meeting. Each PLERG session involves three roles, the Convener, the Leader, and the Guide. (With our small size, the Convener and Leader might be the same person.)
Paper Selection
Essentially we have a priority queue of papers, with priority assigned by members' excitement about that paper. As mentioned above, we will juggle classic papers along with more recent work in order to avoid starvation. No member should expect to read only those papers in his or her research area; indeed, one of the goals is to expose members to other, possibly related areas.
Members are expected to periodically nominate papers to read. The group as a whole will democratically decide the priorities of these nominations over email (or more likely, using some web tool) and at the end of each meeting. Compilations of great PL papers, like Benjamin Pierce's, might be a good place to start, or papers from a recent conference one attended.
Before the Discussion
While reading is obviously the primary activity before the discussion, we draw the following distinction from other discussion formats. In addition to reading the paper, each member is responsible for a list of questions, comments, or issues raised by the paper. Questions may touch on specific technical details or raise broad concerns that will initiate discussion or improve your understanding of the paper. Asking questions does not require expertise, and asking to learn something is a productive way to contribute to the discussion.
No slides are drafted for the discussion. No single member is "responsible" for a given paper, but the Convener selects a Guide, a student that has some familiarity with the subject area and can ensure that general questions can be answered (younger students are encouraged to be Guides, but older students will often fill this role). That said, don't expect to show up and glean the content of the paper from the discussion. As you'll see below, we don't waste any time in summarizing the paper.
Quorum During the height of the academic semester (and the summer months when many students are away), meeting once a week sometimes becomes a burden more than a chance to take a break from the usual run-of-the-mill. To ensure that we have enough people to make the discussion interesting, we have worked out the following "quorum" system:
- On the weekend before the RG, the Convener sends a "Quorum Call" email.
- If you plan to read and attend (the following meeting), respond to the meeting announcement (hopefully by noon Monday).1
- The Convener collects the emails and sends an email to the group announcing the result.
- If the threshold is met, we meet; otherwise, we put the paper off for a week.
Obviously members that don't send mail are still welcome to attend. This is just meant to be an easy way to account for paper deadlines, vacation, etc. We hope to keep the process as lightweight as possible, while still ensuring good discussion at a predictable time. (This mechanism also provides the Convener with an excuse to give gentle reminders to the other members.)
1If you can't attend, but don't want to miss this particular discussion, you may attempt to "veto" the meeting time (but then you're required to attend the following week!)
During the Discussion
We begin each session by "going around the table," sharing our questions and comments and listing them on the whiteboard. At this point, however, we try to limit the discussion of these questions: the goal is to create an outline for the remainder of the session. (By tradition, the last individual to arrive is given the honor of offering the first question.)
We do not spend time summarizing each paper. Instead, questions and comments guide the discussion and serve as the means to understand the paper. Because we do not summarize the paper, each member must come prepared! If a member is too busy to read the paper, then he or she should respond negatively to the Quorum Call.
The Guide and the Leader are the primary roles during the discussion, and should usually be two different people.
- The Guide is there to increase the likelihood that some answer can be found to most questions, but their primary role is answering questions, not proposing them.
- The Leader may be the same person as the Convener, and is also a good role for younger students. He or she leads the discussion, and tries to make sure that the discussion stays on track. In particular, the leader is responsible:
- To ensure that discussion continues to move from one topic to the next and that it doesn't get bogged down, and
- To verify that questions are answered (to the satisfaction of the asker) or that some time in the future is set aside to continue the discussion.
Obviously, there is a balance between giving detailed answers to a small number of questions and cursory explanations to many. Maintaining this balance should be driven by overall goal of the discussion, to include as many people as possible; to this end, priority may be given to more basic questions to ensure that no one gets completely lost.
During the meeting, everyone should be thinking about topics for the next discussion. Difficult questions raised by one paper can be a good guide to determining a paper for the following week.
Revisiting the Format
We want to make sure we explicitly talk about the format at the beginning of a year (to explain it to new students) and the end of the year (to see if we want to change it). Changing of the guard in terms of Conveners has also been a good time to revisit the format of our discussions, but the idea that current students "buy in" to what we're doing is more important than any specifics.
* Conveners:
Scott Kilpatrick (2011-2012)
Imprint / Data Protection