Regarding Graduate Student Service

December 14, 2020

Reposted from an older blog.

People who know me know that I am a huge proponent of treating academia as work. However, we do not work in a system that is particularly compatible with a modern notion of labor. The following advice attempts to balance how things should work (i.e., compensating people on the basis of their labor) with how they actually work (i.e., the reliance on unpaid work that is high value but not valued highly).

We can typically break service obligations into three categories: lab, department/college/university (internal) and external. I will discuss lab service last, since it can be the trickiest to manage.

Internal service

I typically encourage graduate students in their first two years to do only internal service. What is internal service? It's service work within your department/college. The benefits of doing internal service and prioritizing internal service are:

  1. It gives students a chance to meet other students and build a support network.
  2. It builds camaraderie between students who serve.
  3. It (usually) gives students a chance to meet faculty who are neither their advisors nor their professors, whom they may want to ask for letters later on.
  4. It can empower students to feel like they have control over their academic and immediate social environments.
  5. It can be a "trial run" for external service.

The major theme here is using internal service as a structured way to get to learn more about a department and its norms. There are many kinds of internal service. As an example, my PhD program had the following opportunities:

  • Organizers of Monday Morning Coffee and Thursday Tea. These were primarily low-effort weekly social events. If you were a shy person, it was a great way to break the ice with people.
  • Graduate Union Representative. This was my primary service over the years: our graduate students were unionized and serving made it easier for me to be in the loop for contract negotiations.
  • Seminar organizers and volunteers. There were a variety of seminar series for different CS sub-disciplines when I was in graduate school. These seminars ranged from being entirely student-run (e.g., the Machine Learning and Friends Lunch) to entirely faculty-run (e.g., the Systems Lunch). Typically student organizers and volunteers had the chance to select and then later meet speakers. While these positions were certainly work, they were considered very desirable, due to the networking opportunities they afforded.
  • Graduate representative to the faculty. This was a must-do service activity for any graduate student interested in faculty jobs. GradReps attended faculty meetings, interviewed faculty candidates, and were allowed to vote on non-RPT (reappointment, promotion, and tenure) matters. This is one of the best ways to see how the sausage of academia gets made.

Internal service should immediately benefit you, the student. If you are, for example, volunteering at a seminar, but you are just doing setup and cleanup and have no special opportunitites to meet the speakers, you may want to consider putting your efforts elsewhere.

External Service

I typically recommend that students wait until they feel more secure in their position before doing external service. This might mean having published their first research paper, having reached PhD candidacy, or having completed their coursework. The reason is that stakes tend to feel higher for external service, and the load also tends to be higher. It is not uncommon to feel overwhelmed in the first year or so of graduate school, so I am a huge fan of minimizing obligations and focusing on two things in the early years: (1) building your internal PhD support network and (2) growing your expertise in your subfield.

At some point, every student will start to feel more confident, and that is a very good point to begin doing external service. This may happen earlier for some folks than others. Some external service obligations are more intense than others, but all external service is important for building your reputation in your resaerch community. I recommend starting to incorporate external service in the following order:

  1. Student volunteering at conferences. Some students will begin doing this quite early. I think it can be beneficial early on to get a sense of what different research communities are like and what they work on. However, once you are sure of your research area, it is not worth being a student volunteer until you are at a point where you intend to attend conferences regularly. Once you do start attending conferences, being a student volunteer can be an easy way to meet other students. You will eventually "grow out" of this, though. I don't recommend being a student volunteer more than three times in your field; after that, people should start asking you to do more intensive external service.
  2. Student reviewing. Different subfields have different approaches to student reviewing. In SIGPLAN conferences, students who have not published at a conference are typically introduced to reviewing via Artifact Evaluation. In ML/AI venues, where the number of submissions is very high, many conferences have started asking for students to self-nominate as reviewers. I do not recommend reviewing papers for conferences until the latter half of your PhD.
  3. Conference/Workshop Organization. There are some annual workshops or small co-located conferences that involve student organizers, where being asked to chair that workshop or conference is considered a great honor. There are other circumstances where, e.g., you and your advisor may prepare a written proposal that you must submit for review. This kind of service is very labor-intensive and I only recommend it in the last two years of your PhD.

Lab service

Lab service is necessarily low-impact and can involve anything from taking notes during lab meetings, to maintaining the lab website, to mentoring junior students. Lab service can be highly beneficial to students for building camaraderie and trust. However, poorly distributed lab service can completely undermine mutual trust and respect.

One of the unfortunate truths of graduate school is that certain work just needs to get done, and it is simply easier for faculty to lean on "reliable" students they know will follow through. In a fair world, faculty would keep track of the balance between opportunitity and service and ensure that all students do their fair share, while no students are unduly burdened. Unfortunately, this is not common practice. In light of that, students must learn to advocate for themselves. This can be quite tricky, since students and faculty each have incomplete pictures of the situation.


It will always be necessary for students to stand up for themselves. The key is doing so politically. While exploitation can happen in any context, in my experience it is most egregious and difficult to manage when the parties involved know each other well: trust and familiarity are often preconditions.

That said, one person's exploitation is another person's opportunity -- only you can know if the situation is bad for you.

No matter who are you, when asked to do unpaid work, continually ask yourself: what am I getting out of this? Consider the above example, where you are doing setup and cleanup for a seminar series. If you enjoy doing this work as a break from you day, then no one should make you feel bad for not "getting more out of it!" Everyone finds satisfaction in different activities and it is perfectly legitimate to spend time on activities that don't translate to an impressive line on a c.v. or networking opportuntities if you find them fulfilling.

Remember: someone who exploits your labor now, or fails to see your contributions now, is not going to write you a strong recommendation later.

Quitting unpaid work

Quitting service can be politically tricky. Often the folks you report to are fulfilling service obligations themselves and can fail to advocate for themselves. In such circumstances, it's probably best to not come out and say: I feel exploited. It's better to simply say: this position is no longer working for me, and I need to scale back or quit.

If the service obligation has no fixed end date, then you should be able to quit whenever. Term-based service obligation is usually for a fixed length of time, and you should generally try to complete your term before quitting. If you cannot, be firm in your end date. Even if the person you report to is overworked and exploited themselves, they should not be operating in an environment where the margins are so thin that one person quitting prematurely ruins everything.

Finally, remember that in academic computer science, no one's livelihood is on the line if we have to quit a service obligation. If it were really that important, we'd be paid!