— 10 June 2021 —
The summary of this (preliminary) proposal is that we need to have two types of advisors for PhD students at universities.
I argue that decoupling mentoring from research advising (which involves a power dynamic and an incentive tension) has several benefits for the students and for our community as a whole.
A disclaimer of limited-perspective: This mostly applies to the university system in the US and more specifically my field (computer science) and might not be generalizable or even make sense outside of this intersection. For example, in Europe PhDs are usually 3 or 4-years long, and students usually commit to a research advisor before even joining the program since their funding depends on them and so it is difficult to imagine how to apply such a proposal.
The main idea behind this proposal is to shift faculty incentives, so that they better align with the students’. Currently, advisors have strong incentives to publish many1 The word “many” here can be replaced with “impactful” and merely serves as an indication of the requirement for hard work. research papers with their students. This is especially true for younger faculty, who need to “prove” their worth to the department to get tenure and to funding agencies to get grants. This leads to advisors pushing their students to work harder in order to have adequare research output. This might not always align with students’ interests, either due to temporary causes, e.g., if a student is having personal issues for a period of time, or on a more fundamental level, e.g., if a student is interested in an industry or teaching career, both of which have different requirements. In this situation, there is tension between the interests of the student and the advisor, and due to the power dynamic between them, an advisor might choose to push the students for research output even if it does not align with the student’s goals.
Students in this position are usually powerless: the power dynamic makes it very hard for them to argue with their advisor in order to satisfy their own interests and agenda. If a student lacks perspective and has not experienced a healthy advisor-advisee relationship they might even consider that the situation is completely normal and how PhDs should be. Currently, the best a student can do is to seek help through univesity procedures or their support network, and the usual response is that their alternatives are to (i) push through and live with the issues until they finish their PhD, (ii) abandon the program (which might not always be possible if they don’t have another job opportunity streamlined), or in rare cases (iii) switch advisors, a process that is not encouraged in many universities and even when it is, it is stressful and requires significant effort from the student’s side.
Creating an explicit mentor role that is decoupled from research gives us the ability to be more specific about goals such as the student’s well-being and fullfilment; goals that are currently mostly ignored giving their place to more tangible ones like the research output and success of a student. These goals could become part of faculty evaluation (similarly to other criteria like research output, grants, service, and teaching). A mentor would be successful in their role if the mentee student feels that they are empowered to pursue and satisfy their personal needs, even if these needs align with the student leaving from the PhD program to pursue an alternative career. Making the mentor role explicit and part of faculty evaluation also helps by giving mentors more power to argue for the student internally with other faculty, since their incentives are also aligned with their mentees. A mentor’s incentive is to be an ally to their mentee, making sure that their needs and opinions are not swept under the rug.
Another benefit of this role separation is that it lowers the stakes of students’ decision of which PhD group to join, since they can choose the research project that they want to work in without being concerned that their mentoring needs will not be satisfied. Mentors should be assigned to the student from the start of their PhD, while a student could commit to a research advisor later, after exploration and more discussion. Furthermore, this proposal also benefits the university’s reputation as PhD students will be better equipped to pursue their goals and succeed.
Note that such mentorship relations between students and faculty already exist for students that approach faculty and create connections with them. However, the fact that it is not a proper and explicit role gives less incentive to faculty members to commit a lot of time to it. Furthermore, students that have the ability and confidence to seek out help and mentorship from faculty members are often not the ones that need the most help. The less connected and less confident students are often the ones that would benefit the most from such mentoring relationships; even having an external perspective from another faculty member is a important to realize that one’s situation might be problematic.
This proposal is complementary to mentorship services like SIGPLAN-M that focus on inter-institution mentorship relations in a specific field of CS. Intra-institution mentors have internal power to argue for their mentees and give them leverage for their interactions with their research advisors. Furthermore, it is unlikely (and maybe not preferable) that the mentor is in the exact same field as the student, so field-specific mentorship services like SIGPLAN-M are still beneficial since they can offer more targeted mentorship services (including additional networking, internships, connections, etc).
Here are some open questions that the above proposal leaves unanswered (some of which might also affect its feasibility). It would be interesting to think about them and discuss them with others in the community:
What is the exact role of the mentors?
What are the precise faculty evaluation criteria for their mentorship service? Should it be based on explicit student feedback using evaluation forms?
Who funds students? The current proposal (conveniently) ignores this.
Should there be research intersection between mentor and mentee? It would make sense if they are both in the same field, e.g., CS, but not necessarily in the same subfield (PL, systems, ML, theory, etc).
Should the students be able to choose their mentor (as they do with research) advisors, or should they be assigned someone by the university and then have the ability to request for a change, e.g., if they do not feel comfortable with the pairing?
Explicitly naming the mentoring role for faculty members in universities will allow us to focus, precisely define, and evaluate this role’s extremely important goals. This could potentially help students who need it the most; ones that have weaker support networks and could end up in an unfavorable situation with their research advisors without any means to escape from it or anyone to get support from.
This is a very rough proposal and possibly has many holes, so feel free to contact me if you have any thoughts or feedback. I would love to hear everyone’s opinion on this :)
Hosted on GitHub Pages — Theme by orderedlist