This living document is a reflection of my philosophy on Ph.D. advising. I hope it clarifies my expectations and helps you judge whether I am a good fit for you.
When you are my Ph.D. student, I will evaluate you (as a supervisor) and guide you (as a mentor). Once you graduate, we will likely interact on an equal term (except recommendation letters). I will also promote you and your work before and after your degree. Your advisor has a significant long-term impact on your academic life, and you should choose them wisely. Many people use marriage as an analogy.
It is crucial to find an advisor who is an expert in your areas and a person who you feel comfortable working with. Too many people underestimate the importance of personality match.
Mental resilience (ability to cope with life’s hardship) is the most critical skill (above all the academic skills below) and is the key to a satisfying academic experience. The Ph.D. in the U.S. is a multi-year mental marathon, and you are on your own most of the time. You can work on (1) environmental support and (2) your way of handling things.
🔭 Check your future work environment. Collect data to assess whether you will feel comfortable working with your future advisor (me) and other people in the group or in the department. A good support network is generally helpful. Knowing your advisor as a person is as essential as knowing their research; I will now explain my personal view and knowledge so that you can check:
I have personal experience with gender and sexuality and am relatively knowledgeable in related topics. However, I am ignorant about many other topics. I need your help and guidance in navigating the topics you care about.
You might also want to know that the University of Minnesota has put lots of resources on student mental health.
🧘 Ultimately, you decide how to react to events. You can learn to handle undesirable outcomes more skillfully. Here are some possible methods, with their own caveats.
Now that I have explained mental resilience, I believe every researcher should additionally have the following three skill sets:
Here is my standard for (good) Ph.D. graduates:
👉 Your skills are more important than your records. Although you need to fulfill administrative requirements (coursework, written and oral preliminary examinations, etc.) and work on your résumé for your future jobs, they are secondary to these skills. I don’t care about absolute numbers of your publications, as long as your curriculum vitae is not barren. You should focus on your skills and research quality, with the understanding that many people in computer science do care about the numbers.
🧗 You need to practise. A LOT. For example, the answer to “how to read papers” might just be “read a lot of papers.” You can learn the required skills from experience.
📈 We will adjust our plan as we go. There is no particular timeline you have to follow. However, I will use the department’s default path as the baseline, adjusted by the time you might need to familiarize yourself with your research. It is common in my areas to learn for years before doing “real” research, which is different from many subfields in computer science. The evaluations will be personal and situational. Again, the important thing is to make steady progress in improving your skills.
🥱 You should make WPE and OPE trivial to pass. The written and oral preliminary examinations (WPE/OPE) check your research and presentational skills in the middle of your Ph.D. You should improve your skills so much that you can trivially pass them.
🔎 I pay attention to details. I can often provide novel recommendations after careful analysis. You can use me as a careful reader of your plans, papers, or whatever.
♊ It will work best if your attitude is similar. My research is to make software and mathematical proofs more reliable, motivated by my detail-oriented personality. I thus appreciate research that pays great attention to details and logical soundness. For the same reason, I prefer to spend less time on statements that seem to be vague or imprecise. Therefore, it will work best if you deeply share these concerns.
1️⃣ Most of my students work on different problems, even if supported by the same grant. This is different from many areas where team projects are common.
🎯 I strongly prefer direct communication, despite my cultural background (I grew up in Taiwan and know some East Asian cultures) and what you might have heard about “Minnesota Nice.” If you sense there is a difficult (often emotionally charged) issue, I strongly recommend expressing your opinions explicitly. For example, if you get tired of your current research project, and you signal it through sighing, body language, or indirect phrases such as “we should have more varieties,” it will take me forever to pick it up and we will have wasted time beating around the bush. It is best to say “I want to work on other things” directly if you can.
💼 I am more task-oriented than relation-oriented. I tend to show you my respect by doing my job—attending to your research needs, meeting with you, checking your well-being, and giving you advice. In return, you can show your respect via your dedication to work and your steady improvement in your skills. There is no need to build relationship before we start working. In fact, you are encouraged to discuss research or work in our first meeting. Relatedly, I usually do not actively ask you personal questions other than checking your wellness. That said, if you feel comfortable, I am happy to chat with you about stuff totally unrelated to work, and we (as a group) might hang out occasionally.
This may be culture shock to you if you came from a more relation-oriented culture where building trust is a prerequisite to doing business. If so, remember that it is important to meet agreed deadlines and that the lack of certain social aspects is not a sign of being rude or unfriendly. For example, the on-boarding process would focus more on getting you ready to work than knowing your peers, and I will start talking about the research projects without engaging in social events first. You should still know your labmates, but we will talk about work from day one.
🤷 No need to be “polite.” You do not have to be extra polite when talking to me. There is absolutely no need to give me gifts of any kind because I would rather see you spending all your resources on yourself. You can freely drop titles such as “Dr.” or “Professor” when addressing me. The best way to show your respect is to work on your skills and do good research. That said, I understand that the hierarchy based on age or position is critical in many cultures. If using the titles would make you more comfortable, feel free to do so. Just know that they make little or no difference to how I interact with you. “Hi Favonia” is equivalent to “Dear Professor Favonia.”
📅 We will meet weekly. In the beginning, we should meet at least once a week. (You can ask for additional meetings.) Over time, we may adjust the frequency depending on the need. You should treat our regular meetings as opportunities to ask for advice, not your obligation to show me progress (this is a common mistake). It is not recommended to skip a meeting only because you feel you have made no progress. In fact, you probably need my advice most when you are stuck.
📧 You should feel free to email me any time. There is no reason to wait for regular meetings to bring up an issue. If I am swamped, I will simply delay my response, and I usually finish reading your emails way before composing a reply. Regular meetings are only my minimum commitment that a portion of my time is dedicated to you.
😎 You make most decisions. You should feel free to express your opinions and your concerns. I make suggestions, explain the pros and cons, check your understanding, etc. But you are the focus of our meetings, and I will not make decisions for you in most cases. The main exception is funding-related decisions.
🙅 Please avoid careless attitudes, vagueness, and misinformation. For someone who has less power in a relationship, the natural inclination is to “appear good” by quoting others researchers’ words (without understanding them), minimizing problems, or exaggerating progress. This should be avoided because you would be wasting our time and slowing down your own progress. Here are some unhelpful conversations you should avoid:
- Student: I read the paper, and it seems the lemma they proved would not be useful for our theorem.
- Favonia: That sounds bad. What exactly did they prove?
- Student: I have not finished reading the paper.
- Favonia: Well, could you write down the statement of the lemma?
- Student: (appearing to be struggling)
- Student: I have been looking at how people do normalization in this paper.
- Favonia: Sounds good. How do they do normalization?
- Student: I am still reading the paper, but the definition seems related, and we should be able to prove our theorem using the result.
- Favonia: What made you feel that the result might be related?
- Student: The paper is about universes, and the authors are well-known in the field.
- Favonia: Which type theory are they talking about for normalization?
- Student: (appearing to be struggling)
- Student: Is normalization possible for type theory?
- Favonia: Why do you ask? Which type theory are you talking about?
- Student: Mike Shulman once said normalization can be tricky, but Jon Sterling said normalization by evaluation should work for all type theories.
The actual subject of the discussion is irrelevant. These hypothetical students are not helping themselves because they did not exercise careful thinking based on mathematical facts. As a result, there is no actual technical content in their answers. The conversations indicated weakness in their technical/mathematical skills and the lack of intention to drop the defensiveness and work on the substance. On the other hand, here are some conversations you should strive to have:
- Student: I read the paper, and I am not sure if the lemma would help.
- Favonia: That sounds bad. What exactly did they prove?
- Student: They showed the normalization of dependent type theory with a universe.
- Favonia: Why did you think it might not help?
- Student: We have a hierarchy of cumulative universes, and I am not sure if the algorithm in the paper will correctly normalize terms involving lifting operators… (explaining the concerns).
- Student: I have not finished reading the paper we found last time.
- Favonia: That’s fine. Is there something you wanted to tell me?
- Student: I implemented the language in the paper. I want to try out its normalization algorithm.
- Favonia: Sounds good. Is there something that you think might be an issue?
- Student: I am not sure how to represent a closure.
- Favonia: What do you mean?
- Student: Well, I could store a closure as a raw term, or an OCaml function that takes in the actual argument and… (explaining a few possibilities).
- Favonia: Ah, I see. I would recommend… (mentioning a few suggestions).
You are encouraged to apply for distinguished fellowships as they will be viewed highly on your résumé, but it is my job to find funding sources to support your Ph.D. In the “worst” case, you have to be a teaching assistant, but it is impossible to get even worse than that.
I should not be a co-author unless I have offered something intellectual to the work or written a part of the paper. I will explicitly tell you when I feel I have contributed enough to be on the list. Unfortunately, due to the power imbalance between us, you can only rely on my self-control. (Sorry!) You can also invite me to be a co-author, as you should to all your collaborators, and I will respond accordingly.
Being a teaching assistant for a few courses will sharpen your teaching skills and prepare yourself for many faculty jobs. The downside is that a teaching assistantship will eat up your time, just like any other activity. In the desperate cases where my research funding is running out, you might have to be a teaching assistant. Other than the funding crisis, it should be your decision.
You should give me your first draft at least ten days before the deadline. Chances are it would take several rounds to revise the draft.
It is expected that all of my Ph.D. students will share the lab with students supervised by Professor Van Wyk and Professor Gopalan Nadathur.
22 December 2022: Revised the tips about mental resilience and moved the location.
7 November 2022: Replaced the “compromised” suggestions about self-esteem with the more “radical” approach.
30 October 2022: (1) Rewrote the skills section to emphasize more on “you can do” rather than “I can help”, (2) revised many other parts, and (3) added emojis.
12 April 2022: Put in a new section on mental resilience.
12 December 2021: Rewrote the “key secret” about liberated mind and put it in the beginning.
14 November 2021: Rewrote certain parts of the statement to reflect changes in my personal attitudes.
22 Augest 2021: Added a hint about individual projects v.s. big team projects.
25 February 2021: Clarified my WPE/OPE standard given the recent discussions.
30 October 2020: clarified my cultural background and tweaked some words.
26 August 2020: added links to the CS&E department.
23 August 2020: tweaked “task-oriented.”
22 August 2020: elaborated on the “task-oriented” a bit more.
19 August 2020: fixed some typos; added an example in the “direct” paragraph; added a paragraph about “task-oriented;” added a paragraph about titles; replaced the last sentence in the “active role” paragraph with a stronger conclusion.
This page was made with GitHub-like CSS (licensed under MIT and modified by me).