Should We Encourage The PhD?

This is the second part of my feedback document to the steering committee of the Programming Languages Mentoring Workshop. See Part I.

Part II

Grad Skills: the Vague vs. the Concrete

Grad-Skills talks give advice on how to do well in a specific activity that is required from a PhD student.

Two increadibly valuable and memorable grad-skill talks I have seen are How to Write Papers so People Can Read Them by Derek Dreyer and Unaccustomed as I am to Public Speaking by John Hughes. The talks give practical, concrete advice on how to write a paper and how to present a talk. I like how both talks illustrate on live examples what mistakes to avoid in writing and presenting, and how to fix those mistakes. Both talks influenced the way I write papers and give presentations to this day.

Given how useful these talks have been, I would find it great to have more grad-skills talks at PLMW.

Life Story: Survivorship bias vs. More viewpoints

A PLMW life story is an autobiographical talk that aims to demonstrate how doing a PhD can lead to a successful academic career path.

Life-story talks are given by accomplished PL researchers who usually walk the audience through a timeline of their research career. They describe where, when, and with whom they worked on their main research problems and which results came out of that. Throughout the talks, the presenters also insert pieces of advice to the students, relating to the particular topic they discussed. The talks often relate to the audience by showing that the researchers, even though now they are successful, went through difficult periods of failure, which they were eventually able to overcome to get to where they are now.

Life-story talks show the human side of researchers. They provide a refreshing break from the overload of new technical information at the conference. They are interesting to listen to. But from my experience, entertainment has been the only value of these talks. In particular, the talks suffer from two problems that are inherent to the talk’s format:

  • self-centeredness, because the story focuses on the speaker’s persona and does not address the interests of the audience;
  • survivorship bias, which makes it impossible to extrapolate the speaker’s personal experiences and advice to a general audience.

In my opinion, it is so difficult to avoid these problems that it might be better to remove life-story talks from the program entirely. If getting rid of life-story talks is not an option, in the following three subsections, I would like to propose approaches to combat the life-story-format problems. Below, I will illustrate my points through an overview of two example life-story talks that I attended at PLMW.

Focus on audience instead of self

It is probably possible to make a life-story talk useful and insightful to the audience, but the format of the talk makes it hard. Life-story talks are inherently about the speaker, and it is difficult to make them relatable to the audience. Public-speaking guides usually emphasize the importance of making a presentation relevant to the audience, for which the speaker needs to center their talk around the interests of the listeners, as opposed to focusing on their own persona. For example, Stephen E. Lucas writes:

What do people want to hear? very simply, they usually want to hear about things that are meaningful to them. People are egocentric. They pay closest attention to messages that affect their own values, their own beliefs, their own well-being. Listeners typically approach speeches with one question uppermost in mind: “Why is this important to me?” . . . You must relate your message to your listeners – show how it pertains to them, explain why they should care about it as much as you do.

In order to prioritize the audience’s interests, the speaker cannot focus on themselves, as pointed out by Nancy Duarte:

When trying to connect with others during a presentation, you have to remember that it’s not all about you. Audiences detest arrogance and self- centeredness. . . . Instead, embrace a stance of humility and deference to your audience’s needs. Begin the presentation from a shared place of understanding. Make it about the audience. . . . When you’re presenting, instead of showing up with an arrogant attitude that “it’s all about me,” your stance should be a humble: “it’s all about them.”

It might be difficult to avoid turning a talk about the PL researcher’s successful career into a story of their achievements and successes. I do not know how to devise a universal strategy of making anyone’s life-story talk useful to the audience. But I think that the talks can be improved if the speakers consider the following questions when planning their talk: who is my audience, what are they interested in, what moments in my life can they relate to, and what decisions did I make that can inspire them?

Avoid survivorship bias by inviting people who withdrew from PhD

It might seem that accomplished researchers are the best people to provide insights into how to be successful: after all, they applied those insights themselves, and it “worked.”

However, concentrating on the insights of successful people, while overlooking the perspective of the rest, is exactly the reasoning that defines survivorship bias. We do not know what brought the presenter to where they are now. It was likely a combination of great interest in the field, talent, and luck, but these are the circumstances of the speaker; it is difficult to derive what resulted from the speaker’s character, particular circumstances, and coincidences into useful takeaways for a whole audience of students with different backgrounds. For example, if the speaker went to grad school after working in industry, they might say that this is the best career path because one gets to learn about problems in the real world first. And if the speaker went to grad school directly after their undergraduate studies, they will advertise that path because it does not waste time, and because choosing to be poor after earning lots of money is difficult.

Perhaps the best way to counteract survivorship bias is to invite people who decided to quit their PhD. I would find it very interesting to hear from them and learn why they decided to quit.

Somebody who decided to quit the PhD can have insightful and specific answers to how to do well in the PhD, because they know what situations to avoid. They can also provide details into what happens when somebody wants to quit the PhD, which is important because half of PhD students across different fields withdraw from their programs, and the question of quitting is often brought up by students at PLMW, but rarely properly answered.

Finally, inviting such a person would lessen the stigma that we have against quitting the PhD. I think that people who quit their PhD deserve the utmost respect, because additionally to having the skills and interests necessary to become a PhD student, they had the courage and understanding of their own desires, which allowed them to quit the PhD and move on, in spite of the negative reputation of quitting.

By giving people who quit their PhDs a say at PLMW, we allow students to get a better and more realistic picture of the PhD, which can help them make an informative decision.

Talk about personal challenges, add details

Even if we look past the survivorship bias, there is another problem: in order to give advice based on a story, the story needs to contain a problematic situation, like a conflict, or a personal challenge. For example, the most common theme in PLMW-life-story talks is the “ups and downs” of research: presenters try to convey the idea that even they, who are now successful, went through problematic and unsuccessful stages in their careers. But it is difficult for the speaker to relate any details of their personal struggles, especially in front of a large audience of colleagues, and when it requires mentioning problems with specific people. As a result, the advice drawn from personal stories is often superficial (e.g. “work hard,” “love what you do”) and supported by vague examples, if any. I find that merely stating that one had their high and low moments provides few insights. Instead of giving advice that is likely very general or biased, perhaps it might be more effective to talk in detail about some challenging moments in the speaker’s career, about how they felt during that time, and what they did to overcome those problems.

Example talks

To illustrate this section’s ideas, I would like to discuss two example-life-story talks in more detail.

“My 25 years in OO”

The best Life Story talk I have seen is My 25 years in OO by Jan Vitek. I believe that given the constraints of life-story talks, it is difficult to make something that is better than this talk. It tells the story of Jan’s research career. It walks us through the universities where he worked, the people he collaborated with, and the papers he wrote. The talk is funny and entertaining, and is full with little pieces of advice: as an example, he discusses paper rejection, explains that rejection happens to every researcher, encourages us to treat paper rejection as an opportunity to improve our research and especially its presentation: as Jan puts it, every paper is “1% inspiration, 19% perspiration, 80% communication.”

At the same time, I believe that instead of encouraging or teaching the audience about academia, parts of this talk deliver a different, unintended message. I also believe that this is a direct result of the life-story format: the talk is inherently focused on information about the presenter and does not allow us to extrapolate this information onto our own lives.

Here is an example of an unintended message in the talk: Jan mentions having a difficult relationship with his PhD advisor. He tells the story of how once, after a long absence with little or no communication, the advisor enters Jan’s office. The advisor asks “How is your PhD going?”, Jan says: “it’s going well,” and the advisor leaves.

Jan compares their relationship with the interaction between the movie director Werner Herzog and Klaus Kinski, a lead actor in Herzog’s movies. There is a slide with the following quote by Herzog:

I did not love him, nor did I hate him. We had mutual respect for each other, as we both planned each other’s murder.

I enjoyed this part of the talk, because it is funny, and because incidentally, Herzog is one of my favourite movie directors. I felt happy to see a famous PL researcher giving this quote from a rather obscure documentary – maybe there is space for more than PL in a PL professor’s life, after all.

However, what is the message behind this story? I am sure it is intended to be an inspiring one: even a successful researcher can have obstacles on their career path; in spite of having a hard time with an advisor, one can still finish the PhD, write great papers, and become a professor.

But I find that the story is lacking details such as: what was the problem with the advisor? did the advisor not provide any guidance to Jan? how did Jan resolve those problems? As a result, what I actually concluded was that Jan is a brilliant researcher who managed to finish his PhD on his own, without any help. I know that I, by contrast, require my advisor’s help in my PhD. Do I need to be able to do a PhD on my own, without my advisor’s help? If so, I worry that I am not suited for academia.

Another unintended message appears throughout the talk: Jan points out that many of the collaborators during his research career are people whom he met at his first conference. Those people are highly accomplished, famous researchers such as Luca Cardelli and James Noble. This should encourage us to go meet people, talk to them, and find collaborators at the conference: “The first workshop you attend may shape your entire research career, as it did here. These people provided inspiration, advice, acted as reference, and offered employment,” says Jan.

But I was interested in how he met those people. Was it because he proposed them interesting ideas for a collaboration? How did he approach these people? Did they approach him? Without knowing this, all I know is the information that the speaker collaborates with great people, but what can I derive from that for myself?

“Love, Marriage, and Happiness”

Another PLMW talk I would like to cover is named “Love, Marriage, and Happiness.” The talk conveys a deep fascination with programming languages and aims to encourage us to pursue our biggest interests, as opposed to prioritizing trendy topics and lucrative careers. I can relate well to this idea and completely agree with it. At the same time, I wonder how well it achieves this goal.

Before seeing the talk, I thought that the talk would be about balancing work with family life. 1 However, the words “love,” “marriage,” and “happiness” in the title relate to programming language research, and not to a person, as do the 31 occurrences of the word “love” mentioned in the talk. From its abstract:

Wanting to go into research in programming languages must feel like falling in love. Working on your PhD is going to be much more like being in love. So, as you are about to embark on your PhD studies (or wondering whether you should), I suggest you keep this analogy in mind.

Another quote from the talk:

Falling in love is when your body dumps out hormones from 200 to 400 days and you’re out of control. That’s falling in love; that’s the English phrase “falling in love,” as opposed to “being in love,” which comes afterwards. That’s why I said “you have fallen in love with programming languages” and it may not work for the rest of your life . . . cause that’s about being in love. You have to create a sustained relationship with what you’re going to do.”

I find the emphasis on having to “love” work, just as the necessity to have an intense interest in it intimidating. It is okay for the speakers to describe their subjective feelings towards work as a strong emotional attachment, but I find questionable whether these feelings should be presented as the required attitude towards work.

I am not sure that, especially at the beginning of grad school, every PhD student is driven by intense love. For example, strong interest can be related to knowing a subject well and having the freedom to work on one’s own ideas. But at the beginning of grad school, many students work on projects that are proposed to them by their advisors. They might not know enough about a field to be strongly interested in it.

As if to confirm this point, in the same talk, the speaker admits that PL research does not always start with love:

They had to drag me into this area, kicking and screaming. Types? Who wants to work on types? I mean for heaven’s sake, what a boring idea. And then, a little later, [I realized] that is actually pretty cool!

And even after one develops a strong interest, it is unnecessary to demand of somebody to “love” the field. Even somebody who feels enthusiastic, excited, and interested in PL can get intimidated by this rhetoric. When I heard the sentence “if you’re not intensely interested, why are you in grad school?”, I felt like an impostor.

A possible purpose of the talk is to convey the speaker’s interest in the field to the students. However, a great way to do that is to present interesting PL problems and solutions, since they will speak for themselves; and that can be done in a technical talk. Explicitly stating one’s passion does not pass on the “love.” On the contrary, I find these overt expressions of enthusiasm intimidating. They make me wonder whether by not being able to describe my attitude as passion I am missing the prerequisite for being a researcher.

One last example of how the talk relates more to the speaker than to the students: a key takeaway from the talk is that we should follow our biggest interest, as opposed to simply pursuing a hot topic or making money:

That’s a research career. Sorry, that’s a research engagement with a topic. Not a career. Not a career. A career falls out of that. That is my idea. I don’t go for a career.

Even though it might be true, I find it easy to say “I don’t go for a career” after one made a successful career. But how many members of the audience can afford to say the same with that level of confidence? As a result, instead of encouraging us to pursue topics that genuinely interest us, the statement simply tells us something about the speaker, and what attitude happened to work out for them in retrospect.

Technical Talks: Self Advertisement vs. Interesting Problems

The technical talks at PLMW provide overviews of specific sub-fields within programming languages. The talks typically cover the speakers’ area of research. I believe that technical talks can be useful to students who are getting started with PL research and want to learn about the various areas of PL.

I find technical talks at PLMW most helpful if they:

  • provide an introduction to an area of PL research,
  • cover a broad subfield of PL: this will give the students an opportunity to learn about a wider range of problems and prepare them to understand more at the conference.

The technical talks I have seen at PLMW usually fall into one of two categories: a broad introduction to a field, or a narrow focus on the speaker’s research. In my opinion, the more useful talks fall into the first category; they encompass a broader subfield of the presenter’s research, rather than focusing on their current paper.

An example is my favourite technical talk, Two Notions of Beauty in Programming by Robert Harper. This almost philosophical talk discusses the intersection of algorithms and language design. It presents the problem of combining the first field, which focuses on efficiency, and the second, which centers around abstraction and composition. It covers broad problems in PL and automatically conveys a fascination with the field, without the need to talk about feelings.

I find that the other category of centring the talk around the speaker’s narrow field of research should be avoided. Given that the mentoring workshops are co-held with major PL conferences, PLMW attendees already have the opportunity to visit a number of specialized technical talks. However, most of them presume existing knowledge of a field and are too short to give a thorough background and motivation, which makes them difficult to understand for someone outside the field. That is why PLMW can provide a great opportunity for introductory, accessible technical talks.

Furthermore, because PLMW is aimed at beginning graduate students, if the talk focuses just on the narrow area of the presenter, it sometimes looks close to a self-promotion or student-recruitment talk (for example, at the end of one talk the presenter said: “Now some shameless advertisement: if you’re looking for PhD positions, we’re hiring”).

I think that giving interesting technical talks is the best way to convey excitement for PL. There is little other reason to do a PhD than being interested in the field, so all the encouragement should originate here. By contrast, the other attempt at getting students interested in PL by discussing one’s love and passion for the field is only focusing on the presenter’s feelings; it does not apply in any way to the audience and is therefore of little use. I believe that a technical talk is a subtle and the most objective, informative way to encourage students to do a PhD. The encouragement is subtle because it is implicit: rather than stating how great the PhD, PL, and academia are, we allow the students to think for themselves, after seeing an interesting technical talk: “This topic looks exciting to me, and working on open research problems sounds cool, maybe I should consider doing a PhD!”

Should we encourage the PhD?

Given the psychological and financial insecurity that comes with the PhD, plus the poor job prospects for PhD graduates, is it a good idea to universally encourage students to do a PhD?

If the answer is “Yes, it is a good idea to keep encouraging everyone to do a PhD,” then there is still space for improving PLMW to achieve this goal. The encouraging moments at PLMW for me came from the technical talks and two talks that attempted to depict the PhD realistically. I found all explicit encouragement either too vague or actually discouraging.

Another answer is: “No, not everybody needs to stay in academia, and it is fine if some students decide that it is the wrong career path for them. PLMW should encourage the “right” students to stay, and discourage the rest.” If that is our answer, we just need to understand who these students are.

A natural criterion for selection would be the students who are interested in PL. But this is difficult to quantify, so we could let accomplished PL researchers describe their excitement about their work, and target the students who can identify with those feelings. They (the researchers, and thus the right students) love programming languages and are willing to dedicate their lives to the field.

Perhaps PLMW is already achieving this goal, and perhaps it is indeed the purpose of the workshop to attract only the students to whom work is their hobby, or to whom research is the most important thing in life. If so, it is fine to demonstrate that this dedication is what it takes to be a PL researcher. Then the less passionate, undesirable students can choose to stay out of academia. Additionally, this might disillusion students about fitting into our work culture, and decrease the overly high ratio of graduating PhD students to available academic positions.

I have written this document under the assumption that these are not our answers, and that nobody should be encouraged to do a PhD, but rather informed, and I propose to accordingly replace the word “encourage” in PLMW’s purpose statement. Focusing on informing rather than encouraging will help make PLMW’s advice concrete and genuine. It will motivate us to honestly address existing PhD problems that we don’t talk about, but which many students relate to, and for which they need support.

We should not be afraid to bring up the problematic sides of the PhD in fear of discouraging students to pursue it. First, we are already discouraging students by intimidating them with love-and-passion-for-PL rhetoric, and a realistic portrayal of the PhD is more useful discouragement than suggesting that everybody has to have some specific, intense feelings about the field. Second, it is great if knowing something that is true about the PhD will make a student decide that they do not want to sacrifice themselves to research. If that happens due to PLMW, then the workshop will have achieved a great purpose, because it prevented disappointment, health problems, and an overqualified PhD without a job.

Informing, not encouraging, will help students make informed decisions about whether to pursue a PhD, rather than encouraged decisions that came from a person talking excitedly about a great life in academia. And enabling students to go into academia in our field, if that is what they really want, is after all what PLMW is for.


I would like to thank Derek Dreyer, Rose Hoberman, Abel Nieto, Prabhakar Ragde, Ondřej Lhoták, Olga Zorin, and Alexandra Vtyurina for their help in proofreading this document and providing invaluable feedback that helped greatly improve it.

  1. On that topic, the talk contains the following remark: “Being in love can be pretty bad. Things can not go well sometimes, things will not go well, I promise you, especially with a kid, when you’re into grad school.” I am unsure whether it was intended, but to me it sounded like advice against having children in grad school, which I perceive as a strong message against having a work/life balance, and prioritizing school over everything else.