Clarissa's Blog

24 senior mathematicians met at Harvard and agreed on something big:

A PhD should no longer be awarded mainly on the dissertation text. Their reason is that AI can now help write it. They want in-person, written, and oral exams instead.

https://x.com/mathelirium/status/2103843865631293599

I have no idea how it works in mathematics, but in my program, I wouldn’t have been allowed even to present a dissertation prospectus before passing several grueling oral and written exams that were based on a long list of required readings and accompanying secondary sources. I spent years preparing for those exams. I could still probably reproduce my entire reading list by heart.

My secondary concentration for the PhD was in English literature. I was allowed to do that because English is not my first language. On the day when I was supposed to have my oral exam in English, there was a blizzard in New Haven, and the professor didn’t show up in her office. I was freaking out precisely because, without passing that exam, I was not going to be able to proceed with my dissertation proposal. I called her on the phone and persuaded her to allow me to come to her house and do the exam there.

I didn’t have a car and didn’t know how to drive anyway. Cabs were not running because the blizzard was pretty bad. I schlepped on foot, falling into snowbanks, to the professor’s house. She was stunned to see that I actually made it there, handed me a plate of Indian snacks (the professor was from India), a pack of cigarettes and a bottle of wine. I refused everything because I was nauseous with fear over the exam.

The exam was supposed to be 45 minutes long since it was a secondary concentration. But the professor said that since I was already there, we should have a real exam. For over two hours, she grilled me on every single text on my reading list. That professor was a genius. She did not expect me and did not prepare. Neither did she remember the reading list she had given me a year earlier. I had to provide a copy of the list when I arrived. Right there on the spot, drinking copious amounts of wine and smoking like a chimney, she created a web of questions that were all interconnected and led in the same direction. Together, the questions formed an architecturally solid edifice that was a wonder to observe. I learned an enormous amount just from observing how she structured the exam.

I was soaked to the bone, but I answered every single question. “Wow, you actually did read all of the books,” the professor said. “OK, OK, you passed. Now will you start drinking already?”

The professor called a limo to take me home and packed the trunk with Indian food and two bottles of wine.

And hey, this was the secondary discipline. The comprehensive exams in my primary discipline were a lot tougher. In the oral portion that lasted hours, all of the professors were sitting around a table and asking me questions on the 165-book reading list. A professor of medieval literature would ask me something utterly arcane about a medieval text. I would start answering, and the professor of Latin American literature would interrupt and start quoting a poem by an early 20th-century poet and then pause expectantly, waiting for me to continue the poem by heart. After doing that, I would be expected to go back to answering the question on the medieval text.

And then there were the written exams. 20 years later, I still remember every single question on those written exams. And not everybody passed. In the year after mine, only one person passed the comprehensives on the first try. He’s a regular reader of this blog and can confirm everything I’m saying.

It’s not because of AI that the comprehensives are crucial. It’s because a dissertation, by definition, has a pretty narrow focus. If you’re going to get a PhD in a discipline, you need to be very widely read in the entirety of your field, and not just in your focus area. As one of my professors explained, “you have no idea what you will be asked to teach. Yes, you specialize in contemporary peninsular literature, but if you’re told to teach colonial Latin American literature, you need to be able to do that easily.” And guess what? My very first job after completing the PhD was in colonial Latin American Literature.

Another function of the comprehensives is to get you into the habit of reading a lot and extremely carefully. I also had a reading list, albeit a shorter one, and written/oral comprehensives for my master’s degree at McGill University. I spent 6 years of my life preparing for these different sets of comprehensives. And once you get into the habit, you can’t stop. I have read pretty much everything of note that gets published in my field and much of what gets published in contiguous fields.

When you go on the job market in my field, the dissertation is the least interesting thing about you. I don’t even remember being asked anything about the dissertation during job interviews, other than whether I was going to be able to complete it and defend it before my job appointment would start. What people care about is how wide your range is in both teaching and research. What are you going to be able to produce beyond the dissertation?

Sorry for the very long post, but this is a subject that matters to me a lot, and the idea that a PhD is just a dissertation bothers me.