Lessons from the New Martians: Evolving Through Your Graduate School Years

David Brumley

October 16, 2023

In 2003, I arrived in the CMU computer science department for my PhD along with 30 other students. There must have been something in the water, because that one class seems to have had an exceptionally large impact.

Among those 30 were the scientists Elaine Shi in cryptography, Vyas Sekar in networking, Virginia Williams in computer theory, Sonia Chernova in robotics , and James Hayes in Vision and Graphics.

I call it the class of the New Martians.

“The New Martians” as a homage to a quip between Enrico Fermi (played by Danny Defarrari in the Oppenheimer movie) and Leo Szilard (the physicists who first proposed the idea of a chain nuclear reaction). Fermi once asked about aliens “they should have arrived here by now, so where are they?” Leo Szilard (the physicists who first proposed the idea of a chain reaction), quipped back: “They are among us, but they call themselves Hungarians.”

Szilard was referring to the likes of Johnny von Neumann (physicists, game theory, computers), George Polya (mathematics), Paul Erdos (mathematics), and Edward Teller (father of the hydrogen bomb). It was almost like they came down from Mars, and were just explaining to us mortals what they already knew. Dr. Shi, Sekar, Williams, Chernova, and Hayes have all had an exceptional impact, just like the Hungarians.

But this post isn’t what you think. It’s not about how brilliant everyone was. It’s about how those martians are secretly human.

‍

Graduate School Evolution

Right now, many graduate students are starting their journey across the world, and it can be difficult. I’ve seen hundreds of graduate students go through their PhD years now, and there is a common evolution. I call it evolution, not phases, because you really have to go through it to evolve.

Evolution 1: I'm so excited to be here! I'm going to do great things (though play being humble while I figure it out)!
Evolution 2: Oh, wait. Others seem to get things way faster than me. Maybe I should take some more classes… (that’s where I’ve gotten positive reinforcement before!)
Evolution 3: At least two papers are rejected in a row. Reviewers suck. And secretly I don’t think my advisor really understands my work either.
Evolution 4: Everything is trivial. I’m smarter than everyone. The entire field is full of dumb ideas or microscopic ideas or industry is far ahead. Nihilism runs rampant.
Evolution 5: Perspective, and you come out the other side.

At CMU, the average CS PhD is about 6.2 years, last time I checked, and I’d guess it takes about 5 years to really get to evolution 5.

I was friends with the New Martians, and I remember they also had rejected papers, difficulty in research, and times when they made mistakes just like me.

I’ll let them tell their own stories, but I can guarantee everyone goes through a similar evolution in graduate life. I remember my time in graduate school fondly now, but I went through the evolution too.

‍

Evolution 2

In my second year, I took a theoretical cryptography class from Manuel Blum with Virgina. One homework assignment asked us to prove something about generating large primes – I’m pretty sure it was proving a variant of the Miller-Rabin primality test. I remember the difficulty proving that, the long hours, and working with Virginia because it wasn’t obvious to us. Even today I take a little comfort in knowing even a top-rate theoretician like Virginia didn’t get everything immediately.

I remember two others from that class, both of whom bragged about how smart and good at theory they were. And they were quite good and much faster than me, but Virginia didn’t do that. You don’t need to brag when you’re a Martian.

(Funny aside: One of the questions on the final exam was essentially my very own research on an RSA attack I had published at USENIX Security and recently got the test of time award. I got it wrong. Very wrong. Humiliatingly wrong. And I’m pretty sure Virginia and those two who said they were so smart got it right. I don’t know whether it was our TA Ryan Williams or Manuel being nice, but I am grateful to this day someone was a generous grader.)

I think this was Evolution 2 for me. I had done work in crypto, but everyone just seemed quicker than me. I kept wanting to take classes to catch up, but in reality that was a defense mechanism. You see, in undergrad your mastery and intelligence is validated by a course grade. In research, when things get tough, it’s tempting to revert to courses as a defense mechanism. It’s just a phase; you’ll get through it. You can never take enough classes to know everything.

‍

Evolution 3

In my third year, neither Vyas nor I were having any luck getting papers accepted. I was getting rejected in security, and Vyas wasn’t having any luck getting accepted to SIGCOMM.

Rejection does funny things to you. I had conversations that seemed to oscillate between all reviewers suck, my work was trivial, and the entire field was trivial. We all get rejected, and even have years where things don’t go our way. Even Vyas had a down year, and now he has (checks google scholar) 200+ (!) peer reviewed publications.

‍

Evolution 4

In my 4th and 5th year, I was actually getting pretty good at research and publishing. I found myself very judgemental, though. I thought of my own work as exceptional and brilliant, yet 92.3% of the papers I read were just incremental.

It’s really easy to be blind like this, where you measure others' work through what you read, but measure your own work through what’s in your head. I think it’s a little bit like when teens don’t think their parents know anything, just projected onto a graduate student timeline.

‍

Evolution 5

I remember some time later realizing how dumb I probably looked then. I had invited Dr. Manuel Blum to present some of his research in human computation at the USENIX Enigma security conference. He had this theory about humans computing passwords in their head.

I remember vividly during Q&A a cryptography researcher telling Blum he didn’t understand modern theory of cryptographic security and theories of secure passwords at all, blah blah blah. Blum has a Turing award for contributions in theory and crypto, has cryptographic numbers named after him (blum integers), and pioneered many of the definitions in cryptographic security we use today.

It was fun watching that person dig a hole, and I have to appreciate all the times I dug similar holes, around year 4 in my PhD. Luckily, Manuel was very kind to this person, and I am grateful now to all the times my peers, advisors, and professors humored me.

I will say I got one thing right in grad school: making good presentations a priority. At some point I realized I wasn’t very good, and sought out Steven Rudich because he had a reputation for giving clear, concise, and entertaining lectures. I made him a deal: I would take him out to lunch every week if he would spend that hour criticizing my job talk. I think after 3 months we only got through the first 5 slides. I learned a lot about how to think about giving a talk from him.

‍

Coming Out The Other Side

I remember a Simpsons episode where Lisa is asked whether she would rather be a big fish in a small pond, or a little fish in a big pond. She says “big fish! big fish!”.

Grad school is rough on the ego. There can be entire years where you just have to put one foot in front of the other, not because you know you’ll succeed, but because you must go on. And it’s really only useful in very specific ways. But it’s worth it.

And it’s really fun to look back and say “I knew the New Martians when they were first starting out.”

‍