Course-planning @ Brown

Updated 2/18/25 for Fall 2024

I am a student at Brown University studying computer science, so my course selection is heavily biased towards that field. This is a reflection on the courses I have taken.

Fall 2023

Context: I was an incoming freshman with a vague interest in logic, math, and linguistics. I knew I wanted to study computer science.

CSCI 0190

Every year, upperclassmen give a lot of advice about whether to take CSCI 0111, 0150, 0170, or 0190. This is not a simple choice: every intro course differs from the others in size, difficulty, and even content.¹ Difficulty is roughly 11 < 15 <= 17 < 19, and size is roughly 19 < 11 < 17 < 15. As far as I know, 15 is the only one that teaches object-oriented programming, and the rest teach functional programming.

Well, I was already pretty set on taking 19 after doing the summer placement. The first person to introduce me to programming was Zed A. Shaw, who wrote "Learn Python the Hard Way" back in 2013. His authorial voice was very serious, very pedantic, and somewhat sarcastic. I think the professor of 19 gave me the same impression. You don't really have that sort of voice unless you care very much about your teaching.

19 was a small class with a few partner projects, so I knew most of my classmates and made close friends. The professor was also—truly—the best lecturer I have ever had. I can say a lot more about this class, so if you are thinking about 19, please feel free to reach out.²

CSCI 1450

At that time, I was also impatient to get one of my math prerequisites out of the way. After some calculation, I decided that taking probability would unlock more courses than linear algebra.³ There were three courses that fulfilled the requirement: CSCI 1450, APMA 1650, or APMA 1655. I chose CSCI 1450 fairly arbitrarily, mostly on the basis that a friend of mine was taking it. I never attended lecture—this will be a recurring theme for math courses—but CSCI 1450 actually ended up being my favorite class that semester. I have very good memories of visiting office hours every week and hanging out with the TAs and my friends. Really, I spent so much effort on this class that I regret taking it S/NC.⁴

CSCI 1951x

Coming to Brown, I also had a vague idea that I wanted to learn more about logic. I think I was trying to make an (automated) theorem-prover at the time. This led me to take CSCI 1951x, which did not have any formal prerequisites. It touched on everything from program semantics and metaprogramming to the logical foundations of mathematics. I also have very good memories of camping office hours for this course.

CSCI 0120

My last course was COST 0120. I thought it might help me solve a few existential problems I had at that time. It did not help that much, but I did learn a lot about Mengzi, Xunzi, and Zhuangzi.

Reflection

So, a brief summary of what I actually did in these courses: I went to class, skipped class, did problem sets, wrote papers, and spent a lot of time talking to my TAs.

I could have, but did not, take "Introduction to Linguistics" (CLPS 0300), "Diversity of Life" (BIOL 0210), "Form and Formalism" (APMA 1920), and "Writing Graphic Memoir" (ENGL 1050U), and "Advanced Deductive Logic" (PHIL 1635). I do hold a little bit of regret for not shopping APMA 1920 and ENGL 1050U, but I was very busy with CSCI 0190 assignments at that time and missed the opportunity. Here was the full shopping list:

My takeaways from this semester were:

1. Choose the smaller course
2. Choose the more difficult course
3. Choose the course with your friend
4. Do more math because it's fiddly and fun
5. Don't S/NC courses

(I will end up disputing points 3-5 in the future, but that's a lesson for another semester.)

Spring 2024

Context: I was a second-semester freshman who knew a little more about the world. I wanted to know whether I enjoyed systems or AI more, which I thought of as two sides of an extreme. I also thought I wanted to finally try math, logic, and linguistics. I was also working at a neuroscience lab for some reason.

CSCI 0300

I always heard that CSCI 0300 is the best course ever. Well, now that I have taken it, I can become one of those people who say that CSCI 0300 is the best course ever. The professors are some of the most caring and sympathetic people around. You want that in a systems course, because systems are hard. When you're running into a segmentation fault for the nth time, you don't want hard love; you want comfort and support and understanding. I think they get that just right. Here's something I wrote about the transition from 19 to 300 a year later.

Underclassman:

for the people who've taken both 19 and 300 how did you find the latter in terms of difficulty as opposed to 19?

Me:

conceptually around the same difficulty, but 19 and 300 are almost like the antithesis of each other. some things i can point out:

• pyret is functional and C/C++ is not. it's probably easier to pick up imperative after learning functional programming, but that does mean that some things don't transfer. after a few segmentation faults, you'll understand why ideas like "immutable by default" comes about
• you will need to pick up some tooling. by tooling, i mean getting comfortable with your editor (vscode, vim, emacs) and its keyboard shortcuts and with the terminal. the pyret editor is very simple. this is fine because 19 assignments are small and usually contained within a single file. for 300 projects, you'll want to be able to quickly open multiple files at a time and also have gdb running in the terminal. i suspect that 15/17 students might be more comfortable doing that
• 300 projects have a lot of boilerplate code. you'll have to be comfortable with reading and extending code that you have not written yourself. you don't really learn this in 19 because you write all of the code and design all of the data structures yourself. (data-driven programming is amazing, it's just not that useful in 300 because of how structured assignments are)

it is because they are so different that i usually recommend everyone in 19 to take 300 afterward. learning a memory-unsafe language really concretizes how ideas in modern programming languages came about. i would also argue that the only way to really learn software development is by thinking about how a complex system works. good coding practices (which you learned in 19, and you'll learn more of in 320) comes afterward

CSCI 1470

I also took CSCI 1470 because all of my friends were taking it. I did not enjoy it as much as CSCI 0300. I think it faces a very classic problem: the professor teaching it changes every semester, so institutional knowledge does not get passed down, and the course never improves. This is different from CSCI 0190, CSCI 0300, and CSCI 1951x, which are taught by the same professor every year.

Of course, I can't just blame the course structure. I think I'm also just not interested in AI/ML. Some existential thoughts here. In fact, the only part of this course I truly enjoyed was the final project, where I spent three straight days chugging Monsters, debugging legacy Tensorflow v2 code, and spiraling between "it's joever" and "we're so back" with my teammates.

MATH 0540

Abiding by the lessons I learned last semester, I chose MATH 0540 with the thought that it was (1) a small class, (2) a difficult class, and (3) a class with my friend and (4) a math class.

My other option was MATH 0520, which was on more applied linear algebra. It had weekly quizzes. No thanks.

PHIL 1665

MATH 0540 probably would have been a great choice, if I didn't also decide to take PHIL 1665 with another friend.

There were so many problem sets.

Like so many. It was too much math.

I also never truly understood the modal ideas of necessity and possibility. And it became especially hairy when they added in first-order logic, like forall and exists. PHIL 1665 spun my brain around well enough that I'm afraid of taking another analytic philosophy class again.

CLPS 0800

I also took CLPS 0800. It was a great course with a professor, but I realized that I wasn't that passionate about psycholinguistics or clinical research.

Reflection

A brief summary of what I actually did in these courses: I went to class, skipped class, did problem sets, did more problem sets, did so many problem sets, and did not have the time to go to TA hours. There are no achievements for this semester, because I was in the trenches.

My takeaways from this semester were:

1. There is a limit to how much math you do before you start to miss coding. That limit is 16 hours.
2. You should not take too many math classes in a semester.
3. Not lying to yourself about what you like and dislike is very important. While I had suspected that I was not that passionate about linguistics, AI, or philosophical logic, I did not anticipate how burdensome those courses would be.
4. I can work at least 40 hours per week on coursework, but I won't enjoy it, I won't learn as much, and there will be the threat of burnout lurking in the background.
5. The difference between magic and boringness is possibly not the content, but the professor.

But now that I look back on it, I didn't have any particularly good alternatives either. I'll just take a better workload next semester.

Fall 2024

Context: I spent the summer working in a psycholinguistics lab and finally admitted that computer science was the only thing for me. My professor encouraged me to try computer science research.

CSCI 1680

When anyone asks me for course recommendations, I always tell them to take CSCI 0190 first and then CSCI 0300. But my favorite course at Brown is most definitely CSCI 1680.

1. There is no stencil code. You write everything from scratch. You know the codebase intimately. Your design decisions matter.
2. You can choose to do the project in C, C++, or Rust. This course made Rust finally click for me.
3. The two major projects, IP and TCP, are built on top of each other. This encourages you to write clean and maintainable code.
4. IP, TCP, and the final project are also partner projects! You learn to communicate with your partner, and you learn to read and extend code that you have not written yourself.

The workload is fairly high, and I would recommend clearing out your schedule for it, especially if you decide to not do the projects in Go. Here was my breakdown, including time spent watching gearups:

1. Snowcast: 36.5 hours
2. IP: 38 hours (personal), ~60 hours (group)
3. TCP: 52 hours (personal), ~100 hours (group)

CSCI 2952R

I also took CSCI 2952R which is a course on systems research and scientific communication. The specific focus was on systems that could improve other systems. It was quite good; I learned about the entire paper-writing pipeline and heard about tons of cool systems topics.

CSCI 1970

My independent study course, CSCI 1970, was also tied to the research project I was doing in CSCI 2952R. I was in charge of writing the Lean code to prove some correctness properties about automatically synthesized aggregators.

MATH 1530

I wasn't originally planning on taking MATH 1530, but (1) I really enjoyed the first lecture by one of the professors, and (2) I thought I should probably learn about abstract algebra at some point.

Unfortunately, my friends were in another section with another professor, so I switched over. And then they ditched me to take Computer Graphics instead! (This is the last time I'm blindly following my friends with course choices. It never works out.)

I found MATH 1530 to be challenging and existentially abstract, so I think it's going to be a while before I take another pure math course.

EGYT 1310

Finally, I took EGYT 1310. I remember that Red from Overly Sarcastic Productions once said that she took a course on Egyptian hieroglyphs in college. Also, I saw a news article on one of the department monitors about an alumn who worked on deciphering dead languages. It seems like that was enough subliminal advertising for me to want to take this course. It was quite interesting, would recommend.

Reflection

A brief summary of what I actually did in these courses: I spent an upwards of 40 hours/week coding and somehow managed to fit in other things in-between. Then I dealt with my CS-burnout during winter break.

My takeaways from this semester were:

• I like systems.
• No more pure math.
• Research is quite stressful, no matter the field.

In high school, I was very curious about a lot of things. I still believe that there is a high marginal benefit to dipping your toes into a wide range of fields and taking away the most important ideas and models.

But the fact of the matter, at least at Brown, is that there are no longer any survey courses. I checked. We don't have a World History course; it has got to be something like "The Age of Revolutions, 1760-1824" or "Japan's Pacific War: 1937-1945." These are good courses if you really care about the topic, but they are not very suitable for the casual passerby.

To be clear, I think Brown's Open Curriculum is very good at combatting this phenomenon. I have never taken a course that I wasn't the slightest interested in. But I do find my scope of interests narrowing quite a bit, despite my best efforts. Maybe because my first impression of these topics come from drinking from the water hose, I am not that curious about math, philosophy, or linguistics anymore.

Maybe that's just what happens when you get older. You are less confused, and your preferences become clear to you.

Spring 2025

I am still in my Spring 2025 semester, so expected changes to this section later.

Unlike previous semesters, I wasn't sure about my schedule at all. Reflecting back on my decisions in the past, there were usually very shallow reasons for taking or not taking a course. I would take course A because my friend was taking it or I would reject course B because it was held in a dusty classroom. Well, this is even more true this semester. All I learned to do was ruthlessly prune classes that I wasn't enjoying at the start of shopping period.

CSCI 1670/1690

I took CSCI 1670/1690 because all of my friends were taking it. Plus, CSCI 1670 is taught in C, and I was TAing CSCI 0300, which is in C, which made me feel like there's a good kind of synergy there. (I told you, the reasoning was not very sophisticated.) This whole thing was not a easy choice for me though, because I knew that if I waited two more years, it would be taught by a really great professor.

CSCI 1515

I decided to take CSCI 1515 two days before the end of shopping period, without having been to a single lecture. I was originally planning to take CHIN 1010 ("Stories from the Chinese Empire: Scholars, Demons and Swindlers"). But after shopping it for two weeks, I realized that it was probably too advanced and stressful for me, since I would have had to prepare for hours before lecture to understand the texts.

After scrambling to find an available CS course—I also tried to get into CSCI 1650 ("Computer Security") and 1951A ("Data Science")—I finally remembered that CSCI 1515 had very good reviews. Cryptography was never quite on my radar, and C++ is not one of my favorite languages, but I'm quite happy about my choice. The professor puts a lot of care into making sure we understand lecture, and her passion is infectious.

ENGN 0520

I really did not expect to take ENGN 0520. I only decided to shop it because my friend was taking it, and I had a free slot in my schedule. But the first lecture was very good, and it reignited that curiosity I also had toward electronics and hardware. I have had some experience with it in my high school physics class, but I never felt like I really knew how circuits worked. I also thought ENGN 0520 could also solve some existential problems I have about "what is real" and whatnot. It has been going well!

I have had to learn how to do applied math again. There are actually a lot of practical considerations, like:

1. When should you round?
2. Where should you round to?
3. When should you include units?
4. How do you show your work?

In my previous math classes, we just had to write words.

(The lab part is going great too, but I am totally relying on my partner to handle the oscilloscope.)

EGYT 1320

Finally, I continued with Middle Egyptian in EGYT 1320. There is a lot more grammar and vocabulary to learn than I expected, so I'm glad I'm not taking it alongside CHIN 1010.

Thoughts from other people

1. College advice for people who are exactly like me is not about choosing courses, but about all of the other things you can do in college when you're not choosing courses.
2. Bluebooking for happiness argues that only the professor matters.