Disclaimer: The words below are purely opinions of the author. They are not representative of any institutions the author may belong in.
Background
Who am I?
I am Rizhao, a Year 3 Computer Science (CS) student in NUS. I also was a teaching assistant (TA) in AY21/22 Semester 1 for CS2102 Database Systems and CS2106 Operating Systems.
In my first year in NUS, I was an industrial design student where I learnt how to be consistently creative.
What’s this article for?
Times has changed. Zoom classrooms has almost fully replaced all in-person lessons. I for example, didn’t have a single in-person lesson or exam last semester.
While the mode of education has changed, the mindset hasn’t. We are still stuck in old systems from pre-Covid times. This article will propose a few ideas (some crazy, some simple) that hopefully will trigger rethinking cycles of what education could be.
Ideas
The Teaching Assistant (TA)
Teaching assistants are arguably the pillar of NUS CS. TAs are what make NUS CS education scalable. My estimate is for every 20 students enrolled in NUS CS, one would become a TA.

Who are the TAs?
TAs in NUS CS largely comprise of undergraduate students who did well in the module in previous semesters. Another portion of TAs are Masters or PHD students.
From my experience, undergraduate TAs perform much better than graduate TAs. Graduate TAs typically have less experience in the modules they are teaching since most haven’t gone through it before.
Why do people become TAs?
Money
TAs get paid $40/hr. Not a bad sum for an undergraduate. Most new TAs start teaching because they are enticed by the pay. However, they would soon realise they don’t get paid for every hour they work (This point will be elaborated later). Money hence becomes less of a factor for those who have taught before.
Portfolio
For undergraduates who haven’t had any internships, TA experience is something they can show off proudly in their resumés. Being hired as a TA is validation that you understand the content.
Passion
Some just love to teach! Sharing knowledge brings joy. TAs who are motivated by passion usually perform the best. They spent a ton of effort preparing for their lessons. Some arguably teach better than the professor.
Proposal: Performance-dependent TA Pay
$40/hr sounds like a lot right?
Not quite, because lesson preparation time typically aren’t accounted into working hours. I for example, spent 4 hours on average preparing for CS2102 tutorials. I taught 9 tutorials, so that’s a total of 36 unpaid hours!
Some modules also don’t pay for consultation hours. I spent an average of 2 hours per week answering questions on Telegram and Zoom for CS2102. Over 10 weeks, that’s 20 unpaid hours.
Considering I had around 40 paid hours for tutorial time, that means 60 / (40 + 56 + 20) * 100% = 51%
of my time weren’t paid for.
How are TAs paid then?
At the end of the semester, the professor usually determines a constant number (n) of hours for all the TAs under their charge. TAs are then paid n * $40.
Different professors have different pay policy. Some allow you to add a few consultation hours, while others don’t.
The current TA pay system is flawed on 2 aspects
Opaque
I wouldn’t know exactly how much money I would make when I take up the job, since I wouldn’t know what would be considered a working hour.
Wrong incentives
Given the current system, an “economically rational” TA would spend minimal effort on lesson preparation and consultation since they usually aren’t paid for. As you can reasonably guess, a system with wrong incentives results in poor teaching quality.
How should TAs be paid?
I propose TAs be paid a flat amount of money (e.g. $2000 for 1 module) multiplied by their performance rating.
Right incentives
TAs already are given a performance rating by student at the end of the semester. This rating can be bell-curved against the entire CS department to derive a multiplier value
Example:
5th percentile - 0.5X
25th percentile - 0.8X
50th percentile - 1X
75th percentile - 1.5X
99th percentile - 2X
With their pay on the line, TAs would be motivated to put in more effort in teaching. It is also fair to genuinely passionate folks like my friend Christian, who puts in around 8 hours to prepare for each tutorial.


Edit: There are some suggestions on Reddit about giving bonuses instead of a multiplier. I believe that is viable too.
Transparency
We should all know how much money we are making before we take up a job right?
Stating a flat pay on the contract prevents any misunderstanding on how much money one would be earning.
An estimate of how many hours would be required should also be stated on the contract. The number should be derived from the average of a sample of the actual hours put in by TAs for the module in the previous semester.
Why not just pay by hour but account for lesson preparation and consultation time?
There’s no way to verify how many hours a TA put into lesson preparation and consultation. Someone could simply say they put in 100 hours and bankrupt the whole system.
Since these 2 factors have a huge impact on the TA’s performance rating, using the performance rating would ensure that these 2 factors are accounted for.
When TA pay is benchmarked against the teaching quality, students get more motivated TAs and a better education overall.
Proposal: TA union
There’s a review system for TAs but why is there no review system for the system?
TAs should band together to file proposals for how the compensation systems for TAs could be improved.
Feedback is the key to change. The TA union could serve as a platform to bargain for fairer compensation.
Beyond that, having a community of TA helps to reduce the stresses that come from being a TA. During my first class, I almost got drowned out by my imposter syndrome.
Having a support network ensures that TAs don’t feel lonely. It can also be a platform to share innovative pedagogical approaches.
Edit: There are some comments about unions being a little extreme. My idea of a union is more like a support group. Something docile, like NUSSU.
Spend less time marking exams
Marking exams is the activity with the lowest return-on-investment (ROI), especially for finals.
After the final exam is marked, students typically don’t get back their papers or get any comments. Even if they do, few would bother checking it out, since their marks are already locked in. Marking exams hence have the lowest ROI in terms of the student’s learning.
Proposal: Automate the marking
Exams should be designed so that they can be marked automatically (MCQs over open-ended questions). Every hour spent marking is one less hour the professor has to teach students.
What about just removing exams?
After years of conditioning in the Singapore education system, exams are still the biggest motivation for students to study.
Without exams, no one would bother revising the content.
Won’t the quality of the exam drop?
Does the quality even matter? I think as long as the MCQs are set hard enough such that the students still need to revise, the goal of learning is met.
Factor in socialising
I think it’s a safe bet that there are at least 10% of CS undergraduates from the Covid batch (Year 1 and 2 in AY21/22) who had made less than 5 friends in their student career.
Tutorial lessons used to be a great way to make friends in pre-Covid times, but it’s no longer the case.
CS students are a shy bunch. No one turns on their cameras during online tutorials (I’m not lying).
It’s hard to make friends when your classmates are muted dark screens.

Proposal: 5-students Telegram groups
We need to intentionally create spaces for people to make friends.
If you are a TA, I propose dividing your students into groups of 4-5. Add them into a Telegram group, then leave the group so that they are free to discuss whatever they like.
Additional options
Location-based clustering
Place students who are located close to each other in Singapore together
Increases chances of in-person meetups
Balanced gender ratio
Why would it be useful?
New friends, support network (Especially in HARD modules like CS2106)
Help each other in tutorials & assignments
While some may think this would encourage plagiarism, I don’t see what’s the issue since students could plagiarise even if they don’t have such groups
For you own safety, you should add a disclaimer that such groups should not be avenues for plagiarism
You should do a survey during the semester to see if they have made new friends.
Distributed campus
I stay at Khatib. Going to NUS (Clementi) takes a 3-hour round trip by MRT or a $50 1-hour round trip by Grab. Heading to campus just doesn’t make any economic sense.
I have a radical proposal that NUS should create spaces in heartlands around Singapore for NUS students to study and mingle.
The Tzu Chi Humanistic Youth Centre in Yishun is an example to emulate.
Small-sized in-person tutorial classes could even be conducted in these spaces. This lowers the cost for everyone, and would help raise in-person tutorial attendance rates.(Make more friends too!)
Isn’t this going to be expensive?
While I’m not well-versed in Singapore real estate prices, but I would assume a plot of land in Yishun or Jurong cost less than the prime Clementi land NUS campus is sitting on.
Building a space like Tzu Chi HYC with a few study tables and chairs shouldn’t cost a fortune too!
That’s all folks!
I hope this article triggers you to rethink what education could be.
Talk to me if you want to.
I am reachable by
Email: rizhaow@gmail.com
Telegram: t.me/rizhaow (Preferred)
WhatsApp: wa.me/6582186566
Linkedin: linkedin.com/in/rizhaow/
Brilliant