Economist Writing Every Day

I want to share some changes that I’ll make to my game theory course, just for the record. It’s an intense course for students. They complete homeworks, midterm exams, they present scholarly articles to the class, and they write and present a term paper that includes many parts. Students have the potential to learn a huge amount, including those more intangible communication skills for which firms pine.

There is a great deal of freedom in the course. Students model circumstances that they choose for the homeworks, and they write the paper on a topic that they choose. The 2^nd half of the course is mathematically intensive. When I’ve got a great batch of students, they achieve amazing things. They build models, they ask questions, they work together. BUT, when the students are academically below average, the course much less fun (for them and me). We spend way more time on math and way less time on the theory and why the math works or on the applicable circumstances. All of that time spent and they still can’t perform on the mathematical assignments. To boot, their analytical production suffers because of all that low marginal product time invested in math. It’s a frustrating experience for them, for me, and for the students who are capable of more.

This year, I’m making a few changes that I want to share.

1. Minimal Understanding Quizzes: All students must complete a weekly quiz for no credit and earn beyond a threshold score in order to proceed to the homework and exams. I’m hoping to stop the coasters from getting ‘too far’ in the course without getting the basics down well enough. The quizzes must strike the balance of being hard enough that students must know the content, and easy enough that they don’t resent the requirement.

2. Rearranging: I’ve always taught Normal form first, then extensive form. Students get confident on the former, then lost on the latter. Then, they fail to study the normal form well enough for the subsequent exam. This year, I’m instead teaching ‘simultaneous’, including both normal and extensive forms, and then ‘sequential’ games, including PSNE, Subgames, & SGPNE. Maybe, getting their toes wet with the extensive form prior to the sequential games will help dilute the deluge of new terms and concepts.
3. More Time per Topic: Historically, we spend one week per topic. That’s 1 week each for: Game components, normal form, extensive form, MSNE, unobserved payoffs, and continuous strategies. Such a schedule puts the 2^nd midterm in late October and reserves the balance of the semester for other assignments. That clip might be a bit too fast. Students often end up with a confident grip on the content *after* they submit the pertinent homework assignment. So, I’m increasing the time to 1.5 weeks per topic. Hopefully, this will ameliorate some of their frustration while allowing me to move more slowly in class. The tradeoff is that homework deadlines now occur on 2 different weekdays rather than always being on the same day each week. Further, they will know content later, to the detriment of their term paper writing.
4. Better Literature Reviews: As part of the term paper, I require that students find 2 sources each of empirics, theory, and non-scholarly sources. They’re supposed to describe how each paper contributes to their term paper topic. It doesn’t go well. Students have a natural impulse to summarize entire articles – including irrelevant parts. Though they write in paragraph form, they end up with a list of paragraphs that are largely unrelated. Instead, I’m telling them to organize their paragraphs by game components. So, two authors might address the number of players, while three articles address different strategies and information. We’ll see if this helps students write more cohesively and see the connection between the papers more clearly.
5. Term Paper Concision: With a nod toward AI, I am replacing the term paper assignment grading rubric criteria of “Clarity” to “Clarity & Concision”. Social scientists are known for using a lot of words. But economists place a high value on concision. While concision was always an implicit criteria, I’m adding it in the context of encouraging students to use AI to help them write. ChatGPT is famously and unnecessarily verbose. If they lean on it too much, then it will be obvious at least because it will hurt their writing quality. Even apart from AI, I want to discourage students from including entirely correct and entirely irrelevant content in their papers. Without the concision criteria, it’s hard to deduct points for logic when what they write is true, not conflicting, while also worthless.

Those are the changes that I’m making this year. I’ll report back. I suspect that none of it is perfect – I’ll have to adjust some things on the margins. There will be hints as to whether it’s working along the way, but I won’t really know until I see the term papers. Wish me luck! (And my students too!)