Was grateful to share at an #ASSA2021 session today a bit on what I've learned in teaching an undergraduate course on the Economics of Networks.
A short thread to serve as a focal point for any follow-up conversation.
1/
A short thread to serve as a focal point for any follow-up conversation.
1/
What networks is about (very rough and probably somewhat idiosyncratic description)
2/
2/
I taught several variants of an undergraduate elective on this exciting and growing area. It was at the applied math/econ/CS intersection -- sometimes cross-listed, sometimes just economics but open to (and taken by) applied math, CS, other students.
3/
3/
Here is my pitch (from a variety of perspectives) for the course.
On the "strategy" point for econ depts: "big data" and CS courses often appeal to students interested in networks of various kinds.
Economics has some big ideas to contribute - but we have to make the pitch!
4/
On the "strategy" point for econ depts: "big data" and CS courses often appeal to students interested in networks of various kinds.
Economics has some big ideas to contribute - but we have to make the pitch!
4/
I showed a few example slides - here's one from the introductory lecture.
It might be old hat to us, but the idea that the growth of a viral product and the growth of the 🦠 are described by closely related differential equations is a big a-ha if you haven't seen it before!
5/
It might be old hat to us, but the idea that the growth of a viral product and the growth of the 🦠 are described by closely related differential equations is a big a-ha if you haven't seen it before!
5/
As one motivating example, I make the claim that the different phases of the life of a new product/platform can be divided into "infancy," "adolescence," and "maturity" and that different kinds of modeling approaches are suggested by thinking about each.
6/
6/
Materials: On the analog side, my core texts are the left two and the optional one on the right can fill out the picture for more technical lectures/questions/exercises.
Easley and Kleinberg is free online and very affordable in hardcover.
cs.cornell.edu/home/kleinber/…
7/
Easley and Kleinberg is free online and very affordable in hardcover.
cs.cornell.edu/home/kleinber/…
7/
One of my favorite things about teaching this course is getting to use the EK book. Most "innovative" courses don't get a ready-made undergraduate textbook to rely on. Here we have one.
A bit miraculously, it's written without calculus as a prereq!
8/
A bit miraculously, it's written without calculus as a prereq!
8/
On the other hand, most professional researchers who work on networks will learn at least one genuinely new thing from reading their technical appendices. For instance, check this out on coalescent processes (e.g., tracing back COVID strains to common ancestors)
9/
9/
Anyway, students find a textbook very comforting as a "hard source" for the material.
I assign complementary chapters of @JacksonmMatt's (general audience) book The Human Network to connect the technical material in the book to contemporary debates, lively examples, etc.
10/
I assign complementary chapters of @JacksonmMatt's (general audience) book The Human Network to connect the technical material in the book to contemporary debates, lively examples, etc.
10/
Connecting problem sets and writing assignments to this reading has been important to make sure all the materials really play an integral role!
Another huge bonus is that there are awesome online materials paired with the course material from the above books.
11/
Another huge bonus is that there are awesome online materials paired with the course material from the above books.
11/
I use Cornell's videos, available on EdX and YouTube, for INFO 2040 X (the course corresponding to Easley and Kleinberg's book).
They serve as great lectures for important units, especially effective in a flipped classroom format.
courses.edx.org/courses/Cornel…
12/
They serve as great lectures for important units, especially effective in a flipped classroom format.
courses.edx.org/courses/Cornel…
12/
And I use Matt Jackson's Coursera course (also free to access the materials) for lectures especially on more advanced or technical material such as centrality theory.
13/
coursera.org/learn/social-e…
13/
coursera.org/learn/social-e…
The topics vary by semester, but some of the greatest hits I try to cover almost every time include the following.
14/
14/
Here is a (high-level) syllabus of some of the materials that back up these topics.
15/
15/
A challenge is that the background knowledge required and the techniques used to analyze the models are NOT quite part of the standard canon of undergrad economics or CS.
But it grows naturally out of both. So it's an opportunity to teach a cool bundle of methods.
16/
But it grows naturally out of both. So it's an opportunity to teach a cool bundle of methods.
16/
For example, the pictures you use to think through the extent of diffusion/the basic branching process ....
17/
17/
... is closely related to the picture you draw to think through best-responses in a standard platform model!
This is just one example of a certain unity of ideas throughout the course.
18/
This is just one example of a certain unity of ideas throughout the course.
18/
So though networks doesn't fit neatly in the standard *econ* canon of methods that undergraduates already know, it does have a canon that's fun to teach to them and gives the course a lot of technical meat.
19/
19/
Teaching those methods is a challenge -- you can't count on most of the important background having been taught in the prereqs. But also fun and gives the course a lot of value added relative to other classes.
20/
20/
One caution: it's good to make clear to the students from the start that there are a lot of new methods. They don't need a ton of math KNOWLEDGE, but they need to be mathematically mature and eager to learn new and challenging modeling ideas.
21/
21/
A "weird" first problem set with several problems using quite different ideas (probability, basic differential equations, basic optimization) can help signal to students that it WON'T just be turning one crank. Helps for matching with the right students!
22/
22/
So those are intrinsic challenges of the material. But it's all a lot less challenging than building a course on a research area from scratch.
EK is a great and fairly complete undergrad text, with excellent supplemental material and back-of-chapter problems.
23/
EK is a great and fairly complete undergrad text, with excellent supplemental material and back-of-chapter problems.
23/
And for a lot of these topics (e.g., contagion models), people have developed amazing notes, videos, and exercises, since they come up in applied math, probability, engineering, etc.
So it's not a standardized canon but a rich set of things to draw on.
24/
So it's not a standardized canon but a rich set of things to draw on.
24/
To close, I'll share some of my teaching materials. The best-developed ones are problem sets and some suggested presentation topics. These can be grabbed from my website. Solutions are available to instructors by request.
25/
bengolub.net/teaching/
25/
bengolub.net/teaching/
I don't have slides I'm happy with -- I prefer to teach from the board -- but I hope the materials from the EdX and Coursera courses above fill some of that gap!
26/26
26/26
PS/ I am very grateful to Wendy Stock for organizing!
It was fun to share a session with @d_f_stone and the other speakers, whose talks I really enjoyed! Sorry if I missed any co-participants also on Twitter.
Recording here: ativsoftware.com/appinfo.php?pa… #ASSA2021
It was fun to share a session with @d_f_stone and the other speakers, whose talks I really enjoyed! Sorry if I missed any co-participants also on Twitter.
Recording here: ativsoftware.com/appinfo.php?pa… #ASSA2021
• • •
Missing some Tweet in this thread? You can try to
force a refresh
