DCS 2350/CSCI 2350:


Social and Economic Networks


Fall 2023


What is this course about?


Networks: A modern way of looking at, reasoning about, and making sense of the world.

Example: What does a Tunisian fruit seller's self-immolation have anything to do with the Saudi-Iran proxy war in Yemen?

Figure: Mapping the Arab Spring.

Example: Why did Covid impact the northeastern part of the U.S. or countries like Italy so much?

What do these examples have in common? Additionally, how are they even remotely related to things like the U.S. residency matching or kidney exchange programs?

This course examines the social and economic aspects of today's connected world from a multitude of perspectives; namely, network science, computer science, sociology, and economics. The fundamental questions to be addressed are: What are the properties of real-world networks? What are the effects of networks on our behavioral choices like quitting smoking or eating healthy? How do cascades in networks lead to outcomes like videos going viral? How does Google search the Internet and make money doing so? Debates issues around centrality in networks. Uses game theory to study strategic interactions in networks and markets.

Prerequisites:

What is the central theme?

The focus of this course is networks. Graph theory as a mathematical tool for studying networks has been around for over two centuries. However, modern computational power has enabled the study of the types of networks that not only change dynamically but also scale to a previously unimaginable size. Examples of such networks are today's social and economic networks. This course connects a series of questions within the central theme of social and economic networks. (The specific questions are listed in the course schedule.)


Figure: A large-scale network of political blogs. Colors denote political leaning. (Adamic and Glance, LinkKDD 2005.)

How will these questions be addressed?

A variety of inter-disciplinary tools and techniques will be used to address these questions. For example, we will use a software called Gephi to visualize networks and study their properties. We will study the classical as well as the modern theories of network formation, and we will experiment with these theories in the NetLogo software environment. We will use tools from mathematical sociology as well as computer science to study diffusion in networks. We will get an introduction to computational game theory in order to address various interesting questions on the strategic aspects of networks.

Most importantly, elements of computational thinking will be prevalent throughout this course to answer many of the "how" questions. The reason is that in today's socio-economic context, nearly all real-world networks are very complex with many interdependent components. To answer any conceivable question within such complex settings, we need to think critically about devising a solution. We need to consider computational time and space. We need to combine a mathematical approach with an engineering approach. Computational thinking enables us to do the above in order to devise practical solutions to problems. Programming is just one part of computational thinking; it's not all.

Why are networks the way they are?

Several critical questions will be addressed in this course. First, almost all real-world networks exhibit some common properties, such as a "giant component" and a certain type of "degree distribution." Why do the entities in the network connect in this fashion? Second, what are the effects of network connectivity on various social and behavioral phenomena, such as smoking, obesity, or even videos going viral? Finally, what roles do strategic interactions play in shaping the networks that we see today?

Instructors

Professor: Mohammad T. Irfan
Email: mirfan@bowdoin.edu
Office Hrs:
TW 3-5:30 PM
Office room: Mills 209
LA: Charlotte Gehrs
LA Hours: Thu 7-9pm

Room: Mills 105

Time & Place

Lectures: TR 10:05-11:30 PM
Midterm: In-class (10/19)
No final exam

Course Outline

Learning Objectives:

Even after 5 years:

  • Know:
    • How networks are modeled, how they look like in the real world , and how they evolve over time
    • Basics of social network analysis: Small-world phenomenon, clustering, degree distribution, centrality in networks
    • Algorithms like PageRank and web-search auctions running on the Internet
  • Be able to do:
    • Implement network models using computer programs
    • Study the effects of networks, with examples ranging from contagion to smoking and obesity to online videos going viral: How would the effect vary if the network structure were different?
  • Find value in:
    • Networks as an essential DCS tool to better understand today's intricately connected world

What type of course is it?

  • This is a lecture-based course.
  • This is not a projects course. Nor is it a discussion course, even though it will be very much conversational in style.
  • Readings are due after the class, unless otherwise announced.
  • There are some additional notes and video clips that will not be covered in class. However, you are required to utilize these resources to reinforce the key concepts.

Textbooks:

[EK] Networks, Crowds, and Markets, by Easley and Kleinberg (Required)
[Glad] The Tipping Point: How Little Things Can Make a Big Difference, by Gladwell (Required)
[Wat] Six Degrees: The Science of a Connected Age, by Watts (Optional)
[Jack] Social and Economic Networks, by Matthew Jackson (Optional)
[*] Individual chapters from other books (to be provided as hand-outs)

Software:

The following free software will be required. Please install them on your computer.
[1] Gephi

[2] NetLogo

[3] Visual Studio Code with Python 3

Evaluation:

  • 10% Points: One in-class midterm exam

  • 15% Points: Final paper

  • 50% Points: Summative Assessment: Problem solving and programming assignments

  • 25% Points: Formative assessment and class participation: This includes formative assignments, attendance, and active class participation. Taking handwritten class notes is highly recommended for its many benefits. Students will report the status of their note taking.

  • Points to letter grade conversion: 94% A, 90% A-, 85% B+, 80% B, 75% B-, 70% C+, 65% C, 60% C-, 55% D, below 55% F

Late Policy:

  • Late submissions will be accepted up until 24 hours after the deadline with a penalty. Each hour after the deadline would cost 4% of the points for that assignment. Extensions will be given if there is any health or other emergency situation.

Outline of Topics:

Week Question Topics Readings Additional Notes/Links
Week 1
(8/31)
 Course introduction: why study networks? Slides (Intro)
Week 2
(9/5, 9/7)
 What is a network? Basics of graph theory [EK] Ch 2 Slides (continued)
Week 3
(9/12, 9/14)
 What does a real-world network look like? What are some common properties?
 1. Macro-level properties of a real-world social network:
giant component, small-word effect, clustering, degree distribution
2. Micro-level properties: centrality
3. Experiment with Gephi
[EK] Ch 2
[Jack] Ch 2
 1. Slides (Properties)
2. 3.5 degrees of separation in Facebook
3. Power law debate
4. Barabasi's response
5. Petter Holme's blog post on it
Week 4
(9/19, 9/21)
 What does a real-world network look like? (continued)
1. Degree distribution (cont.)
2. Centrality in networks
Assignments 1 and 2 (experiments and theory) out 		[Jack] Ch 2
 Slides (continued)
Week 5
(9/26, 9/28)
 1. What's the effect of different types of edges?
2. How can we use it to detect communities in a network?
1. The strength of weak ties
2. Community detection algorithms 		[EK] Ch 3
Granovetter's paper
1. Slides (The strength of weak ties)
2. Note on Girvan-Newman algorithm
Week 6
(10/3, 10/5)
 1. Are the nodes connected to each other kind of the same?
2. What can we say about a network having friends as well as enemies?
1. Homophily
2. Structural balance: positive and negative relationships
[EK] Ch 4, 5
 1. Slides: Homophily
2. Slides: Structural balance
Fall Break        
Weeks 7 & 8
(10/12, 10/17, 10/19)
 Modeling Networks: How can we model network formation in the real world? And why should we model?

 Network Formation:
Random-graph models and their properties
Midterm exam on 10/19
 [Jack] Ch 4, 5
[Wat] Ch 3, 4
1. Slides (Models of network formation)
2. Prof. Irfan's NetLogo tool to study Erdos-Renyi random graphs
Week 9
(10/24, 10/26)
 Modeling Networks (continued) 1. Watts-Strogatz and preferential-attachment models
2. NetworkX: network programming in Python
Assignment 3 and 4 (modeling networks: theory and experiments) out 		NetworkX tutorial by Prof. Irfan
Week 10
(10/31, 11/2) 		What is the effect of a network as a whole on people's behavior? For example, how does a video get viral in a social network? Similarly, how did the Facebook post "Hi World, we want a puppy!" get over a million likes so quickly?
Diffusion in networks
Assignment 5 (programming: modeling networks, diffusion) out 		[EK] Ch 19
 1. Slides (diffusion)
2. One person starts a dance party
3. Kleinberg's talk
4. Interview of Duncan Watts by Fast Company
Week 11
(11/7, 11/9)
 We've talked about structures of networks and the connection between network structures and various social phenomena. Can we say something about economic networks where the interactions are primarily strategic? How can we model such strategic interactions? Game theory [EK] Ch 6
 1. 1. Slides (Game theory)
2. Game theory on display: USA's information warfare (on the verge of Russia-Ukraine war)
3. Game theory on display: Russia's war of attrition
4. A Beautiful Mind: How it missed Nash equilibrium
Week 12
(11/14, 11/16)
 How can we model one very common type of strategic interaction that we see everyday--auctions?
 Auction
[EK] Ch 9
 Slides (Auctions)
Week 13
(11/21)
 How does Google and other search engines use auction to make money?
 Sponsored search markets
Assignment 6 (game theory, auction, and sponsored search markets) out 		[EK] Ch 15
 1. Slides (Sponsored search markets)
2. Prof. Irfan's video lecture
3. Greg Taylor (Oxford) on the economics of search engines
Thanksgiving Break        
Week 14
(11/28, 11/30)
 Matching: Another example of strategic interactions that received many plaudits due to its humanitarian cause is the kidney exchange program. What is the networked economy foundation of such a program?
 Stable marriage and matching markets
Assignment 7 (matching) out 		1. Stable marriage handout
2. [EK] Ch 10
 Slides (Matching)
Week 15
(12/5, 12/7)
 Let's talk about one special network that we use everyday--the Internet. What does it look like? How does Google search it? WWW: structure
Link analysis and web search--PageRank
 [EK] Ch 13, 14
 1. Slides (Information Networks)
2. Vannevar Bush's Memex
3. Google's Matt Cutts on PageRank


Collaboration Policy:

Assignments will normally be done individually. Unless clearly marked as a group assignment, each assignment is an individual assignment by default. Students are expected to follow Bowdoin's Academic Honor Code.

Following is the collaboration policy for individual assignments. You are encouraged to discuss ideas and techniques broadly with your classmates, but not specifics of assigned problems. Discussions should be limited to questions that can be asked and answered without using any written medium (e.g., pencil and paper or email). This means that at no time should a student read anything written by another student. Violation of this policy is grounds for me to initiate an action that would be filed with the Dean's office and would come before the J Board. If you have any questions about this policy, PLEASE do not hesitate to contact me. This will be a zero-tolerance policy.

It is permissible to use software and materials available from other sources (understanding that you get no credit for using the work of others on those parts of your assignment) as long as: 1) You acknowledge explicitly which aspects of your assignment were taken from other sources and what those sources are. 2) The materials are freely and legally available. 3) The material was not created by a student at Bowdoin as part of this course this year or in prior years. To be absolutely clear, if you turn in someone else's (human or AI) work you will not receive credit for it - on the other hand, if you acknowledge it, at least you will not go to the J Board.

All write-ups, code, reviews, documentation, and other written material must be original and may not be derived from other sources, including AI tools.

If in the future you provide your work to other students, this will also constitute a violation of Bowdoin's honor code.

If Winter comes, can Spring be far behind? -- Percy Bysshe Shelley