Networks II: Market Design
Information Science 4220/Computer Science 4852/Economics 3825
Cornell University, Spring 2018
Networks II: Market Design builds on its prerequisite course, continuing to examine how computing, economic and sociological worlds are connected, and how abstract mathematical models and analysis can inform their design. In this course, we will construct mathematical models for and analyze networked settings, allowing us to both make predictions about behavior in such systems, as well as reason about how to design such systems to exhibit some desirable behavior. Throughout, we will draw on real-world applications such as kidney exchange, college admissions and reputations systems in online marketplaces that illustrate these phenomena.
Answers to FAQs this time of the year:
Note: I am typically unable to respond to individual email due to a chronic medical condition. Once classes start, asking questions in-person, either in or after class or in office hours will be the best way to communicate with me. Until then, this webpage will contain all currently available information about the course.
Note 2 : This FAQ may be updated to reflect any new frequently asked questions (to which I have answers), so please check back if you ask me a question not addressed here. (And if there isn't an update, it probably means I don't have an answer yet!)
- For questions regarding prerequisites, please see the FAQs under Prerequisites below.
- There will be a waitlist for this class this year, in the event that we are able to increase enrollment capacity. Please note: I am not able to predict the probability of this event nor respond to email inquiries about further details on this (as any other email inquiries).
- The waitlist will be managed by the IS department administration. For any questions or issues regarding the waitlist not answered in this FAQ, please email Janeen Orr (email@example.com). The contact person may change in the January add-drop period; please check this webpage for any updates.
- Note carefully that as with all other waitlists and all other classes, being on the waitlist does not guarantee a spot in the class.
- We are unable to answer questions regarding what your chances of getting into the class, based on your number on the waitlist, are (or any other relatives of this question); any emails to Janeen with such inquiries will not be answered.
- The waitlist will be populated and processed on a first-come-first-serve basis (no exceptions) until the start of class.
- There is typically a fair amount of flux soon after class begins, and that leads to seats opening up (sadly, I am unable to predict the extent to which this will happen this year (or any year!)). At this point, we will switch from waitlist to add-drop, to allow for a quicker uptake of freed-up seats (if any). (This is because processing the waitlist at this point can lead to significant delays, if students high on the waitlist have decided on alternate courses and do not respond rapidly to decline.)
- Finally, Networks II will not be offered for technical writing any more; no special exceptions!
Instructor: Arpita Ghosh, 206 Gates Hall
Note (again): I am typically unable to respond to email due to a chronic medical condition. Asking questions in-class, after class, or in office hours (after the course starts) is the best way to communicate with me.
- Networks (INFO2040)
- Familiarity with logical reasoning (at the level of CS 2800 or equivalent),
and basic probability and statistics.
- I haven't taken Networks. Can I still take Networks II?
If you have adequate background in game theory (and have the second prerequisite), you should be able to handle Networks II with the following preparatory self-study: Chapters 6, 10, 22 in Networks, Crowds and Markets. That being said, you will definitely benefit more from this class if you take it after taking Networks, so you should do that if at all possible.
- I don't have the second prerequisite. Can I still take this class?
You can legally take this class, of course, but you might not enjoy it: some familiarity with (or at least aptitude for) simple logical reasoning and basic probability, although not associated with any particular class, is essential for Networks II.
Outline of Topics
Information and networked behavior
- Why matching markets without money?
- Kidney exchange, college admissions, school choice
- Matching markets with versus without money
- One-sided matching markets without money
- Binary preferences
- Perfect matching; Hall's theorem
- Rank-order preferences
- Pareto efficiency and strategy proofness
- No initial endowments ("House Allocation"): Serial dictatorship
- Initial endowments: Core allocations; Gale's Top Trading Cycles (TTC) Algorithm
- Application: Kidney exchange
- Markets with two-sided preferences
- The marriage model; stable matchings
- The Gale-Shapley algorithm and its properties
- Many-to-one matching
- Incentives and preference reporting
- Application: College admissions (NRMP hospital-intern match)
- Information asymmetry and inefficiency in markets
- Adverse selection: Inefficiency; credible disclosure, signaling
- Moral hazard: A prisoner's dilemma model; repeated games
- Alleviating information asymmetry: Reputations
- Quality uncertainty on the Web
- Online ratings and reputation systems
- Application: An empirical study and redesign of the eBay reputation system
There is no textbook for this class, as there is no single book containing all the topics we will cover in this class. However, readings for each topic will be posted on CMS as they become relevant through the semester.
Note that these readings will not substitute for attending class, though, since the material in the lectures will typically at a different level of detail than the readings: the readings should be used as a supplement to, rather than a substitute for, the lectures.
To supplement reading material, the lecture slides will also be made available on CMS from after each lecture. Note, however, that some examples will be worked on the whiteboard in class and these will not make their way into the lecture slides. This means that again, the posted lecture slides are not a substitute for attending class.
Whether you choose to attend lectures or not is entirely up to you, given the information you have about the course structure. However, please understand that (a) you are responsible for all material that is covered, discussed, or announced in class regardless of whether you choose to come to class or not, (b) office hours and Piazza are not a substitute for skipped lectures, and (c) all this being said, please come to class only if you are sure you can stay focused enough so that your in-class behavior does not detract from the learning experience of your classmates. Students doing otherwise will be asked, politely but firmly, to leave the class.
Your final grade will be based on homeworks, a midterm exam, graded clicker questions, and a blog post and course project. The weights on each of these components will be detailed closer to the start of term.
You are expected to maintain the utmost level of academic integrity in the course. Any violation of the code of academic integrity will be penalized severely.
You are allowed to collaborate on the homework to the extent of formulating ideas as a group. However:
Collaboration is not allowed on the other parts of the coursework.
- You must write up the solutions to each problem set completely on your own, using no memory aids whatsoever from your discussions, and understand what you are writing.
- You must also list the names of everyone that you discussed the problem set with.
Finally, plagiarism deserves special mention here. Including text from other sources in written
assignments without quoting it and providing a proper citation constitutes plagiarism, and it is a serious form of academic
misconduct. This includes cases in which no full sentence has been copied from the original source, but large amounts of
text have been closely paraphrased without proper attribution. To get a better sense for what is allowed, it is highly
recommended that you consult the guidelines maintained by Cornell on this topic. It is also worth noting that search
engines have made plagiarism much easier to detect. This is a very serious issue; instances of plagiarism will very likely
result in failing the course.