Doctoral Speaking Skills Talk - Caspar Oesterheld
— 5:00pm
Location:
In Person
-
Newell-Simon 3305
Speaker:
CASPAR OESTERHELD
,
Ph.D. Student, Computer Science Department, Carnegie Mellon University
https://www.andrew.cmu.edu/user/coesterh/
Often when you make a decision in the real world, the outcome (your payoff) is determined by some strategic interaction between a group of other players, a _game_ in game-theoretic parlance. For instance, the outcome of a government regulation in a market depends on how various market participants interact under the regulation. Unfortunately, strategic interactions are often hard to judge because they have multiple strategic solutions (e.g., multiple Nash equilibria). In the talk, I will explain this problem of strategic multiplicity in more detail, as well as some existing solutions. I will then introduce the concept of _safe improvements_ (SIs), a new approach to (sometimes) resolving such questions. The SI approach is to analyze how pairs of games relate. Sometimes this allows us to confidently reach conclusions about which of two games is better (as judged by some specified preferences) without requiring predictions about either of the individual games. For instance, if two games are isomorphic, it stands to reason that they will be played isomorphically.
After explaining the basic idea behind SIs, I will focus mostly on computational questions about them. For instance, how hard is it to decide whether one given game is an SI on another? If we have some way of intervening on a game (say, removing an arbitrary subset of actions), how hard is it to find an intervention that induces an SI on the given game?
Presented as part of the Econ-CS Seminar Series
Presented in Partial Fulfillment of the CSD Speaking Skills Requirement
For More Information:
matthewstewart@cmu.edu