Robustness and Optimization of Scrip Systems
Scrip systems, where users pay for service with an artificial currency (scrip) created for the system, are an attractive solution to a variety of problems faced by P2P and distributed systems. Despite the interest in building scrip systems, relatively little work has been done to help answer basic design questions. For example, how much money should there be in the system? What will happen if some of the users start hoarding money? I present a game-theoretic model of a scrip system and show that this model has Nash equilibria where all agents use simple strategies known as threshold strategies. In fact, the same techniques provide an efficient method of computing these equilibria as well as the equilibrium distribution of wealth. I show how these results provide practical insights into the design of scrip systems. For example, social welfare is maximized by increasing the money supply up to the point that the system experiences a "monetary crash," where money is sufficiently devalued that no agent is willing to perform a service. Hoarders generally decrease social welfare but, surprisingly, they also promote system stability by helping prevent monetary crashes. Furthermore, the effects of hoarders can be mitigated simply by printing more money. This represents joint work with Ian Kash and Eric Friedman.
Other Videos By Microsoft Research
2016-08-08 | Prior Robust Optimization |
2016-08-08 | Reflections on Anonymous |
2016-08-08 | Building a Better Anonymous |
2016-08-08 | Approximation Algorithms for Facility Location Problems and Network Routing Problems |
2016-08-08 | Product Club |
2016-08-08 | Mime/Dance |
2016-08-08 | The Globalization of Sumo |
2016-08-08 | My Little Pony or Porn Star |
2016-08-08 | To Bot or Not |
2016-08-08 | Pew Data, Challenges and Opportunities |
2016-08-08 | Robustness and Optimization of Scrip Systems |
2016-08-08 | Intros and proposed talks from all attendees |
2016-08-08 | Provable Bounds in Machine Learning |
2016-08-08 | Windows Phone Garage: Building Awesome Apps |
2016-08-08 | Fourier Sampling and Beyond |
2016-08-08 | New Approximation Algorithms for Traveling Salesman Problem |
2016-08-08 | Community Cellular Networks |
2016-08-08 | A Combinatorial Prediction Market for the US Elections |
2016-08-08 | The EDSAC Replica Project |
2016-08-08 | Image Classification Using a Background Prior |
2016-08-08 | Tutorial 3: Data-Constrained Environmental Modeling: FetchClimate, Filzbach, & Distribution Modelle |