Matheus Venturyne Xavier Ferreira will present his Pre-FPO "Economics and Computation in Decentralized Systems" on Wednesday, May 12, 2021 at 2pm via Zoom.

 

Zoom Link: https://princeton.zoom.us/j/94626343984

 

Committee:

Examiners: S. Matthew Weinberg (adviser), Ran Raz, Arvind Narayanan

Readers:     Mark Braverman and David Parkes (Harvard University)

 

All are welcome to attend.

 

Title: Economics and Computation in Decentralized Systems

 

Abstract:

While algorithmic game theory studies algorithmic design under incentive constraints, computer security explores the protection of a computer system against adversaries. In one direction, economics provides new perspectives in designing secure computer systems because, in many real-world settings, adversaries are not intentionally malicious but rather rational and economically driven. Notably, Bitcoin has been successful in solving the permissionless Byzantine agreement by monetarily rewarding honest behavior. In the other direction, computer security provides new perspectives to mechanism design. Internet systems are increasingly less transparent and decentralized, which often challenges traditional assumptions from economics. Notably, internet auction lacks transparency, and in many practical instances, auctioneers cheat to improve revenue.

 

In this talk, I will overview my research on designing incentive-compatible systems for an age where Internet systems are increasingly less transparent and decentralized. I will instantiate this agenda by presenting two projects. First, I will demonstrate how to implement incentive-compatible credible auctions where the auctioneer does not have incentive to cheat. I will highlight the connections between credible auction design with the problem of resource allocation on blockchains. Second, I will show that energy-efficient alternatives to proof-of-work blockchains can be ``almost'' as secure as proof-of-work when given access to public randomness by characterizing optimal strategies for the underlying infinite-state Markov decision process.