Price and Allocation Algorithms for Markets and Networked Social Systems

Dartmouth College, Hanover NH

Investigators

Abstract

PRICE AND ALLOCATION ALGORITHMS FOR MARKETS AND NETWORKED SOCIAL SYSTEMS This project seeks a better understanding of the behavior of decentralized systems such as markets and traffic networks, via effective resource allocation and price determination for these systems. This focusses on several problems: equilibrium market price discovery, measurement and control of traffic behavior, auctions for expiring resources, two-sided auctions for ordered goods and bidders (such as web advertisements). The goal of the price discovery work is to explain why price changes in market economies might work effectively by showing simple, independent, distributed algorithms that cause prices to converge quickly toward equilibrium values. The traffic work examines how network performance degrades in the presence of multiple large, selfish network users, and study methods to control this. The two-sided auction work investigates the design of markets where the buyers have agreed preferences for objects to be sold, and sellers have agreed preferences for who to sell to. The goal is to design real-time price and allocation mechanisms that allow bids from both buyers and sellers, while insuring both profitability and market efficiency. The work on auctions for expiring items attempts to assign renewable resources to those bidders (that arrive and depart over time) who can derive the highest value from them. The goal is to design auctions in which it is in the bidders' best interests to bid their true valuation. Auctions with this property are easier both for bidders to figure out their optimal bid, and for auction designers to understand and predict the behavior of bidders.