Research Areas

Engineering Markets includes many different research areas, such as:

Algorithmic Mechanism Design
Mechanism design (in economics) studies how to design economic systems (a.k.a., mechanisms) that have good properties. Algorithm design (in computer science) studies how to design algorithms with good computational properties. Algorithmic mechanism design combines economic and algorithmic analyses, in gauging the performance of a mechanism; with optimization techniques for finding the mechanism that optimizes (or approximates) the desired objective. (Contact: Jason Hartline, Nicole Immorlica)

Applied Mechanism Design
 Internet services are fundamentally economic systems and the proper design of Internet services requires resolving the advice of mechanism design theory, which often formally only applies in ideological models, with practical needs and real markets and other settings. The interplay between theory and practice is fundamental for making the theoretical work relevant and developing guidelines for designing Internet systems, including auctions like eBay and Google's AdWords auction, Internet routing protocols, reputation systems, file sharing protocols, etc. For example the problem auctioning of advertisements on Internet search engines such as Yahoo!, Google, and Live Search, exemplifies many of the theoretical challenges in mechanism design as well as being directly relevant to a multi-billion dollar industry. Close collaboration with industry in this area is a focus. (Contact: Jason Hartline, Nicole Immorlica)

Market Equilibria
The existence of equilibria has been the most important property for a "working" market or economy, and the most studied one in Economics. Existence of a desirable equilibrium is not enough; additionally, the market must converge naturally and quickly to this equilibrium. Of special interest here are decentralized algorithms, ones that do not rely on a central agent, that demonstrate how individual actions in a market may actually working in harmony to reach an equilibrium. (Contact: Hai Zhou)

Prediction Markets
Prediction markets are securities based usually on a one-time binary event, such as whether Hillary Clinton will win the 2008 election. Market prices have been shown to give a more accurate prediction of the probability of an event happening than polls and experts. Several corporations also use prediction markets to predict sales and release dates of products. Our research considers theoretical models of these markets to understand why they seem to aggregate information so well, "market scoring rules" that allow prediction markets even with limited liquidity, the effect of market manipulation and ways to design markets and their presentation to maximize the usefulness of their predictive value. (Contact: Lance Fortnow)

With the increasing popularity of Internet and the penetration of cyber-social networks, the security of computation and communication systems has become a critical issue. The traditional approach to system security assumes unreasonably powerful attackers and often end up with intractability. An economic approach to system security models the attacker as a rational agent who can be thwarted if the cost of attack is high.(Contact: Hai Zhou)

Social Networks
A social network is a collection of entities (e.g., people, web pages, cities), together with some meaningful links between them. The hyperlink structure of the world-wide-web, the graph of email communications at a company, the co-authorship graph of a scientific community, and the friendship graph on an instant messenger service are all examples of social networks. In recent years, explicitly represented social networks have become increasingly common online, and increasingly valuable to the Internet economy. As a result, researchers have unprecedented opportunities to develop testable theories of social networks pertaining to their formation and the adoption and diffusion of ideas/technologies within these networks. (Contact: Nicole Immorlica)

Wireless Networking
Recent advances in reconfigurable radios can potentially enable wireless applications to locate and exploit underutilized radio frequencies (a.k.a., spectrum) on the fly. This has motivated the consideration of dynamic policies for spectrum allocation, management, and sharing; a research initiative that spans the areas of wireless communications, networking, game theory, and mechanism design. (Contact: Randy Berry, Michael Honig)