Price of Anarchy in Algorithmic Matching [Virtual Talk]


Date
Apr 28, 2023 12:00 PM
Event
CRISS-LAB Seminar Series I-2023
Location
CRISS-LAB, UDD, Las Condes, Chile.

Abstract:

[ENG]

Algorithmic matching is a pervasive mechanism in our social and economic lives, used for example in algorithmic job search engines, kidney transplant allocation algorithms, medical residency program allocation, and search engines for finding romantic partners and potential spouses. However, algorithmic systems pose a principal-agent problem with the potential for moral hazard. The agent's (or system's) interest is to maximize the use of the matching website, while the principal's (or user's) interest is to find the best possible match. This creates a conflict of interest: the optimal matching of stable users would lead to fewer users using the site over time, which is detrimental to the online matching industry. Here, we borrow the notion of Price-of-Anarchy from game theory to quantify the decrease in social efficiency of online algorithmic-matching sites caused by the system's self-interest.


[ESP]

El emparejamiento algorítmico es un mecanismo generalizado en nuestras vidas sociales y económicas, utilizado por ejemplo en motores de búsqueda de empleo algorítmicos, algoritmos de asignación de trasplantes de riñón, asignación de programas de residencia médica y motores de búsqueda para encontrar parejas románticas y potenciales cónyuges. Sin embargo, los sistemas algorítmicos plantean un problema principal-agente con el potencial de generar riesgo moral. El interés del agente (o del sistema) es maximizar el uso del sitio de emparejamiento, mientras que el interés del principal (o del usuario) es encontrar el mejor emparejamiento posible. Esto crea un conflicto de intereses: el emparejamiento óptimo de usuarios estables llevaría a menos usuarios utilizando el sitio con el tiempo, lo que es perjudicial para la industria de emparejamiento en línea. Aquí, tomamos prestada la noción de Precio-de-la-Anarquía de la teoría de juegos para cuantificar la disminución de la eficiencia social de los sitios de emparejamiento algorítmico en línea causada por el interés propio del sistema. (Traducido por ChatGPT).

Speaker Bio

Dr. Andrés Abeliuk is an Assistant Professor at the Department of Computer Science, University of Chile and a young researcher at Centro Nacional de Inteligencia Artificial (CENIA). His research focuses on analyzing the impact of algorithms in society and creating social computing systems to improve collective behavior through behavioral modeling, optimization, game theory, and online experiments. He received his Ph.D. in Computer Science from the University of Melbourne, where he worked with Pascal Van Hentenryck and Gerardo Berbeglia. His thesis examined the negative effects of social influence and algorithmic curation of content on the Web and in cultural markets. After his Ph.D., he worked as a postdoctoral researcher at MIT Media Lab and investigated algorithm theory, computational complexity, and combinatorial optimization in social systems. He also worked as a research scientist at the USC Information Sciences Institute, where he collaborated on the development of algorithms and prototypes of socio-technological systems.

Andrés Abeliuk
Andrés Abeliuk
Department of Computer Science, University of Chile