Crowdsourcing refers to the paradigm of eliciting work, typically in small pieces, from a global population of workers, each making dynamic decisions about which task to complete next, and for whom. Crowdsourcing systems are inherently algorithmic because the coordination task would otherwise be overwhelming. Algorithms take the place of the management and incentive structure of traditional firms, responsible for optimizing workflows, determining payments, matching workers with tasks, and learning models of workers and tasks.
With the potential to transform the way in which productive effort is allocated, crowdsourcing is already finding application across a broad range of domains, such as citizen science, the transcription of documents, and collecting training data for use in machine learning for bridging the gap to human-level intelligence.  The goal of this project is to develop a cohesive theory of algorithms and incentives for the design of crowdsourcing systems.
The proposed research addresses key challenges in harnessing decentralized resources and capabilities for productive work, and seeks to develop a theoretical framework to guide the design of crowdsourcing systems that align incentives, are adaptive and robust, and achieve high performance for low cost.  The investigators are interested in the roles of matching and pricing, in developing methods for eliciting high quality, unverifiable contributions, and in embracing realistic models of human behavior.
The investigators will organize workshops on social computing at upcoming Conferences on Electronic Commerce (EC13 and EC14) and they continue to recruit women and other under-represented groups to join the project.
