Designing combinatorial markets for distributed data
Similarly to how the Industrial Revolution was accompanied by rapid developments of markets for natural resources (such as oil and coal), the development of new markets for information goods is crucial to fully realize the benefits of the Digital Revolution. In particular, technological innovations of the last few decades resulted in the production of huge volumes of data by numerous businesses and individuals. Making such distributed data available and easily exchangeable between different
... tween different parties leads to significant benefits for our society. In this thesis, I consider the problem of designing a market for distributed data. I focus on the following specific features of this domain: (1) data providers have high costs for producing their databases and low costs for maintaining these databases; (2) users submitting their queries have combinatorial preferences over which databases are produced and get allocated; (3) the market exhibits a complex two-level production structure; (4) data providers and users can arrive to the market stochastically over time. These factors outline the design space for the market for distributed data that I consider in this thesis. I start with designing a number of mathematical models that tackle some particular aspects of this design space in isolation: from modeling combinatorial preferences of buyers to studying uncertainty regarding availability of goods. I proceed with a formal market design in a static setting (i.e., assuming that buyers and sellers arrive to the market at the same time).