How can we incentivize public agencies to share electronic information effectively? I first argue that the Obama Administration's Web 2.0 methodology of addressing this question (http://www.data.gov) will fail because (1) Federal agencies will not willingly surrender precious information assets that are the source of their political power; (2) valuable data has a currency value in the information age—it is not free; and (3) a Web 2.0 approach to sharing public sector data ignores (and sometimes exacerbates) important problems including how to delete data about citizens, how to create historical versions of the digital image of the citizen, and how to uncover the hidden and powerful algorithms that govern the production of public sector data.

Instead of Obama's Open Data approach, I propose to redefine the public sector as that which holds the legitimate monopoly on primordial data (such as social security IDs). Based on this, I have developed a new information sharing architecture based on three key ideas: (1) bureaucratic politics must take primacy over technology; (2) bureaucratic language can be automated to facilitate information-sharing transactions; and (3) information-sharing transactions can be monetized as buy and sell transactions.

This paper also presents the theoretical foundations for a computational linguistics project to erect a Public Information Marketplace, and discusses several political, legal, economic, ethical, regulatory and technical challenges for building such a marketplace. The paper concludes with a concrete political strategy on how to advance the proposed new information marketplace.