Conventional wisdom holds that government, especially in its executive and administrative capacity, must be “transparent,” disclosing how and why it makes decisions. Transparency, it is believed, limits corruption and encourages public participation. While legal scholarship has examined in detail the policy and legal questions about how much transparency agencies should provide in light of other concerns like deliberative latitude, privacy, or national security, scholarship has not examined the question of what is transparency—a concept that is not, well, transparent. This Article forwards a working definition of transparency and examines the central challenges in creating an administrative transparency regime.
Most legal scholars define transparency as access to information. Finding this definition incomplete, this Article argues that transparency involves two primary elements: one cost-based and the other normative. First, transparency is about lowering the cost of accessing information, particularly the cost of physical access to information in real-time data. In other words, “transparency” or “access” does not really exist if obtaining and securing information is costly in either time or effort. Second, transparency has a “computational” or “complexity” dimension, which has an inevitable functional or normative dimension.
In the administrative context, there is basic agreement about transparency’s normative or political purposes: Transparency limits corruption, protects against opportunistic behavior by officials, and encourages public participation. This Article examines the form of transparency used by numerous statutory and regulatory regimes and suggests reform focused upon lowering the cost of information, both temporally and geographically. “Real time” disclosure will open the smoke-filled rooms to a more democratic cast of special interests. Finally, this Article examines the role of Big Data and its possibly profound effect upon government openness and the relationship between the government and the governed.