Decision Theory

I've gotten more interested in thinking of policy-making (or decision-making in general) in terms of data management. I was inspired by the "Limits of Organizations" lectures by Kenneth Arrow. Managing information and routing information to the right recipients is the main purpose of an organizational structure and processes for decision-making. In other words, who fills out what forms, which fields are actually crucial, who reviews the entered data, and how do decision-makers actually use this data to make decisions.

Different kinds of needs call for different kinds of processes and structures. Hierarchies are best for well-defined objectives that everyone understands and buys into. Horizontal structures are important when the problem is not well-defined and there are many different actors. Everyone's interests can be thought of as a data-point, and at the decision-making stages, no one's interests are inherently more valid than anyone else's interests. Routing the relevant information to "decision-makers" in the policy realm is a special challenge because different actors don't necessarily agree on the problem or even on the state of world.

There is way too much information in any organization. When viewed in this light, it is clear that people often sort through huge amounts of data based on trust. To me, this means that data analysis and trust/nepotism are two sides of the same coin.

Robust decision making is research from the RAND corporation. It is an alternative technique for decision-making other than optimization and sensitivity analysis. An important aspect of these techniques is that the decision-makers are generally people who need to be able to understand the trade-offs. There seem to be concepts and techniques from information theory and robust controls that may be useful for policy-making. Compressing data is like figuring out the minimum information on forms to make decisions. Improving physical links is like the route of information in an organization and the chain of command.