Public policy based entirely on ideology is flawed. Our objective is often to ensure that our citizens have their needs met, but when we change our course dramatic and capriciously (as often happens after an election), we often fail our citizenry. Through the open government movement, and the open data movement, we have an opportunity to change some of this.

What if we decided on public policy based on what works, rather than our feelings about it? We could take 10 randomly selected public policies, all intended to address the same issue. Apply them as public policy to 10 randomly chosen similar jurisdictions (cities would probably have to collaborate to do this) and use the data collected from those jurisdictions to find out how effective the policies are. We would then discard all but the 3 best performing public policies, and randomly select 7 new policies to replace the ones lost, and rerun the experiment. We could tweak the three policies we've chosen (using some random variation on the various aspects of the policies).

Alternatively, once we have enough data collected from enough jurisdictions via the open data and open government movements, we could enter in all of this data into a computer, and run the policy algorithms on a simulation, rather than in the real world. At the very least, this computational approach could narrow down the field to policies which seem effective.

At the very least, the open data movement should allow us to do more effective research on public policy, but it would be interesting to see if any municipalities or governments would be open to an experiment of this kind.