A. Outwards-spiraling iterative development across multiple successive levels.
- First level is the guy that actually comes up with the idea. Naturally, when someone develops a policy, they usually come up with some of the initial objections and work through them.
- The “wisdom of crowds” in some studies ended up being the wisdom of a number much smaller than a crowd. Get 2-5 others to review it and search for holes. Iterate.
- After several iterations, conduct more extensive modelling based on expected behavior.
- The policy is brought to a group of 10-20 people to review and find flaws in. (To get a proper review, incentives may need to reward good flaw-finding, perhaps according to a few supervisors.) Iterate.
- After several iterations, a small “lab-based” experiment is devised to test the policy, approved by some number of the flaw-finders. While this might seem like a toy model, behavioral economists have been able to develop some real findings by just seeing what smallish numbers of people actually do in their simulations. The experiment members should be prevented from suffering any negative repercussions for the providing politically “wrong” answers, and possibly assigned aliases for the experiment.
- Depending on results, go back to 1-4 to incorporate the new data. Iterate.
- Larger experiment with more complex model and more actors.
- Policy is rolled out to a small, real-world group that volunteers for it. Wait some appropriate amount of time to see initial results, mostly to rule out catastrophic failure. Iterate.
- Policy is rolled out to several, somewhat larger groups. Data is collected. Iterate.
- At this point we should have much more confidence in the policy, and can roll it out to a much larger organization, but still something below a whole state/multinational corporation (depending on the policy).
- Continue up/outwards.
Among key factors is that the experiments must have ways for experiment members to act contrary to the wishes of the pro-policy members, or to move sideways within the model as it were. Additionally, experiment members should be rewarded with real-world money to drive an incentive other than just appearing nice/virtuous. To achieve this adversarial nature, the anti-policy forces must be involved in planning or approving the experiment.
A framework of methods for game-theoretical defections (or however you want to put it) could be developed, since in the real world, “cheat and kill the guy” is an option in many scenarios.
While not strictly going to capture every way that a policy could go wrong, this should act as a series of sanity checks for preventing some of the worst policies, and highlight promising policies.
B. Proportional Block Grant Committee.
Have the national government collect some share of national tax revenue for conducting policy experiments. Since most experimental policies would be de facto subsidies relative to other states, issue it to states proportional to some factor like population or size (or maybe population times size). This means all the states are subsidized about the same, at least in terms of the policy spending, depending on implementation.
Use block grants awarded in such a way as to make it difficult to just use the money to offset tax cuts. Generally, give experiments to the subnational governments that most want to attempt them, since those same governments will be less likely to sabotage the experimental policies.
C. Internal migration is an experimental result.
Yes, putting a UBI in a province might result in people migrating to that province to freeload off it. Or it might result in taxpayers fleeing. Alternatively, it might not.
However, unless your country is going to ban emigration and immigration, this is actually important information, as are shifts in jobs, building, etc across the economy so long as your country must compete in the global economy.
None of this will be perfect, but it should be feasible to gather a good harvest of information.
