Hysteresis

Main article: System modeling

We have identified hysteresis as a necessary component of social systems, as it certainly is in natural systems. Generally speaking we want our ratings system, and associated decision-making systems, to not fluctuate wildly in their output. We especially want to prevent rapid fluctuations in policy based on the whims of the moment. So we’d like to build some hysteresis into these systems and understand any hysteresis that’s already built-in to socio-economic systems in general.

One historically good example of this is inflation and the policy responses to it. Inflation is usually characterized by a wage-price spiral. Prices of some good, perhaps a commodity like oil, increase which raises prices across the economy. Eventually wages rise to compensate for the price hike, which leads to another price hike. The cycle continues and, once inflation expectations are established, they are very hard to break. This is an example of hysteresis. Ending the cycle usually involves a prolonged period of high interest rates by the Fed to lower business spending and frequently causing a recession. Once the Fed takes this action, political pressure is brought to bear on it to reverse the high interest rates because they are almost always unpopular. The Fed, however, is politically independent precisely to resist such pressure. It has the ability to enforce hysteresis without flip-flopping. Keep in mind that since higher interest rates take time to have an effect (again, hysteresis), there is often little to do but wait while maintaining the policy (ie staying the course).

The Fed is an example of building organizations that are insulated from politics but this isn’t necessarily our first recourse. We’d like to build our hysteresis in as democratically as possible and retain ratings-based control over official organizations. Decision mechanisms, then, voted by the community, need to be the basis for hysteresis.

In our current system, elections are notorious for electing poor leaders who effectively appeal to momentary concerns or vote out good leaders who haven’t had time to see their policies work. We want to avoid this and educate the voting public about the dangers of succumbing to it.

Perhaps the first mechanism, therefore, is simply educating community members on how the best decisions are usually not made on a whim but are expressions of preference over time. Simulation clearly plays a significant role here. It allows us to step back and study what each policy or candidate can achieve over a decent length of time. Many of Biden’s policies, for example the inflation reduction act (really a climate change bill), anticipate a lot of infrastructure built in the future so haven’t had much effect yet. But simulations into the future vs. other policies can do this.

Furthermore, the ratings system allows for constant voter-sampling, whether for an official election or not, and should be able to alert people when their preferences have changed markedly. The system thereby allows people to recognize their own perhaps irrational fluctuations and correct them. We’ve discussed in the past the notion of building in a time delay between submitting a rating (or a vote) and seeing it officially published.

If these measures alone prove insufficient, then voting rules can be established. Elections, particularly about policy, can be held over a period of time to establish integrated averages. There is nothing that requires us to hold elections on a single day with a single ballot like we do today. For a consequential decision, an election can be held once a quarter for a year (say), to establish a new policy direction.

The voting system can also build in some hysteresis depending on the policy decision. New policies can be evaluated for how drastically they change current policies and make the requirement for enacting them greater than 50% or subject to repeat elections (see above). This would involve modeling the policy to evaluate numerically the magnitude of change that it represents. Another method would simply involve breaking the policy into steps, each of which can be voted on independently and implemented in a step-wise fashion, with enough time to evaluate the effects of each step.

Let’s relate this to a simple and stark problem: what to do when someone decides to commit suicide. Do we let them, because it is their decision to make, or stop them because we know if they just wait a few days their despair will pass and they will be glad to be alive? Clearly suicide is a decision that should be approached with a heavy degree of hysteresis. Chronically sick people, in constant pain, who are going to die anyway may rationally choose suicide and have a consistent preference over time to back their decision up. The teenager who just got dumped probably does not.