[fixermark]

Politics with concrete numerically-quantifiable values is much better politics.

But you can't replace the politics with the math, as the exercise of deciding the alpha value in the Pareto distribution demonstrates. The numbers don't tell us what people will value; that's where the politics comes in. It's the art of figuring out human values and coming to terms to try to satisfy as many as possible.

[logfromblammo]

In theory, the optimal alpha could be determined experimentally. It would be an expensive experiment, though.

For now, we just have anecdotal historical examples, like the Soviets, and the European monarchies, and the ruins of ancient empires, and the banana republics, and the resource-curse countries.

We have already determined that the extremes at either end are bad. Concentrate the wealth too much, and you get slavery or serfdom, and stagnate. Distribute it too much, and you get socialism or communism, and stagnate. Follow some happy medium somewhere between them, and your economy grows.

In the absence of an experiment that would require extensive political maneuvering to even begin to conduct, we substitute less extensive political maneuvering in order to make some reasonable assumptions. It is better than just allowing the person that derails the discussion with the loudest distraction to win.

Besides that, experiments undertaken to determine the optimally fair distribution of wealth are as likely to be used to craft an illusion of fairness as they are to create a real environment of fairness, so even if we knew, we wouldn't necessarily be able to do.