Please note that I’m doing my best to write this from a politically neutral viewpoint. If at some point, my deconstruction fails and it seems like I’m endorsing a certain party, this is unintentional. Assume I’m an alien from another planet.
I was inspired on this post on the Bayesian Conspiracy podcast- as well as by recent election shenanigans- to think about how people vote, and more importantly, why.
We’re all familiar, I think, with the old idea of the rational agent- that people will vote in a way to benefit their self-interests. On the surface, this idea seems like it would work. The ill-informed rational agent would vote in an almost random fashion, based on the gaps in their knowledge, leaving the well informed to pick up the slack, so things may go well even when it’s difficult to wade through the complexities of the proposed policies. We would, if people really vote in their own interests, end up with policies that benefit the majority of citizens. Of course, this would only work in a perfect democracy, where one person equals one vote and everyone is able to vote.
In a system of rational agents, when confronted with an issue that only affects a small minority, the majority wouldn’t have enough incentive to vote contrary to the minority’s interests. The major problems with a system like this is that if a specific class of people were better informed, they would vote in their own interest, and the interest of the ill-informed would be less-well represented. Also, policy that benefits a simple majority isn’t the same as policy that is optimized- that is, that benefits the greatest number of people possible.
The old idea of people acting as rational agents is quickly disappearing, however. Caplan, in his Myth of the Rational Voter argues, among other things, that people don’t vote in a way that will benefit them, because the effect of their vote is so small that there is little cost to voting in a way that makes them feel good. In other words, people vote how they feel they should, which is more prone to bias.
The real question, then, is how to build a system of democracy that benefits the greatest number of people possible, and also has safeguards against systemic bias. One system I’ve seen proposed is futarchy, in which people would bet on beliefs. It seems to me that futarchy would be weighted toward those who are able to invest more. A technocracy based on a computerized model runs into a similar problem: who programs the computer?
For now, I suppose we should continue to rely on our constitutional democracy, as much as the system seems to be malfunctioning. A constitution serves as some control against the more egregious forms of bias, and voters can do their best to educate themselves and update their beliefs based on the available data. In other words, until we’re able to truly optimize the system based on everything we’ve learned, we’ll muddle through.