Q&A: Majority Rule versus Statistical Decision-Making
Majority Rule versus Statistical Decision-Making
Question
Good week, Rabbi,
Recently, to my regret, I ended up watching “Survivor” (a reality show), and it got me thinking about group decision-making. Decisions can be made either by majority rule or by a statistical decision procedure in which each vote gets an equal probability in a lottery that the group will act according to that person’s opinion. In democracies, it is usually accepted that the majority decides. But statistical decision-making seems fairer to me, because even the minority has some chance that its view will be taken into account. I understand that there are decisions where you have to go by the majority, and others where you can decide statistically. But for some reason you hardly see this decision-making mechanism in the democratic world. What do you think about this mechanism of statistical decision-making, and when do you think it is right to apply it?
Best regards,
Answer
I wrote in the past that, for example, on the Supreme Court, when there was proportional representation for religious justices, it would not have helped unless the majority on the various panels were determined by lottery. Otherwise the minority always loses, and 100% of the decisions are made by the majority.
In political decisions in a democracy, this is an interesting question. There is a feeling that things are different there, but I need to think about it.
One could perhaps say that decisions are interdependent, and therefore you need a mechanism that makes all the interconnected decisions together. But it is impossible to know in advance and formalize what depends on what and why, so people look for one single mechanism (a government). But in decisions in themselves, or regarding the choice of the government itself, this does seem like a persuasive and appealing argument. There is a certain problem, though: if the number of statistical decisions is relatively small, then you run into Pascal’s point—that in one of the columns it was argued against using expected value when the number of trials is small and one-time only.