Daniel Wodak has a very interesting paper, Which Majority Should Rule? (PDF), in which he argues for the following position:
POPULAR MAJORITARIANISM If the majority should rule, the option that won the most votes should win, as the popular majority must control how they are governed.
This is, as Wodak notes a position that is often said to be true, but often not defended and often implicitly or explicitly attacked; it is also not strictly true of most national election systems. Most legislatures in the free world, for instance, are elected on the basis of districts or ridings or wards, which often are divisions of even larger constitutional units like states or provinces. The U. S. House of Representatives, for instance, is elected not on the basis of pure majority in a popular vote but on the basis of districts of states, and a party controls the House (and therefore actually controls legislative powers of governance) not on the basis of votes but on the basis of districts won across all the states. The major opposing position to Popular Majoritarianism, and the one that Wodak primarily argues against, is Electoral Majoritarianism.
Wodak unfortunately does not give any strict statement of Electoral Majoritarianism, but he associates it with Mill's description of common electoral systems as based not on rule by majority but rule by the majority of the majority. In effect, the idea is that if the majority should rule, the option that won the most districts on the basis of the most votes in each should win. An important aspect of this, which will be relevant to my comments, is that Electoral Majoritarianism can be seen as two-tier Popular Majoritarianism. You can in fact characterize many forms of Electoral Majoritarianism in the terms that Wodak uses to characterize Popular Majoritarianism, but for Electoral Majoritarianism, there are two kinds of voters in two kinds of elections: the individual voters (what we normally call voters) and the districts. The individual voters decide the vote of their district, and then the option that won the most district-votes wins. (It's less clear whether all forms of Electoral Majoritarianism can be characterized as such.)
I do not think that any of the arguments for Popular Majoritarianism are sound. It is particularly impossible, I think, to affirm Popular Majoritarianism on gounds of 'self-governance of the people' while also accepting the principle of representative government, because representative government is like Electoral Majoritarianism two-tier in a way that makes almost all, and perhaps all, of the arguments for Popular Majoritarianism easily adaptable into arguments against representative government; this is particularly true in a federal system. Representative government might not be logically inconsistent with Popular Majoritarianism as a general system, but I think it is practically inconsistent.
One argument sometimes brought against Electoral Majoritarianism is that it violates what is sometimes called anonymity, the principle that the outcome of the election should not depend on which voters vote for or against. This is very important, because in fact most people who accept Electoral Majoritarianism do so precisely because they reject anonymity. For instance, in the United States our electoral system is at several different levels set up to obtain spread-around majority. In the United States, elections are run by states, with each state running its election its own way; the President Elect is the candidate who wins the majority of elections, as weighted by the Congressional Representation (and thus partly by the population) of the states that run them. The entire purpose of this is to guarantee that the President of the United States is not put into power entirely by the citizens of a few states. Virginia was the state that specifically worried the Founding Fathers: Virginia at the time the Constitution was ratified was approaching one-fifth of the entire population of the United States, with the next most populous states, Pennsylvania, North Carolina, and Massachusetts, each around 10%. (As a comparison, today California has about one-eighth of the U. S. population.) In any pure majority election, the interests specific to Virginia residents would pack around twice the weight of the interests of even a very populous state; and more than ten times the weight of the interests of the smallest states. For the Founders, it did matter which voters were voting, because voter interests mattered; a system in which large-state interests could swamp small-state interests was regarded as intolerable. But while the U. S. system is particularly devoted to the principle of a spread-around majority, other election systems do in fact tend to get divided up on principles that at least make a spread-around majority more likely (for at least some offices) than it would be under Popular Majoritarianism. Large-state and small-state, urban and rural, traditional ethnic and subnational groupings: there are lots of reasons why one would prefer that at least some offices are manned by people with broadly distributed support, and all of those reasons are reasons why one would reject anonymity.
This is relevant to other arguments that Wodak considers, utilitarian and epistemic arguments. In fact, almost all utilitarian arguments for Popular Majoritarianism make the elementary mistake of assuming that successful-vote-in-a-single-election maximization is utility maximization; this is not a sense of 'utility' than anyone uses in any other context, and there is no reason why one would accept it unless one were already a Popular Majoritarian. Other, more natural conceptions of utility maximizations could, depending on the circumstances, just as easily conclude for a spread-around majority as a bare popular majority. Epistemic arguments usually run into analogous problems, since a concern that voter outcomes be not merely based on strongly shared judgment but also on well-rounded judgment might lead one to the spread-around majority conclusion -- if your interest is informed judgment, you want governing authorities who aren't merely judged by a lot of voters to be the preferable governing authorities, but are judged to be the preferable governing authorities by a lot of voters in a lot of different contexts with a lot of different backgrounds. Contrary to the Condorcet Jury Theorem, the probabilities of voters being right vary wildly and these probabilities are not independent of each other; you therefore want to dampen the extent to which a single very large group's biases could be biasing the whole result, and prefer a system that considers a lot of groups with different biases. (Wodak later considers something like this point, but claims that larger groups are more likely to be diverse, but (1) this is obviously not something that can just be accepted, because it depends on how the larger group forms, and large majorities do not form randomly; and (2) it is not likeliness of diversity but of politically relevant diversity that matters. This is what people mean when they say they don't want to be ruled from Los Angeles County; people in Idaho don't want their voice on farming policy to be swamped out by Californians thinking mostly about California, for instance. Indeed, I can guarantee you that you can get an anti-California popular majority by asking voters in every other state whether they want a system in which they can be consistently outvoted by Californians; they will, I am certain by a large majority, say no. We are already grouped politically by states, and this affects our interests on agriculture, industry, immigration, housing, and many other things, in a thousand different ways.)
The only arguments for Popular Majoritarianism really worth taking seriously -- and in practice, I think the ones that are most often actual reasons for accepting it, as opposed to an argument one could make in its favor -- are egalitarian arguments. Utilitarian and epistemic arguments about majority rule, I've suggested, only get Popular Majoritarianism as a conclusion if you are assuming something very much like Popular Majoritarianism to being with, and procedural arguments against Electoral Majoritarianism, like those based on anonymity, are the same. Although I will not go into it, I think Wodak's major negative argument against Electoral Majoritarianism, that it creates arbitrary or perverse results, is also question-begging; besides the fact that there are no unique solutions in electoral systems (even popular majority systems are affected by things like distance to polling station and ballot order), the 'arbitrariness' and 'perversity' classifications are based on the assumption that an electoral system should have the shape that results if you assume something like Popular Majoritarianism. However, I do think that people have independent reasons for accepting principles like "one person, one vote" and "every vote counts equally", reasons that do not necessarily depend on anything like Popular Majoritarianism. My view is that this is not necessarily a problem for Electoral Majoritarianism, which does not reject these ideas outright, but holds that they are true in being applied to particular contexts rather than universally, and most Electoral Majoritarianism has at least some concern with diachronic issues, which allows for the argument that while in any given election 'electoral influence' may be unequal, this is not a problem as long as the inequalities are transient rather than stable. But Wodak is right, I think, that at least many people's understanding of "every vote counts equally" directly implies something like Popular Majoritarianism, so that if your understanding of 'majority rule' is egalitarian in these ways, it does require that the most votes win.
Despite my disagreements, it is all in all a very nice paper, with the arguments mostly laid out very well. And more important than that, I think he has the stick by the right end, which is uncommon in these discussions. "Which majority should rule?" is indeed the key question of democracy; the correct answer, which is not Wodak's answer, is that it is the majority that, in the first and most important place, most likely avoids political breakdown and civil war, and, after that, gets us as close to the kinds of campaigns and compromises that give us some kind of general consensus on major policy issues, as far as possible. Numbers are going to be important to that, but mere numbers are not always going to be enough. But to see that we first need to be asking the right question.
