Voting, in our system, is supposed to be deliberative, so for all of the hue and cry from progressives about “voter suppression,” there is nothing wrong with requiring voters to do things like register to vote in advance, show up on Election Day, and bring an ID. That said, democracy is for everyone. By definition, half of the voters are of below-average intelligence. By definition, half of the voters are of below-average education.

The fundamental American system of elections has a basic simplicity that inspires confidence. The essential model is that each party picks a candidate, there is a choice between the two, and whoever gets the most votes wins. All you need to do is look at the totals for the top two, and you can see that. At the presidential level, this is done in 50 statewide races (plus D.C.) rather than aggregating those races nationally, for the same reasons why members of the Senate and House are elected by the people of districts or states, but the principle is the same. Sure, there are third-party candidates, but those are a familiar feature, and they usually lose. After two centuries, Americans are long accustomed to the simple idea of two choices, one vote, the most wins. We should not mess with what works simply because we think that tweaking the system in ways that look clever in a political-science class will have predictably good effects.

A consistent preference for the basic American binary-choice, first-past-the-post model is why I have argued consistently against jungle primaries, against general-election runoffs, against recalls, and against the Maine and Nebraska systems of running presidential elections by congressional district. At most, we should use runoffs in party primaries, and only when the leading vote-getter falls below a low threshold (say, 40 percent) that suggests that the party has not really made a choice. But primary runoffs are about picking the best candidate for the party, not respecting the democratic legitimacy that is central to general elections.