An old post from The Grumpy Economist, John Cochrane, Why the Electoral College Is a Great Idea

But every theorem has assumptions. The median voter theorem assumes that political outcomes can be placed on a one-dimensional line, and that preferences are “single peaked,” people liking the outcome closest to their preference. Ask any libertarian where they stand on the left-right continuum and you get a long lecture. Our parties are coalitions of very strange bedfellows, not points on the median voter line.

My worries are about outcomes where there are many dimensions to what we care about — foreign policy, economic policy, environment, social questions. Most of all, my worry is geographic concentration — red states vs blue states, and candidates who win by getting 90% majorities in one or the other.  My worry is that the cost of getting a deeper majority in your own state is less than the cost of getting a slim majority in a battleground state. Costs are absent in the median voter theorem.

The point of an election is not really about selecting a winner and a loser, or a set of policies. It’s about building a consensus, that the losers  agree to live together under the winner, and try again next time. A geographically broad set of 51% wins with a popular loss is more important to that goal than a 51% popular vote win based on overwhelming majorities in narrow areas, and overwhelming losses in the rest of the country.

from a much more recent post in the same blog, Electoral College:

The electoral college forces candidates to attract geographically dispersed support. Moving a swing state from 45% to 55% is much more important than moving a solid blue or solid red state from 75% to 85%.

The deep point: When you set up rules for anything, there is a tension between measurement and incentives. Once people show up at the polls on election day, there is a strong case that “each vote should count the same.” But if you do that, the incentives, and hence the outcomes will be much worse.

In the end, we care more about a good outcome — good policy, and politics that keep the country from flying apart. So, swallow hard, measure imperfectly but set up better (slightly less bad) incentives.

This is a point worth repeating. It’s the kind of thing that economists really do have to offer to students of politics and the world at large.  Good incentives for politicians, parties, and agendas  matter more than good measurement.