The principle of democracy is equality, and the political expression of that principle is majority rule. Majoritarianism is democratic because it treats every vote cast by every citizen as equal.

The Constitution departs from the principle of majority rule in several ways, such as requiring overwhelming rather than narrow majorities for constitutional amendments, or giving each state two senators, whether it has 523,000 people, like Wyoming, or 36 million, like California. The rationale for these qualifications of majoritarianism is not opposition to democracy but support for good government. The authors of the Constitution were good democrats, and, as Harvey Mansfield has written, good democrats “must not think that government is automatically good merely by being democratic.” Rather, they understand that “democracy can be good, and when they see that it is not, they take responsibility for reforming it. To do this they must think that good government as a standard is above democracy.” National cohesion and reassuring as many groups as possible that their concerns will be respected are among the aspects of good government that can be advanced by compromising the principle of majority rule.

If the Electoral College is abolished in favor of giving the presidency to the candidate who wins a majority of the national popular vote, American government will be rendered more democratic but less good. The most important virtue that will be lost is that the Electoral College penalizes candidates who amass deep but narrow support, and rewards those who secure broad support. A national popular vote doesn’t, and could lead to unpleasant consequences.

Suppose that access to fresh water becomes one of the nation’s leading controversies, with most of the nation favorably disposed to increased irrigation, but the states with shorelines on the Great Lakes – Minnesota, Wisconsin, Illinois, Indiana, Michigan, Ohio, Pennsylvania and New York – strongly opposed. If we abolish the Electoral College and elect the president who gets a majority of the national popular vote, a candidate could get 60% of the vote in those 8 states, lose the other 42 states and the District of Columbia with an average of 45.6% of the vote, and still win the presidency. (This calculation is based on the number of votes cast in each state in the 2004 presidential election.) If a candidate gets two-thirds of the vote in the Great Lakes states he could win despite receiving only 42.7% of the vote in the rest of the country; 75% majorities in those 8 states and he would need only 39.1% elsewhere.

The winner-take-all feature of the Electoral College, by contrast, makes it impossible for a candidate to win the presidency if he is widely opposed but also wildly popular in a few states with a common concern or outlook. There is no incentive to run up the score by increasing your majority in the states that already favor you. Instead, you have to campaign in enough states, and appeal to enough voters in different kinds of places and situations, to win. The eight states in the above example cast 141 votes in the Electoral College, far short of the 270 needed to win. The Electoral College, in the scenario sketched out, would give a clear victory to the candidate who received the second-largest number of popular votes, but who demonstrated a much broader degree of popular support, or at least popular tolerance. It would be moderating rather than polarizing.

The problem with the Electoral College is that it is also capable of denying the presidency to a first-place finisher who did not run up the score in just a few states, but who merely had the bad luck to get a national majority distributed among the states in a way that was a little less favorable for the purposes of the Electoral College’s math than his opponent’s. No one has ever contended that Al Gore’s defeat in 2000 was a good thing because his plurality of the popular vote – 48.38% to Bush’s 47.87% - was narrowly drawn from just a few states. Gore won 20 states plus the District of Columbia, including states on the West Coast, New Mexico, the upper Midwest, the middle Atlantic, and New England. Leaving aside the Florida controversy, his margin of defeat in 5 other states was less than 5%. America may have been spared, or denied, many things because of the Gore presidency that wasn’t, but it did not avoid the misfortune of president whose popular support was narrowly rather than broadly based. By the same token, if 60,000 voters in Ohio in 2004 had switched their votes from Bush to John Kerry, Bush would have lost the election despite receiving nearly 3 million more popular votes than Kerry out of the 122 million that were cast, winning 30 states, and coming in second by a margin of less that 5% in 7 others.

Such outcomes, real or hypothetical, are rightly seen as capricious and arbitrary, rather than curtailments of the principle of majority rule for the sake of promoting good government in some intelligible way. Under the Electoral College, winning a majority of the national popular vote means nothing. If we abolish it in favor of direct popular election, it will mean everything. The best system would be for it to mean something, but only something.

Modifying the Electoral College by means of a “national bonus plan” would accomplish that goal. When I learned that one of the advocates of this reform was Arthur Schlesinger, Jr., I was skeptical, too. But good ideas can come from surprising precincts. The national bonus plan calls for keeping the Electoral College, but awarding the presidential candidate who finishes first in the nation-wide popular vote an additional number of electoral votes. The idea is to prevent the silver-medalist from becoming president in an election like 2000, but enable him to do so to thwart something like the Great Lakes coup described above.

How big should the bonus be? Schlesinger suggested 102, two votes for each of the 50 states plus Washington, DC. If you make the bonus that generous, however, you might as well just go ahead and abolish the Electoral College. The number of electors will grow from 538 to 640, meaning the majority needed will increase from 270 to 321. Since the candidate who finished first in the popular vote will already have 102 electoral votes, he would need only 219 of the 538 awarded on a state-by-state basis – 40.7%. Such an outcome advances only slightly the idea that the president should have broad, and not just deep, support.

A smaller bonus promotes both goals. Suppose the winner of the popular were awarded a bonus equal to the number of electoral votes cast by the most populous state. Under the allocation in effect since the 2000 census, that would mean a bonus of 55 votes, California’s total. The new Electoral College would have 593 votes, and a candidate would need 297 to win the presidency. The first-place finisher in the popular vote would have 55 in his pocket, meaning that he would have to win 242, or 45%, of the 538 electoral votes awarded the old-fashioned way.

If you think a president who could have only secured 242 votes under the existing Electoral College still has a base too concentrated for the good of the republic, make the bonus smaller. If it’s 27 votes, one tenth of the number needed for victory today, the number of electors grows to 565, the majority needed for victory increases to 283, and the number a candidate would need to win on a state-by-state basis is 256, 47.6% of 538. If the bonus is only 11 votes, the mean number cast by the 51 states (including DC), then the magic number becomes 275 electoral votes; the winner of the national popular vote would still need to get 264 electoral votes by winning individual states, 49% of the total available by that avenue.

Even the 11-vote bonus would have given Al Gore a 278-271 victory in the 2000 election. Had John Kerry won Ohio in 2004, he would have secured an Electoral College majority of 272-266. The 11-vote national bonus, however, would have given Bush a 277-272 victory.

The national bonus modification of the Electoral College has the additional virtue of making the election of the president partly national and partly federal. When we cast our ballots, we would be voting in our dual capacity as Texans (or Virginians or New Yorkers), and also as Americans. It will prevent miscarriages of political justice, like the one that happened in 2000 and nearly happened in 2004, but also retain the Electoral College’s protections against political competition that becomes divisive and extreme.