The following is adapted from a speech delivered on April 30, 2019, at Hillsdale College’s Allan P. Kirby, Jr. Center for Constitutional Studies and Citizenship in Washington, D.C.

Once upon a time, the Electoral College was not controversial. During the debates over ratifying the Constitution, Anti-Federalist opponents of ratification barely mentioned it. But by the mid-twentieth century, opponents of the Electoral College nearly convinced Congress to propose an amendment to scrap it. And today, more than a dozen states have joined in an attempt to hijack the Electoral College as a way to force a national popular vote for president.

What changed along the way? And does it matter? After all, the critics of the Electoral College simply want to elect the president the way we elect most other officials. Every state governor is chosen by a statewide popular vote. Why not a national popular vote for president?

Delegates to the Constitutional Convention in 1787 asked themselves the same question, but then rejected a national popular vote along with several other possible modes of presidential election. The Virginia Plan—the first draft of what would become the new Constitution­—called for “a National Executive . . . to be chosen by the National Legislature.” When the Constitutional Convention took up the issue for the first time, near the end of its first week of debate, Roger Sherman from Connecticut supported this parliamentary system of election, arguing that the national executive should be “absolutely dependent” on the legislature. Pennsylvania’s James Wilson, on the other hand, called for a popular election. Virginia’s George Mason thought a popular election “impracticable,” but hoped Wilson would “have time to digest it into his own form.” Another delegate suggested election by the Senate alone, and then the Convention adjourned for the day.

When they reconvened the next morning, Wilson had taken Mason’s advice. He presented a plan to create districts and hold popular elections to choose electors. Those electors would then vote for the executive—in other words, an electoral college. But with many details left out, and uncertainty remaining about the nature of the executive office, Wilson’s proposal was voted down. A week later, Elbridge Gerry of Massachusetts proposed election by state governors. This too was voted down, and a consensus began to build. Delegates did not support the Virginia Plan’s parliamentary model because they understood that an executive selected by Congress would become subservient to Congress. A similar result, they came to see, could be expected from assigning the selection to any body of politicians.

There were other oddball proposals that sought to salvage congressional selection—for instance, to have congressmen draw lots to form a group that would then choose the executive in secret. But by July 25, it was clear to James Madison that the choice was down to two forms of popular election: “The option before us,” he said, “[is] between an appointment by Electors chosen by the people—and an immediate appointment by the people.” Madison said he preferred popular election, but he recognized two legitimate concerns. First, people would tend toward supporting candidates from their own states, giving an advantage to larger states. Second, a few areas with higher concentrations of voters might come to dominate. Madison spoke positively of the idea of an electoral college, finding that “there would be very little opportunity for cabal, or corruption” in such a system.

By August 31, the Constitution was nearly finished—except for the process of electing the president. The question was put to a committee comprised of one delegate from each of the eleven states present at the Convention. That committee, which included Madison, created the Electoral College as we know it today. They presented the plan on September 4, and it was adopted with minor changes. It is found in Article II, Section 1:

Each State shall appoint, in such Manner as the Legislature thereof may direct, a Number of Electors, equal to the whole Number of Senators and Representatives to which the State may be entitled in the Congress.

Federal officials were prohibited from being electors. Electors were required to cast two ballots, and were prohibited from casting both ballots for candidates from their own state. A deadlock for president would be decided by the House of Representatives, with one vote per state. Following that, in case of a deadlock for vice president, the Senate would decide. Also under the original system, the runner up became vice president.

This last provision caused misery for President John Adams in 1796, when his nemesis, Thomas Jefferson, became his vice president. Four years later it nearly robbed Jefferson of the presidency when his unscrupulous running mate, Aaron Burr, tried to parlay an accidental deadlock into his own election by the House. The Twelfth Amendment, ratified in 1804, fixed all this by requiring electors to cast separate votes for president and vice president.

And there things stand, constitutionally at least. State legislatures have used their power to direct the manner of choosing electors in various ways: appointing them directly, holding elections by district, or holding statewide elections. Today, 48 states choose their presidential electors in a statewide, winner-take-all vote. Maine and Nebraska elect one elector based on each congressional district’s vote and the remaining two based on the statewide vote.

It is easy for Americans to forget that when we vote for president, we are really voting for electors who have pledged to support the candidate we favor. Civics education is not what it used to be. Also, perhaps, the Electoral College is a victim of its own success. Most of the time, it shapes American politics in ways that are beneficial but hard to see. Its effects become news only when a candidate and his or her political party lose a hard-fought and narrowly decided election.

So what are the beneficial effects of choosing our presidents through the Electoral College?

Under the Electoral College system, presidential elections are decentralized, taking place in the states. Although some see this as a flaw—U.S. Senator Elizabeth Warren opposes the Electoral College expressly because she wants to increase federal power over elections—this decentralization has proven to be of great value.

For one thing, state boundaries serve a function analogous to that of watertight compartments on an ocean liner. Disputes over mistakes or fraud are contained within individual states. Illinois can recount its votes, for instance, without triggering a nationwide recount. This was an important factor in America’s messiest presidential election—which was not in 2000, but in 1876.

That year marked the first time a presidential candidate won the electoral vote while losing the popular vote. It was a time of organized suppression of black voters in the South, and there were fierce disputes over vote totals in Florida, Louisiana, and South Carolina. Each of those states sent Congress two sets of electoral vote totals, one favoring Republican Rutherford Hayes and the other Democrat Samuel Tilden. Just two days before Inauguration Day, Congress finished counting the votes—which included determining which votes to count—and declared Hayes the winner. Democrats proclaimed this “the fraud of the century,” and there is no way to be certain today—nor was there probably a way to be certain at the time—which candidate actually won. At the very least, the Electoral College contained these disputes within individual states so that Congress could endeavor to sort it out. And it is arguable that the Electoral College prevented a fraudulent result.

Four years later, the 1880 presidential election demonstrated another benefit of the Electoral College system: it can act to amplify the results of a presidential election. The popular vote margin that year was less than 10,000 votes—about one-tenth of one percent—yet Republican James Garfield won a resounding electoral victory, with 214 electoral votes to Democrat Winfield Hancock’s 155. There was no question who won, let alone any need for a recount. More recently, in 1992, the Electoral College boosted the legitimacy of Democrat Bill Clinton, who won with only 43 percent of the popular vote but received over 68 percent of the electoral vote.

But there is no doubt that the greatest benefit of the Electoral College is the powerful incentive it creates against regionalism. Here, the presidential elections of 1888 and 1892 are most instructive. In 1888, incumbent Democratic President Grover Cleveland lost reelection despite receiving a popular vote plurality. He won this plurality because he won by very large margins in the overwhelmingly Democratic South. He won Texas alone by 146,461 votes, for instance, whereas his national popular vote margin was only 94,530. Altogether he won in six southern states with margins greater than 30 percent, while only tiny Vermont delivered a victory percentage of that size for Republican Benjamin Harrison.

In other words, the Electoral College ensures that winning supermajorities in one region of the country is not sufficient to win the White House. After the Civil War, and especially after the end of Reconstruction, that meant that the Democratic Party had to appeal to interests outside the South to earn a majority in the Electoral College. And indeed, when Grover Cleveland ran again for president four years later in 1892, although he won by a smaller percentage of the popular vote, he won a resounding Electoral College majority by picking up New York, Illinois, Indiana, Wisconsin, and California in addition to winning the South.

Whether we see it or not today, the Electoral College continues to push parties and presidential candidates to build broad coalitions. Critics say that swing states get too much attention, leaving voters in so-called safe states feeling left out. But the legitimacy of a political party rests on all of those safe states—on places that the party has already won over, allowing it to reach farther out. In 2000, for instance, George W. Bush needed every state that he won—not just Florida—to become president. Of course, the Electoral College does put a premium on the states in which the parties are most evenly divided. But would it really be better if the path to the presidency primarily meant driving up the vote total in the deepest red or deepest blue states?

Also, swing states are the states most likely to have divided government. And if divided government is good for anything, it is accountability. So with the Electoral College system, when we do wind up with a razor-thin margin in an election, it is likely to happen in a state where both parties hold some power, rather than in a state controlled by one party.

Despite these benefits of the current system, opponents of the Electoral College maintain that it is unseemly for a candidate to win without receiving the most popular votes. As Hillary Clinton put it in 2000: “In a democracy, we should respect the will of the people, and to me, that means it’s time to do away with the Electoral College.” Yet similar systems prevail around the world. In parliamentary systems, including Canada, Israel, and the United Kingdom, prime ministers are elected by the legislature. This happens in Germany and India as well, which also have presidents who are elected by something similar to an electoral college. In none of these democratic systems is the national popular vote decisive.

More to the point, in our own political tradition, what matters most about every legislative body, from our state legislatures to the House of Representatives and the Senate, is which party holds the majority. That party elects the leadership and sets the agenda. In none of these representative chambers does the aggregate popular vote determine who is in charge. What matters is winning districts or states.

Nevertheless, there is a clamor of voices calling for an end to the Electoral College. Former Attorney General Eric Holder has declared it “a vestige of the past,” and Washington Governor Jay Inslee has labeled it an “archaic relic of a bygone age.” Almost as one, the current myriad of Democratic presidential hopefuls have called for abolishing the Electoral College.

Few if any of these Democrats likely realize how similar their party’s position is to what it was in the late nineteenth century, with California representing today what the South was for their forebears. The Golden State accounted for 10.4 percent of presidential votes cast in 2016, while the southern states (from South Carolina down to Florida and across to Texas) accounted for 10.6 percent of presidential votes cast in 1888. Grover Cleveland won those southern states by nearly 39 percent, while Hillary Clinton won California by 30 percent. But rather than following Cleveland’s example of building a broader national coalition that could win in the Electoral College, today’s Democrats would rather simply change the rules.

Anti-Electoral College amendments with bipartisan support in the 1950s and 1970s failed to receive the two-thirds votes in Congress they needed in order to be sent to the states for consideration. Likewise today, partisan amendments will not make it through Congress. Nor, if they did, could they win ratification among the states.

But there is a serious threat to the Electoral College. Until recently, it has gone mostly unnoticed, as it has made its way through various state legislatures. If it works according to its supporters’ intent, it would nullify the Electoral College by creating a de facto direct election for president.

The National Popular Vote Interstate Compact, or NPV, takes advantage of the flexibility granted to state legislatures in the Constitution: “Each State shall appoint, in such Manner as the Legislature thereof may direct, a Number of Electors.” The original intent of this was to allow state legislators to determine how best to represent their state in presidential elections. The electors represent the state—not just the legislature—even though the latter has power to direct the manner of appointment. By contrast, NPV supporters argue that this power allows state legislatures to ignore their state’s voters and appoint electors based on the national popular vote. This is what the compact would require states to do.

Of course, no state would do this unilaterally, so NPV has a “trigger”: it only takes effect if adopted by enough states to control 270 electoral votes—in other words, a majority that would control the outcome of presidential elections. So far, 14 states and the District of Columbia have signed on, with a total of 189 electoral votes.

Until this year, every state that had joined NPV was heavily Democratic: California, Connecticut, Hawaii, Illinois, Maryland, Massachusetts, New Jersey, New York, Rhode Island, Vermont, and Washington. The NPV campaign has struggled to win other Democratic states: Delaware only adopted it this year and it still has not passed in Oregon (though it may soon). Following the 2018 election, Democrats came into control of both the legislatures and the governorships in the purple states of Colorado and New Mexico, which have subsequently joined NPV.

NPV would have the same effect as abolishing the Electoral College. Fraud in one state would affect every state, and the only way to deal with it would be to give more power to the federal government. Elections that are especially close would require nationwide recounts. Candidates could win based on intense support from a narrow region or from big cities. NPV also carries its own unique risks: despite its name, the plan cannot actually create a national popular vote. Each state would still—at least for the time being—run its own elections. This means a patchwork of rules for everything from which candidates are on the ballot to how disputes are settled. NPV would also reward states with lax election laws—the higher the turnout, legal or not, the more power for that state. Finally, each NPV state would certify its own “national” vote total. But what would happen when there are charges of skullduggery? Would states really trust, with no power to verify, other state’s returns?

Uncertainty and litigation would likely follow. In fact, NPV is probably unconstitutional. For one thing, it ignores the Article I, Section 10 requirement that interstate compacts receive congressional consent. There is also the fact that the structure of the Electoral College clause of the Constitution implies there is some limit on the power of state legislatures to ignore the will of their state’s people.

One danger of all these attacks on the Electoral College is, of course, that we lose the state-by-state system designed by the Framers and its protections against regionalism and fraud. This would alter our politics in some obvious ways—shifting power toward urban centers, for example—but also in ways we cannot know in advance. Would an increase in presidents who win by small pluralities lead to a rise of splinter parties and spoiler candidates? Would fears of election fraud in places like Chicago and Broward County lead to demands for greater federal control over elections?

The more fundamental danger is that these attacks undermine the Constitution as a whole. Arguments that the Constitution is outmoded and that democracy is an end in itself are arguments that can just as easily be turned against any of the constitutional checks and balances that have preserved free government in America for well over two centuries. The measure of our fundamental law is not whether it actualizes the general will—that was the point of the French Revolution, not the American. The measure of our Constitution is whether it is effective at encouraging just, stable, and free government—government that protects the rights of its citizens.

The Electoral College is effective at doing this. We need to preserve it, and we need to help our fellow Americans understand why it matters.

[This post originally appeared on Imprimis, a publication of Hillsdale College]