The electoral college is a unique and somewhat arcane institution, but its history and function are not mysterious. The electoral process is embedded in the Constitution and its structure specifies precisely how much influence each state can have on the election of the president of the United States.

Suggestions to abandon the electoral vote in favor of a national popular vote are not new and, for reasons discussed here, have never come close to implementation. Recently, however, a strategy has been devised to effectively nullify the electoral college: the so-called National Popular Vote Interstate Compact. Constitutional scholars, state lawmakers, and voters who are contemplating this strategy should understand its implications.

538 electors / First, a little history. The Constitution directs that the number of electors for each state shall equal the “Number of Senators and Representatives to which the State may be entitled in the Congress.” Every state has two senators and at least one representative, so every state has at least three electoral votes.

The number of representatives was fixed at 435 by the Apportionment Act of 1911; that number was affirmed by the Permanent Apportionment Act of 1929. The 23rd Amendment (ratified in 1961) assigns three electoral votes to the District of Columbia. Unless a new state is added to the realm or Congress revises the number of representatives—something it has not done in over 100 years—the current total of 538 electoral votes will prevail. A simple majority—270 votes—is required to win the election.

So how are the 385 additional representatives, beyond the minimum of one per state, apportioned among the states? As specified in the Constitution, the apportionment process occurs every 10 years, based on state populations as determined by the official decennial census. The current apportionment method, in place since 1941, is credited to the mathematician Edward Huntington and to Joseph Hill, chief statistician of the U.S. Census Bureau. The Huntington–Hill method, also known as the “equal-proportions” method, applies a straightforward mathematical weighting formula to determine how many representatives each state shall have. Adoption of the Huntington–Hill method does not appear to have been controversial; it simply made formulaic the ad hoc methods that had been applied over the preceding decades.

One attribute of the electoral college’s design is apparent. The method for apportioning representatives among the states, and consequently determining the number of electoral votes, ensures that small states (as measured by population) will have shares of the electoral vote that are greater than their shares of the national popular vote. Conversely, large states will have electoral vote shares that are less than their shares of the national popular vote. For example, the smallest state, Wyoming, with one congressional representative, has 0.6% of the electoral vote and 0.2% of the U.S. population. At the other end of the size distribution, California, with 53 representatives, has 10.2% of the electoral vote and 12.0% of the U.S. population.

In 2016, 32 states and the District of Columbia had an electoral vote share greater than their population share. Those states have a greater potential influence on the electoral vote than they would have on the national popular vote. This is not a new pattern. Throughout U.S. history, the majority of states—the smaller ones—have had an electoral share greater than their population share.

Compared to a national popular vote, a small state is likely to prefer the electoral college system because of the relative advantage conferred. Any state, large or small, may well prefer the system that is not dominated by a handful of the largest states. (When considering what system a state “prefers,” I am referring to what the voters of the state are likely to prefer in general, without consideration of the outcome in any particular election.) Because a majority of states are likely to prefer the electoral college, it has never been susceptible to removal by constitutional amendment, which requires the assent of not just a majority but three-fourths of the states.

Overruling state voters / Recently, however, another consideration has come into play. Several states have joined the National Popular Vote Interstate Compact (NPVIC). The NPVIC is an agreement, among the states that join, to award all of their electoral votes to the presidential candidate with the highest share of the national popular vote. The agreement goes into effect if enough states join the NPVIC to account for 270 electoral votes, the number needed to win the election. As of now, the NPVIC includes 12 states and the District of Columbia. Together, they account for 181 electoral votes.

Is the NPVIC in accordance with the Constitution? Article I, Section 10, Clause 3 declares, “No State shall, without the Consent of Congress, … enter into any Agreement or Compact with another State.” It seems clear, therefore, that the NPVIC would require congressional approval to become operational. But, as someone once said, in law nothing is certain but the expense. Let’s put this issue aside for present purposes.

Article II, Section 1, Clause 2 of the Constitution, already quoted in part, declares, “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.” Accordingly, it is for the states to direct how their electors will vote. Every state—except Maine and Nebraska, which split their electoral votes—has determined that all of its electors shall vote for the winner of the popular vote in the state; winner take all. The dominance of a winner-take-all allocation should not be surprising. Each state would presumably want to maximize its influence on the election. I cannot explain Maine and Nebraska.

Since the Constitution gives each state the exclusive right to assign its electors as it sees fit, joining the NPVIC could be seen as just another way for a state to assign its electors. Let us consider, however, the possible outcomes if NPVIC membership grows to account for 270 electoral votes, so that it becomes operative.

If the winner of the national popular vote would have won the electoral vote (absent the NPVIC), then the NPVIC would not affect the outcome of the election. Nonetheless, an NPVIC state that voted for the loser of the national popular vote would have its electors assigned to the winner of the national popular vote. The majority of that state’s voters might well feel cheated, but the outcome of the election would not be affected by the state’s participation in the NPVIC.

What happens if the winner of the national popular vote would have lost the electoral vote (absent the NPVIC), as happened in the most recent election? If the winner of a state’s popular vote also wins the national popular vote, then that candidate already has the state’s electoral votes; participation in the NPVIC would not change the state’s electoral vote. But if the winner of a state’s popular vote loses the national popular vote, then that state’s electors would be shifted from the state’s popular vote winner to the national popular vote winner. That, along with the shift of other NPVIC states’ electoral votes, would change the outcome of the election. The majority of the state’s voters would definitely feel cheated.

Joining the NPVIC means that the assignment of a state’s electoral votes in a future election might not be in accord with the will of the majority of the state’s voters. That outcome is, indeed, the entire and explicit purpose of the NPVIC. If the winner of a state’s popular vote is the loser of the national popular vote, then the electors will be allocated contrary to the majority of that state’s voters. By joining the NPVIC, therefore, a state commits itself to disenfranchising its own voters in some future election.

Regardless of the legal issues that raises, why would any state join a system that is intended to allocate electors contrary to the preferences of the state’s voters? There are several possible reasons. Some of the largest states might join because they see the NPVIC as an effective repeal of the electoral college—which it is—and they prefer the national popular vote for the reasons discussed above. Some small states might join because they see the NPVIC as a way of avoiding a replay of the 2016 outcome; those would be the small states that see themselves as politically aligned with the largest states. Indeed, participation in the NPVIC so far reflects these explanations. Large states California, New York, and Illinois have joined, and all of the joining states voted for Clinton over Trump in 2016.

Nullifying the Constitution and the will of state voters / The electoral college does not correlate perfectly with the national popular vote. That is not a flaw; that is its purpose. And there is a powerful reason for the persistence of the electoral college: a majority of states prefer it, which has been the case throughout U.S. history.

If the NPVIC becomes operational, two outcomes would result. First, the only vote tabulation of consequence would be the national popular vote. The electoral college would be effectively nullified without constitutional amendment, likely by an agreement among a minority of the states. Second, NPVIC states would be enacting a system designed, under specific circumstances in a future election, to disenfranchise their own voters.

It is important to remember that the distribution of congressional representatives (and consequently of electors) was not happenstance, but rather the result of purposeful compromises among the states, in consideration of the wide variation in state populations. The Constitution and subsequent laws embody several such accommodations, so as to prevent a small number of large states from dominating political outcomes. As one of those accommodations, the electoral college has proven to be durable, so far.