In Michigan, the National Popular Vote Interstate Compact (NPVIC) has qualified to begin gathering signatures to appear on the ballot next year. It appears likely the petition will succeed, since it enjoys the backing of prominent and experienced Democrats.

The basic premise of the NPVIC is to have states each pass an identical piece of legislation under which they would award their electoral votes to the candidate who receives the most popular votes nationwide. The states would implement this plan once enough states controlling 270 electoral votes have signed on, guaranteeing that the national popular vote winner will win the presidency. Currently, the NPVIC has been adopted by 15 states plus the District of Columbia, together controlling 195 electoral votes.

The NPVIC is not a new idea. Cato’s John Samples criticized its constitutional and policy merits in a 2008 policy analysis. But in light of the crisis over the 2020 election and the prospect of a repeat or worse in 2024, it’s worth taking a skeptical look at the nuts-and-bolts mechanics of how the NPVIC would work.

There are lively debates about if a national popular vote is desirable at all, and if the National Popular Vote Interstate Compact is a constitutional way to get there. But even for those who support a national popular vote and think the NPVIC clears all the constitutional hurdles, there’s a more concrete practical problem which has been neglected by the plan’s drafters.

The problem is best summed up by a simple question: Who won the national popular vote in 1960?

The usual answer, and the figure you’ll find reported in most places, is that John F. Kennedy narrowly defeated Richard Nixon, 34,220,984 votes to 34,108,157 votes. “I didn’t pay for a landslide,” Joe Kennedy, JFK’s wealthy and well-connected father, supposedly remarked. But the reality is not so simple, even aside from credible claims of corruption and fraud in key states.

The ambiguity comes from a single state which chose an unusual system for picking its presidential electors. Instead of voters choosing from slates of electors pledged to a particular party’s ticket—the system we’re used to in most states today—Alabama’s legislature chose a different method. Each of Alabama’s eleven members of the Electoral College were chosen individually by voters statewide, with each voter allowed to pick any eleven candidates. The names Kennedy and Nixon did not actually appear on the ballot.

Each party could nominate eleven candidates for the state’s eleven elector seats. Driven by opposition to civil rights, Alabama Democrats nominated five Kennedy electors together with six who were unpledged to any candidate. All eleven Democratic electors won, with each of them garnering more votes than any of the eleven Republican candidates, but the number of votes they received varied.

The six unpledged Democratic electors ended up voting for segregationist Sen. Harry Byrd of Virginia, along with all of Mississippi’s electors and one from Oklahoma. Their intent was to deny either major-party candidate a majority in the Electoral College, letting conservative southern Democrats play kingmaker when the election was thrown to the House under the Twelfth Amendment. Kennedy still won a majority without them, making it moot. Similar unsuccessful attempts were made in 1948 and 1968 by segregationist candidates Strom Thurmond and George Wallace.

Under the system as it stood in 1960, and still stands today, that was that. How things would have played out under the NPVIC plan is not so clear. In fact, it’s hopelessly ambiguous.

If the total number of voters who voted for Kennedy electors and the unpledged electors are divided proportionally between them, the result is that Nixon actually won the national popular vote by about 50,000 votes. This was the method used by Congressional Quarterly at the time, the source usually taken as the closest thing we have to an official national popular vote tally. Another way of interpreting the result allocates to Kennedy the vote total of his most popular elector, which has him winning the national popular vote. The problem is this method effectively double-counts voters who voted for both the Kennedy electors and the unpledged electors.

The bottom line is there’s no objectively correct answer for which way to count Alabama’s popular votes in 1960. And because of that, there’s no way to say for sure if Nixon or Kennedy really won the national popular vote.

This might seem just a bit of obscure electoral trivia, and indeed it didn’t change the outcome of that year’s election. It’s also unlikely the exact circumstances which led Alabama Democrats to nominate a mixed slate of electors will be repeated any time soon. But it highlights a hidden weakness in the NPVIC: It has no way to handle a scenario like 1960. And that precedent points the way for any one state, even a state not participating in the Compact, to render the whole scheme inoperable.

The Compact’s language simply assumes the existence of a traditional popular vote total in each state but it provides no details on how that is to be ascertained. This hasn’t gone unnoticed by opponents. It opens the door for a state to throw a wrench in the works by adopting alternative methods for choosing their electors, methods which are not conducive to producing an unambiguous national popular vote total.

Republicans state legislators, who mostly oppose using the national popular vote, have already figured that out. In North Dakota, the Republican-controlled state senate passed a bill saying their state will withhold its popular vote totals for president until after the Electoral College has voted in December. Instead, the state would only publish the rough percentages. This is deliberately aimed at making it impossible to properly calculate the national popular vote total in time to award electors on that basis. Similar bills have been introduced in other states.

The North Dakota proposal is subject to fair criticism on transparency grounds. Refusing to release the precise vote totals seems unlikely to inspire confidence and trust. Indeed, the state House amended that proposal out of the bill, instead passing a statement urging Congress to oppose the NPVIC and studying other options for the state to block it. But that’s just one example of the ways in which non-participating states could make the operation of the NPVIC impossible, at least as it’s currently written.

Another unintentional example is Maine, where Democrats have successfully implemented ranked choice voting for federal elections, including for president. Under that system, each voter’s one vote can be reassigned to different candidates through multiple rounds of tabulation until one candidate has an absolute majority. The NPVIC does not specify if votes on the first round or the last round should be counted, and either one would defeat the purpose of ranked choice voting.

Maine’s system raises another possible perverse incentive under the NPVIC: it might reward a candidate for doing worse. If a candidate wins an outright majority of over 50% of the first-choice ballots, the tabulation stops then. But if they fall short of 50% and the votes for lower-ranked candidates are then redistributed, the winning candidate might accumulate many thousands more votes in the final round. If those final-round votes are what’s counted for the NPVIC, it could be a huge boon for a candidate to not do too well on first choice votes, so that they get to add those second- and third-choice votes to their national total.

These are just some examples of how the NPVIC could be confounded by variations in state election procedures. Creative and motivated legislators could undoubtedly come up with many more, and they surely would if the NPVIC ever nears its 270 electoral-vote trigger. For example, states could simply take the names of the presidential candidates off the ballot, letting voters pick only between a “Republican” and “Democratic” slate of elector candidates. While this ruse would be transparent, with everybody knowing which party is backing which candidates, it’s not clear that the NPVIC would have a way to count those votes. Worse, there’s no guarantee all of the states in the Compact would reach the same conclusion. 

The NPVIC’s authors could have provided some mechanism for adjudicating this problem, either discounting a state’s votes altogether in such cases or else providing some way to conclusively determine how many votes to count for each candidate. But the Compact doesn’t do that, nor does it have any way for states to harmonize their decision, leaving it up to the election authorities in each state to come up with their own interpretation. Instead of letting voters pick the president, the NPVIC could effectively hand the decision over to a handful of little-known (and partisan!) secretaries of state deciding how to read a vague law with no obvious answer. That’s a recipe for disaster.

In a worst-case scenario, different states in the Compact could make conflicting decisions, throwing the entire scheme into chaos. The only way to fix that now would be to go back and have each state adopt an amended NPVIC law filling in the details, an arduous hurdle. That prospect becomes even more daunting if states begin adopting the NPVIC by popular referendum, leaving their legislatures unable to easily make needed changes.

If the trigger for the NPVIC is ever reached, this practical problem could make the Compact effectively inoperable before the courts even get the chance to rule on questions of constitutionality. Worse, it could throw the outcome of a presidential election into doubt without any clear rules for determining who rightfully won.

In an era when disputed elections are becoming more common and where the laws governing presidential elections are already riddled with dangerous ambiguities, the NPVIC is an invitation to a constitutional crisis. With the world’s most powerful office at stake, America’s presidential election procedures need certainty and stability. The National Popular Vote Interstate Compact threatens to bring us the exact opposite, an election in which the rules would be extremely uncertain and unsettled.

Rather than risking such a high-stakes catastrophe, opponents of the Electoral College should accept there’s no easy shortcut around the difficult task of amending the Constitution.