Julie Jacobson / Reuters

Twice this century—and we are only two decades in—the person who has won the White House has received fewer voters nationwide than his opponent. As a direct result, more and more people are interested not just in who should be president but also in how that person ought to be selected.

One of the most significant reform proposals—what’s known as the National Popular Vote (NPV) interstate-compact plan—has gained momentum. As of today, 15 states and the District of Columbia have joined the agreement. As signatories, each jurisdiction pledges to select Electoral College members who support the presidential candidate who won the most votes nationally, regardless of which candidate won the most votes in that particular jurisdiction. By its terms, the compact will go into effect only when states totaling at least 270 Electoral College votes—the number needed to win the White House—enact the plan. So far, jurisdictions accounting for 196 votes—more than 70 percent of the needed 270—have signed on.

The idea is to get the country closer to having a national popular election for president within the current constitutional framework and without the need for a constitutional amendment. Under this system, the Electoral College would still exist, and it would still pick the president. As a constitutional matter, Article II gives each state the power to select its Electoral College members—what the Constitution calls “electors”—by whatever means the state chooses. It is perfectly within a state’s authority to decide that national popularity is the overriding substantive criterion by which a president should be chosen. And there is no constitutional problem with a state using other states’ voting tallies, even if the states have different voting rules and ballot forms. As long as each state treats people within its own borders equally, there is no equal-protection issue.

In this way, the NPV retains the constitutional device of the Electoral College but also moves the country closer to a less convoluted national vote. As a policy matter, the NPV helps address the two biggest criticisms of the status quo. The first criticism is that the current system does not respect the one-voter, one-vote principle, because voters across the country don’t have an equal voice in selecting a president. The second is that the margin of victory or loss within a state is irrelevant, given nearly every state’s winner-take-all approach to allocating presidential electors, so presidential campaigns are not national at all. Instead, they are focused on a handful (anywhere from four to eight) of reasonably sized “swing” states where the median voter is in play.

The NPV could really work, and I’ve been writing and talking about it (or some variant of it) since 2001, when my brother (and fellow law professor) Akhil Amar and I—and, separately, another law professor, Robert Bennett—laid out the plan’s intellectual foundations. Twenty years ago, probably only a few people expected states to act on these ideas. But now, thanks in large part to an organization headed by the Stanford mathematician John Koza, the plan is moving forward.

Of course, there are critics. Some people oppose the NPV because they think it will lead to more third-party candidacies. Leaving aside for the moment whether third-party candidacies are undesirable in the first place, there’s no reason to think the NPV would increase them. All states—even large states like California, Texas, and Florida—use the popular vote, rather than a state-level electoral college, to determine gubernatorial races, yet meaningful third-party candidacies are no more common in governors’ races than in presidential contests.

Nor is the NPV likely to cause insurmountable recount difficulties; large states handle recounts regularly, and so could the nation. Moreover, often (as in 2000), the national-vote winner will be clear despite claims that vote counts in particular states are flawed. Indeed, an NPV system makes recounts less likely in the first place, as elections would not turn on a handful of close calls. For example, the 2000 Florida recount would have been irrelevant, since no one denies that Al Gore had more votes nationally than George W. Bush no matter what the exact outcome in Florida was.

But while the plan is a smart one, there is always a risk of unforeseen and undesirable collateral consequences. The Electoral College, though far from perfect, has served the nation reasonably well, and change always brings the risk of glitches. This is why the details about how to implement the NPV matter greatly, and why I’ve advocated for what I call a “deferred-implementation approach.” Such a strategy would mean that the NPV would not go into effect until three presidential election cycles after the requisite number of states had signed on to it, which would give Congress enough time to figure out how to patch any holes in the plan.

Polls suggest that many Republican leaders (though not necessarily rank-and-file Republican voters) oppose the NPV, possibly because they fear it will hurt their party’s chances of winning presidential elections, or because they think NPV supporters have partisan motives. Partisanship should neither drive nor stymie principled electoral reforms, of course, but, in fact, NPV has no partisan skew. In 2000, many analysts expected Bush to win the nation’s popular vote and lose the Electoral College, and in 2004, Bush beat John Kerry soundly in the national-vote tally but would have lost the Electoral College vote, and thus the White House, had 60,000 voters chosen differently in Ohio.

Moreover, no one can easily know which candidate would have won past presidential elections had the NPV plan been in effect, because the candidates would have campaigned differently. For example, it is possible that Donald Trump could have run up the margin of victory in some solidly red states and narrowed the margin of loss in some deep-blue ones.

To the extent that, notwithstanding the factual and historical data, many people still view the NPV as a partisan trick to benefit Democrats, a phased-in approach to implementation should help. Setting the plan’s effective date so far into the future makes it impossible for either side to reasonably know what the partisan consequences will be after full implementation. Red-state voters and elected officials who believe that all votes should count equally could support this delayed implementation without feeling that it will necessarily hurt their side.

Perhaps the greatest risk for “glitches” arises from the fact that there is substantial nonuniformity among the states that are included in the national-vote count on the questions of who votes, how votes are cast, and how votes are counted and recounted. This lack of uniformity undermines the normative appeal of the move to a national popular vote and also raises the specter of electoral crises that must be avoided. My brother and I noted—indeed, highlighted—these problems and identified workable solutions when we first analyzed the idea of state-level movement toward direct election almost two decades ago.

But drafters of the NPV plan in play today did not incorporate any of the solutions to state nonuniformity that we proferred; they did not build into the plan uniform rules of voting eligibility, uniform presidential ballots, and an election-dispute procedure. They also didn’t delegate authority to do so to a nongovernmental commission in the way we suggested could be done. Nor did they affirmatively invite Congress to step in. But Congress will need to do so. If and when the NPV comes into being, Congress should supplement it with a system of uniform rules for vote tallying in all 50 states.

Delayed implementation would also direct attention to where it should be—in the future, not the past. The historical evidence overwhelmingly suggests that placating Southern states and the institution of slavery was a primary driver of Electoral College design, but even if that were not the case, the modern arguments against the Electoral College should carry the day. An innocent original intent would not cure the problems with the Electoral College in modern American democracy, any more than such innocent origins cure other defects of the original design (such as the indirect election of senators) that have since been undone by formal or informal amendment. The NPV can cure us, in due time, of the Electoral College—something needed not because of the institution’s poisonous beginnings, but because of its poisonous effects.

This story is part of the project “The Battle for the Constitution,” in partnership with the National Constitution Center.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.