Given that Hillary Clinton is on pace to outperform Donald Trump in the national popular vote but lose the electoral vote (and the presidency), this currently-meaningless scenario is renewing calls to alter or abolish the Electoral College. The most pressing plan is the National Popular Vote, a compact between the states to award their electoral votes to the winner of the national popular vote once 270 electoral votes' worth of states agree to do so--effectively circumventing the Electoral College and using the selection of electors on a nationwide rather than statewide basis. Having recently been approved in New York, 165 electoral votes' worth of states have approved the plan.
I want to set aside the issue of whether the Electoral College should be (actually or effectively) abolished for a moment. There are some good theoretical grounds, I think, why the Electoral College--at least, a system designed to recognize individual states and all residents in a state as a proxy for political power rather than simply raw voter totals--retains some (admittedly, imperfect!) merit. Instead, this argument will focus exclusively on the means of the National Popular Vote, as a legal matter and as a practical matter. On both grounds, I think it falls quite short.
As a legal matter, I have written extensively that the compact is unconstitutional absent congressional consent. States are prohibited from entering into interstate compacts with one another without the consent of Congress. The Supreme Court has slowly carved out exemptions to this provision and now (or most recently) only requires consent for compacts that affect the balance of power between the federal and state governments or among the several states. The decision of some states to change the balance of power among presidential electors--essentially, prior to an election, ensuring that non-compacting states' electors are irrelevant to the presidential election--is the kind of shift of power among states that requires congressional consent under even the most generous construction of the Compact Clause. As this process was designed to avoid Congress--and because I think congressional consent is unlikely in any event--the compact would fail. (More details can be dug out of those articles.)
There have been other concerns raised by other commentators--that the compact would improperly strip the House of its power to choose a president in a contingent election when no candidate secured a majority; that the state legislature's plan to award electors on a basis other than the decision of the people in that state is prohibited; that the compact may run afoul of the Voting Rights Act for certain jurisdictions with sufficiently minority populations.
As a practical matter, the decision to change presidential elections at a state level without including a uniform national plan for elections, or empower Congress to do so, is deeply problematic. A recent, and quite significant, effort to amend the Constitution took place in 1970, and even it fell short of the likely required federal power we would need to regulate presidential elections. A nice summary from CQ Almanac shows some of the things a federal amendment was designed to do. For instance, the constitutional amendment guaranteed some uniformity in voter qualifications:
"The electors of President and Vice President in each State shall have the qualifications requisite for electors of the most numerous branch of the State legislature, except that for electors of President and Vice President, the legislature of any State may prescribe less restrictive residence qualifications and for electors of President and Vice President the Congress may establish uniform residence qualifications."
Additionally, the proposed amendment provided a times, places, and manner provision for presidential elections, similar to such a provision for congressional elections:
"The times, places, and manner of holding such elections and entitlement to inclusion on the ballot shall be prescribed in each State by the legislature thereof; but the Congress may at any time by law make or alter such regulations.
But there are no such provisions for the National Popular Vote. I've written about the "invisible federalism" that exists within the Electoral College. Right now, fifty-one jurisdictions set their own rules. That means a true national popular vote would require some significant centralization of our electoral process, some of which was baked into the 1970 proposed constitutional amendment. Consider some things that ought to be nationalized in a presidential election:
Uniform voter qualifications. States have some fairly dramatic differences in the qualifications for voters. Some bar anyone ever convicted of a felony from voting, others permit felons imprisoned to vote. Some in the future may lower their voting age, as they have done in the past. Some have different standards for the mentally ill. When faced with such an option to the proposed amendment, Congress rejected it; but it would likely need to be an element of any future effort.
Uniform voting procedures. States set different absentee and early voting requirements. They set different residency requirements. Some have voter identification laws and others don't. Some open their polls at different hours. Some have all mail-in elections. Recount procedures trigger differently in different states. The proposed constitutional amendment got part of the way there in permitting Congress to regulate such procedures if it deemed such laws necessary to preserve uniformity.
Uniform ballot access standards. Even more so in an election like the one that took place in 2016, state permit different candidates to appear on the ballot. Evan McMullin and Jill Stein appeared on the ballot in a handful of jurisdictions. States like Texas have extremely early ballot access deadlines for independent candidates. Indeed, an effort to keep Donald Trump off the ballot in Minnesota would have wreaked havoc in a national popular vote total.
Standard ballot content. Consider, for instance, California including a dozen ballot propositions on its presidential ballot, boosting turnout; or hotly-contested United States Senate races, which influence turnout. We may see some oddities in turnout simply because of the other races that happen to appear on the ballot--and if we want a truly national election, we ought to aspire to more uniform standards.
A requirement to hold a popular election in the first place. And under the current system, states are not even required to hold a popular election in the first place. A state legislature could strip its citizens of the right to vote in presidential elections if it so desires, something like many states near the Founding did, or like Colorado did in 1876. Of course, such a decision might cut off the nose to spite the face... but remains within the realm of possibility. (A version of this challenge would be a state's future decision to use instant run-off voting, which offers different challenges in tabulating the votes from that jurisdiction--in a true national election, we would need to decide whether first past the post or an alternative system would be used. This is hypothetical at the moment because every state currently uses first past the post.)
One rejoinder I've heard goes something like this: "Yes, the votes are all commingled, but that doesn't really mean anything. After all, the electoral votes are all commingled in the final totals, too, even though there are separate elections that have occurred using different systems."
But I've never understood this rejoinder. It's something like saying, "Yes, I understand you planned to eat roast beef, mashed potatoes, green peas, and apple pie tonight. But because they're all going to end up in your stomach anyway, there's no difference if we simply blended them all together before consuming them."
You see, our present electoral system holds fifty-one separate contests. States can do whatever they want within their electoral process--they're going to get a fixed number of electoral votes, and whatever they do to determine who gets those votes (or, more precisely, who the electors will be) is up to each state. Yes, at the end of the day, the votes are combined, but only after the results have been limited to the boundaries of that particular state.
Now, it's also the case that many elections are administered at the county level, and those county-level decisions affect statewide elections--the format of the ballot, the type of ballot used (optical scan or direct-recording electronic voting machine), local races, the training for poll works, and so on. And perhaps we have to accept some lack of uniformity in our electoral process simply because of the expansive country we live in. But it may be worth thinking carefully about how uniform our elections should look (perhaps drawing inspiration from other countries that maintain both national and local election systems), and what that ought to look like in a constitutional amendment. The National Popular Vote offers absolutely no opportunity to ask these questions.
Finally, I've mentioned before that we tend to prefer majority winners in our electoral system, and we have seen that the Electoral College can turn fairly small margins into rather lopsided victories, a kind of affirmation for the president (mostly, I think, a happy and incidental effect of the system, not a part of its design). Only some presidents secure more than 50% of the popular vote but all secure a majority of the electoral vote (or the House chooses the president by a majority of states). A runoff or some guarantee that the winner gets 50% of the vote--or a nationalized instant run-off process. But the National Popular Vote makes no such guarantee. Perhaps we are ready to say that the person who gets the most votes wins, as we often do in other elections where a plurality winner takes office. But it is worth considering whether that, too, should be a part of the conversation of a new presidential election system--something that simply cannot be accomplished under the National Popular Vote.
It may well be that Americans are ready for a new system of electing the president, the normative or theoretical reasons a matter beyond the scope of this post. But in doing so, the National Popular Vote is a fairly ineffective way of doing so, and, as a cure for the problems perceived by many, may ultimately be worse than the disease.