Let us be quite clear from the beginning: The Democratic Party deliberately nominated a candidate who was undergoing federal investigation for matters under national security law, and who was associated with a charity, the Clinton Foundation, that was also under federal investigation. The Clinton campaign actively worked to make sure that Donald Trump would get the Republican nomination and be on the ballot. Clinton then ran a campaign heavily governed by an algorithm. In the meantime, both the campaign and the party failed to make any serious attempt to re-integrate Sanders supporters, despite their vehement complaints about her tactics during the primary campaign. The campaign repeatedly assumed that it could turn out blacks and Latinos in the neighborhood of Obama's turnout while doing almost nothing specific to help local groups make it happen; and she passed over several Latinos to pick as her running mate the weak and barely helpful Tim Kaine. They did next to nothing shoring up the Democratic Party in Wisconsin despite the fact that it was known that the state party there was in disarray, and despite warnings that the entire Rust Belt was in imminent danger of being captured. One could make the list much, much longer. If one wishes to find something to blame for the rise of Donald Trump, one need not look beyond a campaign whose chief characteristic was the arrogance of complacent incompetence.
But a tough election leaves the losing side flailing around for a ground of complaint, and when there were a lot of close races, they look in particular for a reason to complain that they did not lose fairly. So, as the returns on individual races slowly reach more precise numbers, it was perhaps inevitable that their flailing would lead to an attack on the big target, the Electoral College. Such is Andrew Prokop's recent Vox essay, which handily brings together almost all of the false things people often say about the venerable institution.
Let's start with an essential issue, however, and one that is so often overlooked in these discussions that it needs to be highlighted. The first and most important rule for adding up numbers in the real world is that only relevant commensurable units can be added. If I add 5 degrees Celsius to 6 oak trees, I don't have 11 of anything real. Unit-wise, adding degrees Celsius to oak trees requires taking the unit of addition to be degree-Celsius-or-oak-tree, which is nothing real whatsoever. Degrees Celsius and oak trees are not commensurable; they have no common measure.
The reason this point needs to be remembered at the outset is that the United States has no mechanism for generating a generic vote across state lines. Votes are artificial things that are created by specific processes constructed on purely conventional criteria set by law. They have no other existence. I can fill out a ballot and bury it in the yard, and I have not voted, because I have not undergone the right process; I can scream out that I support someone for President and I have not voted, because I have not undergone the right process. I can write-in a vote for U.S. President in an election in Norway and while I have voted, my vote is not a vote in the same process and is not commensurable to the votes in the American election. But in the United States votes in a Presidential election are created by processes established by state laws, which vary from state to state, sometimes in considerable ways. Votes are not commensurable across state lines. The difference, of course, is not territorial but jurisdictional. When you vote, you do not create a generic American vote, you create a vote-in-California or a vote-in-Maine, or the like. We call them both votes because they are created by a voting process, and this is helped by the fact that there is a lot of similarity between the voting processes due to historical tradition, but we do not create the same kind of vote if we are voting in different states. Each state is its own election system.
Thus there is no national popular vote. Unlike, say, Canada, we have no mechanism for producing one.
So what is this 'national popular vote' that we keep hearing about in talk about election results? It is a violation of the unit rule for real-world addition. It's simply taking the votes in different election systems and adding them together. This is not completely senseless. One can think of the number as evidence for counterfactual votes -- that is, under some set of assumptions, we can translate the votes from state returns into votes-that-would-have-been-cast-if-we-did-have-a-national-popular-vote, and then add these together to say something about what the national popular vote might have been like. But one can see immediately that this is not a straightforward process. First, since state election laws differ, simply taking the numbers straight across creates uncertainty. In some states, for instance, felons can vote; in others, they cannot. Would our counterfactual national vote be a felon-voting system or a non-felon-voting system? Obviously, there is no answer to that question, and, what is more, whichever answer to that question we gave, it would mess with our starting numbers. If we assume that the counterfactual system is a felon-voting one, our numbers for the non-felon-voting system no longer provide exact evidence for what the votes would be in the counterfactual system, because we don't know how many felons would have voted if given the chance, nor do we know exactly how they would have voted. Treating the numbers as if they could be just translated across to our hypothetical national voting process is wrong. The difference in our starting systems has created an uncertainty in the final numbers.
Felons are just one case. The uncertainties add up considerably. And this is all before we get to normal error -- voting machine glitches or paper ballot misread -- or moral wrongdoing -- voter suppression and voter fraud. And all of this is important when we talk about the relations among the real state popular vote numbers, the real Electoral College vote numbers, and the fictional national popular vote numbers. More on this in the next part, when I can get to it.