With the National Popular Vote compact, the popular-vote count from each state would be added up to obtain the nationwide total for each candidate.
And in every close election by popular vote there would be a national recount/reexamination of every vote.
Recounts are far more likely in the current system of state-by-state winner-take-all methods.
The possibility of recounts should not even be a consideration in debating the merits of a national popular vote. No one has ever suggested that the possibility of a recount constitutes a valid reason why state governors or U.S. Senators, for example, should not be elected by a popular vote.
The question of recounts comes to mind in connection with presidential elections only because the current system so frequently creates artificial crises and unnecessary disputes.
We do and would vote state by state. Each state manages its own election and is prepared to conduct a recount.
The state-by-state winner-take-all system is not a firewall, but instead causes unnecessary fires.
Given that there is a recount only once in about 160 statewide elections, and given there is a presidential election once every four years, one would expect a recount about once in 640 years with the National Popular Vote. The actual probability of a close national election would be even less than that because recounts are less likely with larger pools of votes.
The average change in the margin of victory as a result of a statewide recount was a mere 296 votes in a 10-year study of 2,884 elections.
No recount would have been warranted in any of the nation’s 56 previous presidential elections if the outcome had been based on the nationwide count.
The common nationwide date for meeting of the Electoral College has been set by federal law as the first Monday after the second Wednesday in December. With both the current system and the National Popular Vote, all counting, recounting, and judicial proceedings must be conducted so as to reach a “final determination” prior to the meeting of the Electoral College.