[dragonwriter]

> The number of overall votes it would take to change the outcome of the election, compared to the total size of the electorate, is small.

Sure, with a geographically optimum shift of votes to maximally leverage the anti-democratic nature of the electoral college, it would take about ~125k votes (37 EVs need flipped, ~40k gets PA for 20, ~31.5k gets GA for 16, and ~3.5k gets ME-1 for 1; at least I think that's the lowest-vote-change scenario, and assumes reversals of votes where every change reduces the margin by 2.)

Of course, that would also be by far the biggest popular vote loss by an electoral college winner since the election of 1824, which had four candidates clearing over 10% of the vote, and the only electoral college defeat of the majority (not merely plurality) winner of the popular vote in US history.