Common Cause tweeted me yesterday about my post about the ability of the Electoral College to limit fraud. Common Cause suggested that there would be a much greater incentive for fraud under the Electoral College because one state (or a few swing states, which are usually known in advance) could matter so much. So if you know that, say, Florida is a swing state, you might be much more likely to commit fraud for your side there.
This is not an unworthy point. However, as I suggested in my original post and as Tara Ross tweeted back, the Electoral College localizes the incentive for fraud. Under the Electoral College, the benefits of fraud are really only there if you're in a swing state. If the state is clearly trending toward one side, you're just adding meaningless numbers to your candidate. This localization allows for increased public scrutiny. The public (including the media, watchdog groups, national organizations, federal lawyers, etc.) would know where to keep its attention fixed for voting irregularities. So not only would there be a more limited motivation for fraud; the public would be more likely to find out about this fraud due to the concentration of potential areas for fraud.
Under a National Popular Vote scenario, every precinct in the nation could be an opportune place for electoral fraud. And there have been a lot more elections where the popular vote was close than where the winner of the popular vote lost the electoral college. In the past 50 years, there have been 3 presidential elections which were decided by a fraction of a percent: 1960, 1968, and 2000. And 1976 and 2004 were also very close; they were decided by less than 2.2%. A few hundred thousand votes either way could definitely have altered the outcome. Under NPV, those votes could come from anywhere in the nation; under the Electoral College, those votes could only come from a few close states.
Because the Electoral College places buffers around fraud, it may help protect the public sense of legitimacy of presidents elected in close races. Because the opportunities for fraud are more limited, the public has fewer places to look for irregularities. Of course, the Electoral College is not a panacea in this regard: certainly, much of the vitriol of the 2000 election was unhappily facilitated by the Electoral College's structure. But consider an alternative National Popular Vote scenario of a close race, where partisans of both sides would have thousands of precincts across the nation about which to cast aspersions of skulduggery and fraud. The poison of 2000 would be magnified. Every single precinct would become a battleground. The ensuing firestorm (and, if history is any guide, there would be plenty of close elections) could very much damage the sense of the legitimacy of the office and the holder of this office.