Countless commentators have declared the American Electoral College to be an obsolete anachronism, and have suggested selecting American Presidents by simple majority vote. I argue that the Electoral College has a major advantage that has not been understood.
My interest in this area was stimulated by a popular magazine article by Will Hively (1996) about Alan Natapoff, and by a more technical paper which Natapoff kindly sent me (1996). However, I won't dwell on these papers except to say that I disagree with much of their argument, and I believe that Natapoff's concept of "voting power" is seriously flawed. Natapoff attempts to maximize individual "voting power" while still giving individuals equal power. As was noted in the letters to the editor of Discover magazine (March 1997), that goal can actually be attained by randomly selecting one voter, who is then allowed to name the President all alone -- obviously an unsatisfactory procedure. I found Natapoff's response to this argument unconvincing. Also, in those papers Natapoff explicitly avoids the issue of electoral fraud, which to me is the major single argument in favor of an Electoral College system for selecting the American president.
It's important to distinguish between two types of fraud. I'll call it minority fraud when the minority within a state manages to fraudently gather enough additional votes to make itself look like a majority. I'll call it majority fraud when a majority makes its majority look bigger than it really is, as when they stretch their candidate's vote from 60% to 70%. Majority fraud is probably far easier than minority fraud: for minority fraud to be successful the minority must fool the majority party within its own state, which probably controls the election machinery. Majority fraud requires merely that the party which probably controls the election machinery must manage to fool or silence a group that is already a minority. Therefore majority fraud seems like a far greater danger.
It is therefore worth noting that majority and minority fraud have completely different effects in electoral college and majority-rules systems. In an electoral college system, majority fraud conveys no advantage at all; only minority fraud is even worth trying. But in a majority-rules system, majority fraud is just as effective as minority fraud -- boosting a candidate's vote within a state from 60% to 70% helps tip a national election just as much as boosting it from 30% to 40% or from 45% to 55%. Thus under an electoral college system the easiest type of fraud is totally useless, while under a majority-rules system it is just as effective as other types of fraud.
This problem compounds itself. People living under a majority-rules system would quickly recognize all these points. Consider a set of states that all have quite different social and political climates, such as Maine, New York, South Carolina, Florida, Louisiana, Illinois, California, and Montana. People in each of these states think of people in all those other states as quite different from themselves. Thus people in California will naturally wonder whether elections in New York are being conducted fairly, or whether a reported 70% margin for the Eastern presidential candidate was produced partly by fraud. Therefore there would quickly be a demand to nationalize the entire electoral system, and that would probably be done.
But that brings in a whole new possibility of fraud -- fraud by federal officials. When we count the number of top federal officials who have gone to jail in the last 25 years, and reflect that winning the presidency is the most tempting prize of all, we can have little faith that nationalized electoral machinery would be run fairly. One of the remarkable features of our present system for choosing a president is that the federal government has virtually no control over the system. If a sitting president chose to "postpone" the next election because of a war or social turmoil, he might succeed under a nationalized system. But under the present system his opinion would have little effect, since all the electoral machinery is in the hands of the states. This scenario is not as outrageous as some may think; during Richard Nixon's first term as president (1969-1973), some liberals seriously suspected that Nixon was thinking about using the turmoil of those times as an excuse to "postpone" the 1972 election. (I don't need a historical citation for this point; I was there and I remember it clearly. I recall it was discussed in a small liberal political newsletter called I. F. Stone's Weekly.) As it turned out, the Republicans were able to pull a quite different type of dirty trick, influencing the Democratic primaries to get the weakest major Democratic contender (George McGovern) nominated, with the predictable result that Nixon swept the election, less than two years before his dishonesty was exposed and he was forced to resign.
In tense times like the time around the 1972 election, serious suspicion of major vote fraud in a federally-controlled presidential election could well produce major social unrest and perhaps even civil war. Thus the most important single point about our system for choosing presidents is that it must remain primarily in the hands of the states. For that reason, there is no serious alternative to the Electoral College.
Hively, Will. Math against tyranny. Discover, November 1996.
Natapoff, Alan. A mathematical one-man one-vote rationale for Madisonian presidential voting based on maximum individual voting power. Public Choice, 259-273, 88, 1996.