Is there a case for a system that sometimes produces undemocratic outcomes? I think so, on two grounds. First, it creates incentives for political parties and candidates to seek supermajorities rather than just playing for 50.1 percent, because the latter play is a losing one more often than in a popular-vote presidential system.
Second, it creates incentives for political parties to try to break regional blocs controlled by the opposition, rather than just maximizing turnout in their own areas, because you win the presidency consistently only as a party of multiple regions and you can crack a rival party’s narrow majority by flipping a few states.
According to this — admittedly contrarian — theory, the fact that the Electoral College produces chaotic or undemocratic outcomes in moments of ideological or regional polarization is actually a helpful thing, insofar as it drives politicians and political hacks (by nature not the most creative types) to think bigger than regional blocs and 51 percent majorities.
That is from the NYT, he also considers some arguments against.