This would be a truly wretched outcome. The electoral-vote system was a terrible idea when the Framers created it, and it has gone wrong four times. It failed in 1824, when John Quincy Adams won nearly 40,000 fewer popular votes, and 15 fewer electoral votes, than Andrew Jackson, but was nonetheless elected president by the House of Representatives. It failed in 1876, when chaos at polls in the South permitted a Supreme Court commission to grant the election to popular-vote loser Rutherford Hayes (known to many thereafter as "His Fraudulency"). It failed in 1888, when Benjamin Harrison defeated incumbent Grover Cleveland with the same kind of win Obama might achieve -- narrow Harrison majorities in big states against huge Cleveland winds in the one-party South. And, of course, it failed in 2000, when George W. Bush lost the popular vote, by 47.8 to 48.4 percent, but was installed by a vote of 5-4 in the Supreme Court. An electoral-popular split also seemed possible in 1988, and we had a close encounter with one in 2004, when John Kerry might have been the electoral winner had he managed to get 150,000 more votes in Ohio.
If Obama wins without a popular plurality, we can expect Fox News to begin demanding that he concede anyway, to which the sane answer is the one Bush gave in 2000: No. If the president were picked by popular vote, both candidates would have run very different campaigns; in technical terms, the winner of electoral votes in an electoral-vote system is called the winner. (For those who doubt my sincerity, consider this quote I gave a reporter in 2000, when I was a volunteer with the Gore campaign: "Saying Gore should win because he won the popular vote makes as much sense as saying that Bill Clinton should get a third term because polls show he's more popular than either of these guys.")
2. Narrow electoral victory for one side or the other: Every few cycles, one or two electors refuse to support their party's candidate. It's never swung an election, but it could. GOP vice-presidential nominee Bob Dole once told a congressional committee that, had Gerald Ford carried Ohio in 1976 to come within seven electoral votes of Jimmy Carter, "We were shopping -- not shopping, excuse me. Looking around for electors ... We needed to pick up three or four after Ohio."
Neither side is above this kind of shenanigans, but this year the temptation might be great and, with Crossroads GPS and the Koch brothers in the mix, the amount of money available to produce a switch might be prodigious.
3. Tie in the electoral vote: Numbers guru Nate Silver on Sunday gave this outcome only a 0.4 percent chance of occurring, but that's more risk than a stable country ought to be running. If there is a tie, and neither side can flip a treacherous elector, the choice of the president would fall to the House. In that process, under Article II of the Constitution, each state -- from mighty California to tiny Wyoming -- would have one vote, determined by a majority of that state's House delegation. An evenly divided House delegation casts no vote; the winner must receive the votes of 26 states. The choice will be made by the new Congress, which takes office in early January. The math of the current House suggests that the next one would likely produce 26 votes for Romney.
In the meantime, the Senate would pick the vice president. Current political math gives an edge to Joe Biden in that count. That's important because if the House doesn't produce a winner by January 20, the vice president-elect "shall act as President until a President shall have qualified."