Electoral College vs. Popular Vote mis-matches
National polling has a fine track record in predicting presidential races, despite the fact that the Electoral College presidential selection process uses only state-by-state results. This was a practical reality. It's isn't a natural law that national polling works, but rather an observation about American politics.
It would be easy to "redistrict" the states in such a way that that the popular vote never matched the electoral college result. So the century of close matching we had is to some degree happy accident for the system... the absence of pressures on the system that could have just as easily been manifest.
The ongoing polarization of the nation, particularly in the South, may well be creating a new reality where popular vote/electoral college mismatches may become the norm, and with Democrats as the winners, despite getting fewer votes.
In 2000 Gore got half a million more votes but the election was decided by 1,000 votes in Florida. (It is irrelevant here whether Gore should have won or lost. The point is that the electoral college race was a coin-flip, despite the popular vote.
In 2004 Bush got millions more votes than Kerry, but small changes in Ohio would have given Kerry the presidency.
In 2008 it was a blow out so there was no chance of a mismatch.
In 2012 we have Gallup giving Romney a strong national lead (correctly or incorrectly) despite Obama leading in three of the four national regions in their polling. But Romney's immense lead in the south makes for a substantial popular vote lead. Whether the Gallup national headline number is right or wrong is irrelevant to this particular question. What is noteworthy is the concentration of Republican voting in the south, and in the red western states. The electoral college system does not reward concentrated support. If you win Alabama by 40% you get the same electoral votes as if you won it by 1%.
State versus national polling has never been this far apart, since the start of modern polling. In 2012, the odds going in of an an electoral college/popular vote mismatch are unusually high (though still slim), and this is after mismatchy-type elections in 2000 and 2004.
So the historical connection between votes and EC electors appears to be, though still strong, the weakest in has been in any of our memories, and it could get worse before it gets better.
Changing demographics should tend to smooth out the south some, rather than polarizing it more, so we will not develop a permanent mismatch-nation. But there is nothing in the rules to prevent it.
What would happen if we did have a permanent mismatch? Civil War or nothing. The mismatch disadvantages the concentrated states on the losing side of the mismatch, and thus empowers less concentrated states. In general terms, it helps more states than it hurts. So there would be no way to amend the Constitution unless a lot of states chose to reduce their power. Unlikely.
Yesterday I posted that the South region in Gallup polling is 37% of the national population. That was incorrect, and I apologize for the error. Gallup does not use the South region as defined by the census, which is 37%. They put Maryland, DC and Delaware in the Northeast. So though Gallup's south is still the most populous of the four regions, it is less than 37%. Probably closer to 32%.
= new reply since forum marked as read
