>FWIW, there is nothing wrong with the electoral college per se since it does depend on the popular vote within a state. Seems to me if you get the underlying problem fixed, the electoral college issue goes away.

Not quite. Independent of this current mess, which most of us agree your "universal" solution would repair much of, there is an inherent flaw in the electoral apportionment system as it is, where 48 states and DC use the "winner take all" method. It simply is mathematically flawed.

Mathematically, as the percentage of difference in the popular vote between the two major candidates increases, the elctoral winner is magnified, and there's a nice mathematical curve, in fact. Fine, that IMO is a positive thing.

However, below a certain popular vote margin of difference (in the vicinity of 1%, is my estimate), the current electoral system winner begins to fluctuate wildly with respect to the popular vote winner, there is no longer a positive correlative relationship. As you go below a 1% margin, it becomes increasingly likely that the popular vote winner will lose electorally, due to the combination of these factors: large swings caused by "winner-take-all" states, the huge popular vote margins in some states, and the disproportionate clout small states have. This assumes a perfect vote tally.

Now some "fix" ideas, such as the direct within-state apportionment idea (as Peter suggested) simply will create chaos, I agree. And going to a national popular vote is not a likely scenario either, it requires an Amendment, almost certain to be blocked by smaller states. Let them keep their extra clout, fine.

Given that, the only logical solution (as has been suggested by myself and others) is the "mathematical" electoral solution: improve the current system by apportioning the electoral votes by CD within state, exactly as Maine & Nebraska do now (the presidential winner in each CD gets the electoral vote, and the statewide winner gets the extra two). Thus the major two candidates will still receive all the electorals (though Perot 92 might have picked up a few), but the electoral count will much more accurately reflect the popular vote since they're assigned at a less aggregate level. Small states will retain the same clout, so they're not a problem. It doesn't require an Amendment, so that's not a problem.

The only obstacle is getting the state legislatures to approve the idea in more states (the smallest ones don't need it, of course) - and that could be accomplished quickly, certainly, if the two major parties agreed on the idea.

I'd like to see a little further analysis on past elections, see how the numbers under this method stack up compared to the current system...but I will wager we will see a significantly better correlation between popular and electoral vote, especially in close elections.