In an electoral college system, however, the campaigns are induced to focus less on the sheer size of a state and more on its political moderation. The so-called swing or battleground states are those states with roughly equal numbers of voters potentially willing to back different candidates, such that an extra successful effort by one candidate could tip the balance. This is not a theoretical concept, but rather an observable phenomenon. For example, this year the candidates repeatedly visited swing states such as Florida, Pennsylvania, North Carolina, Ohio and New Hampshire. Tiny New Hampshire, with its mere four electoral votes, received more candidate attention than California with its 55. Why? It is because New Hampshire was open to being persuaded by either candidate, whereas California was not.
As I noted to Chuck in some Facebook comments, I don't think this is true. A state can be a "swing state" because it contains a lot of voters who are political moderates and thus swing voters. But a state can also be a "swing state" because it contains a roughly equal number of Democratic voters and Republican voters, who are no more persuadable than their counterparts in other states that have less equal numbers. In the latter case, the "political moderation" benefits of the electoral college don't materialize. From my experience, the Democrats in New Hampshire (where I work) are just as partisan and no more open to persuasion to vote for a Republican than the Democrats in Vermont (where I live). Chuck is correct that candidates respond to incentives and deluge the New Hampshire media market with advertising, but these advertisements may simply be efforts to turn out their base of partisans in larger numbers than their opponent's base of partisans.
To add some further support to my hypothesis, I would note that the political advertising done in New Hampshire during the general election was not particularly targeted to swing voters. All the negativity seemed like an effort to drive turnout by the candidate's base or to depress turnout by the opponent's base. I cannot recall an advertisement that tried to appeal to moderates who might be undecided (e.g. here's a problem we need to solve in a way that respects the valid points on both sides of the political spectrum.)
Much of Chuck's article is about what would happen in the counterfactual case where the national popular vote determined the winner. He is right that candidates would likely gravitate to their strongholds to try to drive turnout by their base. (President Bush remarked, when queried about his failure to capture the popular vote majority in 2000, that if that was the objective, he would have campaigned in Texas and "run up the score.") My simple model of what candidates are doing, now and in the counterfactual, is to devote their resources where the "price per vote" is the lowest. The price they pay is with their own time and with advertising. Population density matters for economizing on their time, but I presume that media buys are more expensive when they reach more people, so that it is not clear that media dollars would flow to the largest markets. Further, the media buys presumably depend on total population, not "total population likely to be sympathetic to your message." In this case, more politically diverse areas -- swing states -- would be further disadvantaged.
Addendum: How sad, I have finally repeated a post title without realizing it. Fortunately, the message is consistent.