Anthony Downs taught us that the candidate closest to the median voter, with respect to their positions, always wins (An Economic Theory of Democracy, 1957). But contrary to popular belief, the median voter model doesn't require candidates to move to the center. Candidates can also move the median voter towards themselves.
Around 40 percent of eligible Americans don't vote. Aside from being embarrassing, it means that there is a lot of opportunity to increase turnout. But only if a candidate seeks to connect with and turn out non-voters. We are talking about young and minority voters. And only one candidate is speaking to them.
And they have been turning out in record numbers. Rock the Vote reported youth turnout was up 109% in the primaries. African-American turnout was way up, as well. If these patterns repeat themselves on November 4, we will have a whole new median voter, one that is to the left of the 2004 median voter.
Reaching out to new voters in the long tail of the electorate will increase turnout and pull the median towards the candidate, eliminating the need to pander to the center. The location of the median voter with a 55% turnout is not the same as the location of the median if a candidate can pull another 10% of eligible voters from the left (or right) out to the polls.
My sense is that the left will turnout more new voters than the right, partly because young and African-American voters are more likely to vote Democrat, partly because Karl Rove and company, already pushed turnout from the right to its near limits in 2004, and partly because McCain still has issues with turning out the far right (at least among those who have gotten over the bright and shiny VP-nominee).