Today is the last day for voter registration in nine Southern states -- AR, FL, GA, KY, LA, MS, TN, TX and VA.

If any of you may be wondering if last-minute registrations really make a difference, Nate Silver at leads with some analysis today that shows how small registration and turnout gains could tip the election.

And to make the point, he uses an unlikely example: Georgia.

No polls show Georgia to be very close for the presidential race. But Silver shows how even a modest gain in African-American turnout could make Georgia -- and other states with growing numbers of black voters, like North Carolina -- a lot closer than you'd think.

Assuming that about 95% of African-Americans will vote for Obama, and about 70% of non-black voters for McCain, Silver comes up with this very compelling analysis:

In 2004 ... black voters made up 25.4 percent of election day turnout (this means that they participated at slightly lower rates than white voters). Applying those 95/5 and 30/70 voter splits to the 25.4 percent figure would work out to a 7.0-point win for John McCain, about where polls seem to have Georgia now.

Now suppose that black and nonblack voters each turn out at the same rates as they did in 2004, but that we account for the increase in black registration. According to our math, John McCain's 7.0-point lead is now cut to 4.9 points.

But that is probably too conservative an assumption. Newly-registered voters -- and nearly half of Georgia's newly-registered voters are black -- turn out at higher rates than previously registered voters. In addition, one would assume that the opportunity to vote for the first African-American nominee might be just a little bit of a motivating factor for black voters. Suppose that African-Americans represent 29.0 percent of Georgia's turnout, matching their share of active registrations. Using the splits we described above, McCain's lead is now cut to 2.3 points.

Even this, however, may be too conservative. For one thing, the registration window in Georgia is not yet over ... it concludes today. The statistics I cited above only reflected registrations through September 30. There is typically a surge of registrations in the final few days before the deadline. In 2004, Georgia's active voter rolls increased by about 150,000 persons in the first four days of October, before the registration deadline closed.

So suppose that by tonight, black voters have increased to 30 percent of Georgia's registered voter pool. Plugging that 30 percent number in, McCain's advantage is a mere 1 point.

Silver doesn't conclude that Georgia will go blue -- but the exercise shows how this election is changing many of the old assumptions about Southern politics. Indeed, as Silver notes, given that African-American voters have made up 40% of Georgia's early voters in 2008, Obama right now holds the lead in Georgia.

What does this mean for other Southern states like North Carolina? I'll be doing a more in-depth piece on this later, but here's a quick take.

According to the latest NC State Board of Elections statistics, 422,200 new voters have been added to the NC rolls in 2008.

33.4% of that new growth has come from African-American registrations. The share of the NC electorate that is African-American has grown almost 1%, to 21% of NC voters.

Starting with the assumption that 95% of African-American voters will go for Obama and two-thirds of non-black voters go for McCain -- a generous number -- Obama's starting baseline is 46% in NC.

But then you start adding in all the other factors that are working in Obama's favor in NC:

* As Silver noted in Georgia, we can assume a higher rate of turnout among newly-registered -- and newly-energized -- voters. This will be especially true in North Carolina, which will have two weeks of same-day registration and voting this month -- ensuring that new registrants will be voting. If current trends hold, those new registrants will disproportionately be voting for Obama.

* Leading this trend is the growing share of African-American voters. If the African-American share of the electorate rises 1% by Election Day, Obama's overall baseline rises to 47% in NC.

* Obama is also doing better among certain non-black voters, especially younger whites -- think of all the colleges -- Latino, Asian-American and Native American voters. A recent Public Policy Polling survey found Obama doing 10 percentage points better than Kerry in the NC mountains, Obama's toughest territory. If Obama increases his support among non-black voters by just 1.5% to 36.5%, his baseline rises to 49.8% in NC.

And so on. These are just estimates, but you get the picture.

There's a reason why North Carolina -- no matter what the New York Times might say -- is a battleground state. Of the seven latest NC polls, four show Obama with the edge, one gives the edge to McCain, and two have a dead-even tie.

It all has to do with the changing South and the new 2008 election map.