Barely the favorite but still the favorite, according to FiveThirtyEight’s “polls-only” model. Thanks to a new survey out today, she now has a 51.3 percent chance of winning a state that Romney won by eight and which the GOP has won in seven of the last eight presidential elections. The good news for Trump fans is that the site’s “polls-plus” model, which incorporates economic data and historical trends, still has him a comfortable favorite in Georgia at 68.5 percent. But then, the “polls-plus” model also had Hillary as the national favorite even at the height of Trump’s post-convention bounce, when he surged into the lead in various surveys (and inched ahead in FiveThirtyEight’s own polls-only model briefly), and some Trumpers cried foul about that. It’s only the polls that should count, they said! If you believe that then you’re stuck, at least for the moment, believing that she’s ahead in Georgia.

The last three surveys of the state have had the race tied, then Clinton ahead by four, and now, via JMC Analytics, Clinton by … seven:

JMC is not a highly rated pollster, drawing a “C” rating from FiveThirtyEight, and there’s some reason to be skeptical of these numbers based on the gender gap. They have Hillary ahead by 13 points among women, which is in the ballpark of her national lead, but among men she trails Trump by just one point. Trump usually leads Clinton comfortably among male voters. Proof that the poll is faulty? Not necessarily: The reason she’s competitive in Georgia to begin with is that the state has a huge black minority, approximately 30 percent of the total population. Black voters of both sexes almost unanimously favor Clinton. The heavy number of black men supporting Clinton are bound to offset part of Trump’s lead among white men, meaning a narrower spread among male voters overall. In fact, look back at the graph above of preferences by race and you can see in a glance why Trump is having trouble in the state. Romney won 59 percent of the white vote in 2012; Trump is at a mere 52 percent, with Clinton and Johnson each siphoning off chunks of that vote. In fact, Trump may be underperforming Romney more dramatically than that: There was no exit poll of Georgia in 2008 or 2012 but it’s logical to assume that white voters in a reliably red state within the GOP’s southern base tend to vote Republican at higher rates than voters nationally do.

You would expect that some white voters in Georgia who are currently undecided or are supporting Johnson will shift later this fall and deliver the state to Trump, which explains why he’s still the favorite to win there (albeit not at this particular moment). But the more that purplish states like Virginia and Colorado turn safely blue, the more Team Clinton can shift money and organizers from those places to new battlegrounds like Georgia and Arizona to try to make them competitive. That’s already happening, in fact. Normally you’d expect undecideds in Georgia to come home to the GOP in October but it’s unclear what happens if Democrats are contesting the state aggressively and targeting black turnout. And there’s more reason to worry: Nate Cohn asks a good question in wondering what South Carolina looks like right now if Georgia looks like this. That state hasn’t been polled this year for the obvious reason that it’s one of the safest Republican jurisdictions in the country and usually isn’t worth wasting money on during the general campaign. But maybe it is this time. After all, South Carolina’s demographics look a lot like Georgia’s, with blacks composing 28 percent of the population. If Trump’s support among Republicans is tepid and Democrats go all out to organize black voters there, that race could be competitive too. And suddenly, instead of camping out in Pennsylvania and Ohio, Trump has to split his time between the Rust Belt and … the south.

One other hard-to-fathom wrinkle in all of this. Team Clinton may decide that two of its best weapons in South Carolina and Georgia are Barack and Michelle Obama. It’s hard to imagine a president as deeply disliked by conservatives as O campaigning in the heart of the GOP’s southern base but Obama’s job approval is at 52 percent or better in six of the last seven national polls. He’s not the liability that he has been in the past. And obviously, a direct appeal to black voters from the first black president would carry weight like little else. If things go sideways for Trump, and they’re already leaning that way, Democrats are going to try to flip some southern states. And there’s a nonzero chance that they’ll succeed.