Trump is putting the House in play. Gerrymandering may save it.

Make no mistake, Trump was already collapsing in support and heading towards defeat before the release of his sexual assault bragging on tape. The weekend of Republicans dropping their support of him led up to a flawed, unapologetic performance at the second debate. The infighting is creating a “no-win situation” for Republicans according to Real Clear Politics’ Sean Trende as House Republicans have to decide if they support Trump and risk losing the support of swing voters, or dump Trump and alienate his hard-core backers.

Before this meltdown most prognosticators were estimating that the GOP would lose about 15 House seats, around half of what the Democrats need to take back the chamber. Geoffrey Skelley, of the University of Virginia’s Center for Politics, offers up the possibility that Clinton victory of 6 points may be enough to put the House in play. Should Clinton win by 6 points and improve uniformly on Obama’s 2012 margin, she would win around 50 currently GOP-controlled House seats. Based on the decline of split ticket voting, this could potentially tip the chamber to the Democrats.

But this analysis assumes a simple, uniform improvement over Obama’s 2012 margin and also ignores the host of seats where Democrats are noncompetitive. David Wasserman of the Cook Political Report looked at the numbers in early September and threw some cold water on Democratic hopes. His argument is that Democrats simply haven’t put enough seats into play to take back the House.

Let’s break down the various scenarios.

Skelley uses a simple metric, Obama’s 2012 performance, and then how many additional districts Clinton would win based on a uniform improvement. Wasserman instead looks at the actual performance by House Democratic candidates in 2014. Given that this is a Presidential election with far higher turnout the former makes more sense than the latter, but the latter is revealing for looking at changes in the seats that are competitive this year versus two years ago.

Let’s start with seats that Democrats lost in 2014 that are not in play this year. Republicans defeated incumbents John Barrow in Georgia and Nick Rahall in West Virginia, both entrenched incumbents who had been holding on while their districts shifted out from under them. They also won the open seat in North Carolina left by Mike McIntyre. Similarly, there are a number of districts where Democrats were competitive in 2014 and make Wasserman’s list of seats winnable under a 6 to 8 point swing in favor of Democrats, but that aren’t being contested this year. They include two districts in Arkansas and one in West Virginia. In other words, six districts that have historically elected Democrats, but are continuing to shift to the right as white working class voters, particularly in the South and Appalachia, embrace the GOP.

Next, comparing the districts that make Wasserman’s list from 2014, Skelley’s list of Obama surge districts, and ratings from Cook and the Center for Politics, we can identify three districts that Democrats contested in 2014 but are not in play in 2016. McArthur in New Jersey, Bishop in Michigan, and Donovan in New York. There are slightly different reasons in each, Donovan is a popular figure in contrast to the the scandal-plagued Grimm who held the seat in 2014. Democrats recruited Little House on the Prairie actor Melissa Gilbert in Michigan, but she dropped out in May. A small list, but every seat is crucial for Democrats.

Having addressed the changes from 2014 to 2016, we can also look at the consistency. There are nineteen seats that make Wasserman’s list of competitive results from 2014 that are also in play according to prognosticators. Here they are:

Alaska (Young)

Arizona 2 (McSally)

California 10 (Denham)

Colorado 6 (Coffman)

Florida 26 (Curbelo)

Illinois 10 (Dold)

Illinois 12 (Bost)

Iowa 1 (Blum)

Iowa 3 (Young)

Maine 2 (Poliquin)

Michigan 1 (Open)

Michigan 7 (Walberg)

Nevada 4 (Hardy)

New Hampshire 1 (Guinta)

New Jersey 5 (Garrett)

New York 1 (Zeldin)

Pennsylvania 6 (Costello)

Texas 23 (Hurd)

Utah 4 (Love)

I’ve bolded the three seats that don’t make the list of Obama surge districts from Skelley’s list. In other words, the vast majority, sixteen out of nineteen, are districts either won by Obama or winnable if Clinton performs two to three points ahead of Obama’s 2012 performance. The three exceptions are special, like snowflakes, in their own way. Don Young is a bizarre, buffoonish member who narrowly won reelection in the Republican wave of 2014. And internal Republican polling shows Clinton down only three points to Trump in the state, a vast improvement from 2012. Michigan’s Upper Peninsula is a historically Democratic, white working class region that has been trending Republican but still has shown strong Democratic performance locally And Mia Love in Utah has struggled to connect with her district, appearing more like a national show horse than a local work horse, and is not being helped by Trump’s alienation of Mormon voters.

But the fact is, these are not enough seats for Democrats to take back the majority. Luckily, there are a few key differences from 2014. First, they are picking up at least two seats due to court ordered redistricting in Florida and Virginia. Republican Jolly in Florida 13 is similarly vulnerable from both the combination of redistricting and doing everything possible to alienate the NRCC. And there are a number of seats that are considered in play right now that did not make Wasserman’s list of close races from 2014.

California 21 (Valadao)

California 25 (Knight)

California 49 (Issa)

Colorado 3 (Tipton)

Florida 27 (Ros-Lehtinen)

Florida 7 (Mica)

Indiana 2 (Walorski)

Indiana 9 (Open)

Kansas 3 (Yoder)

Minnesota 2 (Open)

Minnesota 3 (Paulsen)

Montana (Zinke)

New York 19 (Open)

New York 21 (Stefanik)

New York 22 (Open)

New York 23 (Reed)

New York 24 (Katko)

Pennsylvania 16 (Open)

Pennsylvania 8 (Open)

Virginia 10 (Comstock)

Virginia 5 (Open)

Wisconsin 8 (Open)

Again, the bold seats are outside of the Obama surge universe. Seventeen out of twenty-two will likely be won by Clinton given her polling ahead of Obama from 2012. The question for incumbent Republicans is how they can separate themselves from Trump without alienating base voters. Many, like Florida’s Ros-Lehtinen, have strong independent brands that should keep them in office. Others, like the Republican candidates in the open seats, are practical nobodies who could easily be crushed by a Democratic wave. A handful of these seats were contested by Democrats in 2014, but were not close as the election broke for Republicans.

Forty-one seats are in play for Democrats, while they are only behind in one, the open seat in Florida left by Graham’s retirement, because of redistricting. Three other incumbent Democrats (Ashford, Bera, and Nolan) are in close reelections but are benefiting from the anti-Trump wave. This could give them control of the House, but would be dependent on a Democratic wave sweeping away incumbents who are only marginally vulnerable now, based on polling primarily before the infighting has impacted public perception. The best comparison is 2006, when Republicans started October believing they were going to lose seats, but confident that they could minimize their defeats and keep control of Congress. Then scandal after scandal broke out, and the election produced a wave of just over thirty Democratic pickups. That’s exactly the bloodbath scenario Democrats need this year.

The Democrats could be doing better. As noted by Skelley, over fifty districts could be won by Clinton if she exceeds Obama’s 2012 performance. But only thirty-three of them are currently in play. In other words, Democrats have failed to make competitive a host of seats, either due to popular incumbents (like Upton in Michigan, where a Sanders backed college professor has failed to catch fire) or recruiting failures (like the open seat in Virginia Beach, which will almost certainly go to Clinton by big margins). This list of roughly twenty-one seats includes the three aforementioned districts that were competitive in 2014 but not 2016, meaning that the majority have seen consecutive Democratic failure.

Here’s a list of winnable districts that Democrats tried to contest in 2014, but the results were underwhelming and failed to make Wasserman’s list of winnable districts.

Illinois 13 (Davis)

Michigan 6 (Upton)

Michigan 11 (Trott)

Not too extensive. Democrats failed in both 2014 and 2016 to even try in many of these districts. I can even go back to 2012 to list out the seats that were contested.

Illinois 13 (Davis)

Michigan 3 (Amash)

Michigan 11 (Trott)

Wisconsin 7 (Duffy)

Virginia 2 (Open)

Democrats need to figure out how to go after a number of incumbents in Michigan, New Jersey, Ohio, Pennsylvania, Washington, and Wisconsin that they are simply not even trying to defeat. Some of these states, particularly Ohio, have been performing more Republican this election cycle than other swing states, and Senator Portman has successfully put away a race that many thought would be competitive. It may be that after the dust settles on election day we’ll have a new list of competitive districts, as the great sorting of white voters by education continues.

Continued class-based polarization of the white vote puts Democrats at a disadvantage in many competitive seats, but it’s gerrymandering that puts Democrats in the position of trying to contest in these areas in the first place. Republican gerrymandering translated a Democratic popular vote win in the House in 2012 into a Republican stranglehold of the chamber. But a glimmer of hope comes from recent court cases involving the Voting Rights Act in Virginia, where a GOP gerrymander was unpacked and Democrats will be picking up one vote. A similar approach in other Southern states would likely produce pickup opportunities in Alabama (+1), Georgia (+1), Louisiana (+1), South Carolina (+1), Texas (+3), North Carolina (+2), and the possibility of a swing district in Mississippi that could be going Democratic this year. At nine seats, this would put Democrats significantly closer to taking back the House.

A stronger Voting Rights Act is at the top of the list for Democratic priorities should they take back the House this fall, and even without the House an aggressive strategy could be pursued through the courts. A bigger goal could be the courts striking down all political gerrymandering, not just racial gerrymandering, which would significantly help Democrats in states like Wisconsin, Michigan, Ohio, and Pennsylvania. Such a scenario would have won the House in 2012 for the Democrats, and would put them probably ten to fifteen seats short of a majority going into this year. With the Trump meltdown they’d probably be able to pick up twenty to thirty seats, pushing them well into the majority. Luckily for Speaker Ryan, gerrymandering will likely save the GOP majority this year … giving him an opportunity to figure out how to patch things up with his conservative colleagues and maintain rule in the House. Good luck.