The next census is still 5 years away, but population changes since the 2010 census already allow us to guess at apportionment changes the next redistricting will bring. Some changes are already assured (e. g. Texas will definitely gain seats, probably 3), whilst others are possible but not assured (e. g. New York is currently on course to drop to 26 districts, but it break a streak that goes back to the 1940 census and manage to hold steady instead.)
One change that falls into the latter category is Virginia gaining a 12th district, which would make its delegation as large as it's been since before West Virginia's secession. So what would a twelve-district Virginia look like?
The first thing we have to take into account is that recent population growth in Virginia has been distributed unevenly. Whilst outlying areas of the DC metro area are growing exponentially, in much of rural Virginia growth is stagnant. In fact a quarter of counties and cities actually saw falls in population between 2010 and 2014. Just dividing the state up into twelve based on the 2010 census therefore isn't going to produce an accurate picture. Instead I've taken growth between 2010 and 2014 and done a linear extrapolation to 2020. This is a fairly blunt instrument and could easily be out by 10,000 here or there, especially if something like a recession affects migration patterns.
The second thing to consider is who draws the lines. In Virginia, this is the General Assembly, comprised of the State Senate and the House of Delegates. The House of Delegates is an extreme Republican gerrymander and there's no realistic prospect of it being broken prior to 2020. The State Senate is more finely balanced, but at present the Republicans have a narrow majority and few of their seats are remotely competitive, making a Democratic regain of the chamber challenging.
Thus the balance of probabilities suggests a Republican-drawn map, which in Virginia means an aggressive gerrymander. However, just because they draw the map, doesn't mean they can pass it. The Governor can veto redistricting bills and Republicans won't have the votes to override it in the Senate. Given Virginia's trend towards the Democrats, they're probably favoured to hold the governorship in 2019, although as it's an off-year that's far from assured.
If the Governor or the State Senate blocks a Republican gerrymander, the Republicans could either pass a compromise map or let the courts step in and draw the map. That's what I've drawn here, because predicting a Republican gerrymander is impossible without extremely accurate population estimates or in-depth knowledge of poitical changes at a sub-county level (neither of which I have).
The third issue is the VRA. The situation is in flux since VA-3 has been struck down, but if the decision isn't overturned then there's no need for a VRA district to be over 50% black VAP and much more need for it to be compact. This, and the fact there's one more district to go round, means that there really ought to be a second district in Virginia where the black population have the ability to elect the candidate of their choice. It can't be guaranteed this will happen, because a Republican DoJ won't care and Democratic DoJs have shown a regrettable tendency not to play hardball. However, I've taken an optimistic view here and assumed a second district will be drawn.
Not that the deviations are on current population, not my 2020 estimates (though I allowed a fair amount of leeway there, given the rough nature of my estimates).
VA-1: The first pulls out of Fauquier and Prince William, instead grabbing Spotsylvania and Orange Counties from VA-7. As this removes areas rapidly trending towards the Dems, Wittman would find much to like here. Safe R
VA-2: In 2020 Virginia Beach and Chesapeake will likely have almost exactly the right population for a congressional district. I didn't notice this until I'd almost finished drawing the Hampton Roads VRA district (VA-3), but it's so neat that I can easily imagine a court drawing it. However, it combines Scott Rigell and Randy Forbes in a district that isn't 100% safe, so I'm fairly sure this would have state Republicans up in arms if it were drawn.
Likely R
VA-3: If VRA districts don't need to be 50%+ African-American, then it's much easier to make them compact in Virginia. This combines the core of the Hampton Roads metropolitan area with the Eastern Shore. The Democratic primary will be comfortably black-majority, but if the white percentage is felt to be too high, it'd be possible to rectify this by giving the Eastern Shores to VA-2 and picking up black areas of Chesapeake. That said, this is almost black-plurality by total population, so by 2020 I suspect it'll probably be fine. Safe D
VA-4: As in just about every other fantasy map I've ever seen, Virginia's 2nd VRA district combines Richmond with the Southside. However, with 12 districts and more flexible rules, you can draw it surprisingly cleanly. Henrico and a sliver of Chesterfield (to take in I-295) are split, but every other county and city is kept whole. Black-plurality by total population.
Safe D, Dem gain
VA-5: The shape of the district is neatened, but it remains a relatively cohesive Piedmont district, where white rural areas comfortably outvote Charlottesville and the black vote. In fact, it's a couple of points redder than the current district, so Hurt would face no problem whatsoever. Safe R
VA-6: As I'm not gerrymandering too aggressively, a compact Shenandoah Valley district certainly ought to be attempted. The north of the valley is kept out to help Comstock in VA-10 and in the south it extends to take in just about all of Roanoke's hinterland, but otherwise this is a pretty clean CoI district. Safe R
VA-7: By losing its arm to the north, this becomes a suburban Richmond district. Overwhelmingly white and overwhelmingly Republican, a few tweaks would probably make this the best McCain district on the map (though by the 2020s VA-9 will assuredly be redder.) Safe R
VA-9: Skipping northern Virginia for now, we come to the south-western district. This area is losing population, so even though the state is split between one extra district, it still needs to pick up electors. It does this by adding Danville to the east, which can be done without rendering the district even remotely competitive. Safe R
VA-10: Comstock's home in McLean is drawn out, but I guess that with better estimates at a precinct level it'd be easy enough to stick that back in in exchange for more Democratic areas of Loudoun. This is a couple of points redder than the current VA-10, so given trends she might prefer to move anyway. Likely R, though with better precinct data Safe R might be achievable.
VA-8: The lines here shouldn't be taken too seriously, as I wasn't able to establish groups of whole counties smaller than Fairfax + Arlington + Prince William and therefore used a lot of guesswork as to where growth is actually occurring. In addition, there's a case that I should have tried to draw a minority-majority district here. I decided not to because I see no reason to believe that it would perform, but that might be different by 2020 and a Republican gerrymander would certainly attempt to (not least because it'd make not drawing one in the Southside easier.) As it is, this is already bluer than both VRA districts I've drawn.
Safe D
VA-11: Again, lines were mostly about getting something that looks clean, rather than a concrete suggestion. Maintains Connolly's home in Mantua. Safe D, though a Republican gerrymander would pump up the margins here in order to bolster VA-10 and perhaps take a shot at VA-12.
VA-12: On current trajectories, Prince William County will be solidly blue by 2020 and will have sufficient population for two-thirds of a congressional district. This being the case, I think it deserves to anchor a district and even a Republican gerrymander would probably abandon most precincts here to the Dems. That said, this isn't that strong a district (because it also includes some of the reddest bits of Fairfax), so it could be vulnerable in wave years, particularly with some slightly cleverer line-drawing. Likely D
So my map postulates a gain of two seats for Democrats and one loss for Republicans. However, the gain/loss comes about from enforcement of the VRA, not from reapportionment and hence can't be relied upon. Moreover, a strong Republican gerrymander that sliced and diced Prince William could probably keep the new seat in their column initially, though it might risk becoming a dummymander by the end of the 2020s.
Thoughts, criticism?