Trump would have won if the United States looked like this
Four years ago, I built a tool called Redraw the States to ask a simple question: How few counties would you need to move from one state to a neighboring state to change the outcome of the Presidential election. The answer turned out to be three. But in exploring these quirks of American political geography, I found lots of other little oddities as well.
For instance, can you tell what is wrong with the above map? If it helps, here’s a link to a map of the United States. Still can’t spot the difference? A few hints: there are four changes. One to the President-Elect’s home state, one to the state that’s currently hosting two elections for Senate, one in the Upper Midwest, and one in the Pacific Northwest.
If you still can’t spot the difference, I wouldn’t worry about your grade in US geography. The differences are extremely tiny spatially but make a huge difference in the outcome of the election. Specifically, I’ve moved Delaware County, PA, into Delaware, Savannah, GA, into South Carolina, a couple counties from Washington to Idaho, and finally, I’ve moved the southernmost counties of Minnesota into Iowa. The result: a 269–269 tie in the Electoral College, which would throw the election into the House, which Trump would then win.
The rules that govern the US Presidential election are extremely arcane. Each state gets a number of Electoral College Votes equal to the number of representatives in they have in the House of Representatives plus two. DC, not being a state and thus having no representatives, gets 3 Electoral College votes. To allocate these votes, each state and the District holds an election on the first Tuesday after the first Monday in November. In 48 of the 50 states plus DC, the person with the most votes (but not necessarily a majority of those votes) in the state gets all of the state’s Electoral College Votes. But in Maine and Nebraska, the person with the most votes gets 2 of their Electoral College Votes. The rest of the votes are given out to the person with the most votes in each of the state’s Congressional districts. What happens if no one gets a majority of Electoral College Votes is a whole other Rube Goldberg mechanism that we won’t go into.
There are many, many, many reasons, to dislike the Electoral College, but to me this instability of the results due to small changes in the map is the most egregious. State lines were drawn for all sorts of reasons ranging from geographic pragmatism (a river runs along the border) to 19th Century compromises involving slavery (the Missouri Compromise chief among them) and bad legal drafting (see the Toledo War). And while gerrymandering is bad, there’s more method to that madness than Alabama failing for 100 years to get the Florida panhandle to secede.
Which brings us back to the original question: How many counties would need to move to a neighboring state to swing the election to Trump. The answer it turns out is the same as in 2016: three.
But which three counties and how many ways can it be done? Well, we’ve already seen above that moving Delaware County, PA, to Delaware would flip Pennsylvania. But similarly moving Philadelphia itself or (if you allow a little fudging) Montgomery County to New Jersey, you would also flip Pennsylvania.
In Georgia, we’ve seen that moving Chatham County to South Carolina would flip the state red. But so would moving Richmond County, home of Augusta, Augusta State University, and the Medical College of the University of Georgia.
If we make these changes, Biden’s lead slips to 271–268 in the Electoral College. So we can either try to flip another state, or we can muck with the number of representatives each state gets in the House. The formula for House apportionment also has a convoluted history and is sufficiently arcane that we used it above to get a 269–269 tie. There we netted Iowa and North Carolina one more representative each at the expense of Minnesota and Washington.
(No, that’s not a typo. While I moved the counties from Washington to Idaho, the apportionment formula is such that North Carolina picks up the extra House seat.)
But if our goal is to move as few counties as possible, we should look for states that were close. For that, look no further than Arizona, which turned out to be nail biter. Here, Biden won 5 of 15 counties. If any one of four of these counties (Apache, Coconino, Maricopa, and Pima) were moved to another state, Trump would have won.
There are many other ways to redraw the map that reveal all sorts of oddities. I encourage you to play around yourself with the updated 2020 Edition of Redraw the States and Tweet me @khayeswilson on Twitter with your best maps. Last time there were some great ones!
But all of this leaves us with the question of what we should do to fix the problem that is the Electoral College. Short of moving to a better constitutional system like the one New Zealand uses, I think the solution is simple: just let the popular vote winner be president. It’s simple, equitable, and obviously democratic. There’s even one weird trick called the National Popular Vote Interstate Compact that would get us there if only a couple more states would sign on.
In the end, no government can long survive without the consent of the governed, and one of the biggest drivers of consent is feeling that the system is reasonably fair. We know from the gerrymandering wars that people don’t think it’s particularly fair when legislators choose their voters instead of the other way around. And while U.S. state boundaries are harder to redraw, that changing so few of them in such small ways can change the outcome of the presidential election no less absurd. Let’s move beyond it before we can no longer call ourselves a democracy.
Correction (2021–01–07): Thanks to several people in the comments who pointed out that Augusta is not the home of UGA but ASU. Apologies for the error.