The High-Risk Strategy That Could Hand Democrats the White House

Chris Wilson

Over many years of following our beloved Philadelphia Phillies, my dad and I have often wistfully imagined that, in games where the Phils disassemble their opponent by a lopsided margin, the team could store some of that surplus in a “run bank” to be allotted in future games where they lose by one or two runs. It would be a great system, so long as only the Phillies could use it.

I thought of that idea on the night of November 8, 2016, when Hillary Clinton won the popular tally by 2.86 million votes but lost the Electoral College, 306 to 232 (putting aside the 7 faithless electors who cast their support to someone else, which would presumably never stand if it swayed the outcome).

The 2016 election hinged on three states: Pennsylvania, Michigan and Wisconsin. They were worth 46 electoral votes, and went to President Trump by a combined 77,744 ballots. These three states are so often mentioned in the same sentence as one monolithic bloc of disappointment for Democrats that we really ought to start calling them “Pennsylconsigan.”

If both candidates had been allowed to deposit their surplus votes from runaway states into a “vote bank” to draw from in close states, the election would have gone to Clinton by a healthy number of electors. Her combined surplus margin of victory in the 21 states she won (including D.C.) was 11,226,150, compared to Trump’s 8,357,682 in the 30 states he won, giving her plenty of capital to both shore up the Rust Belt and defend narrow victories in states like Minnesota and New Hampshire.

Here’s the secret: There actually is a vote bank. It’s just a special kind of bank that never tells you your balance and has a strange, laborious way of reallocating funds. It’s called campaigning. And if the eventual 2020 Democratic nominee invests wisely, it very well could decide the outcome.

The biggest challenge facing Trump’s eventual opponent is not popular support. It’s a distribution problem. A large percentage of reliably liberal voters live in a small number of states like California and New York. As I noted last week, the 2.4% of eligible voters who move every year further complicates the problem, since some states export more desirable demographics for either party than they import. While both campaigns are expected to wage aggressive registration and turnout efforts, nothing short of mass state-to-state migration will change this fact. (Reporter Alec MacGillis told “young hipsters” to move west in October 2016, if more for the sake of Congress, noting that the “hyper-concentration of Democratic votes has long hurt the party in the House and state legislatures.”)

For the Democratic candidates, the best way around this challenge is to play the “vote bank” strategy, focusing all of his or her energy on ginning up support in the fewest places where electoral return on investment is the highest, at the cost of reallocating “funds” — time spent campaigning and advertising — from places that need them less.

On the face of it, this doesn’t sound much different from setting up shop in swing states, the way all candidates do. But this plan involves finer tuning. The smart campaign will make its best guess as to the handful of states where the smallest growth in support will generate the largest boon in the Electoral College, spend as little time and money as possible everywhere else, and hope for the best.

It’s a gamble, yes — akin to an investor making a small number of high-risk bets on a few companies rather than diversifying their portfolio through interest funds and the like. But the current electoral map is sufficiently baked against the Democrats that it could be their only path to victory.

For example: If Democrats were to cede “Pennsylconsigan” (which could easily flip back on its own) and put all their chips on Texas (which, depending on who you believe, could in fact go blue), that single victory would provide precisely the 38 electoral votes that Hillary Clinton needed to get to 270 and become President.

Sound reckless? Consider this, with the benefit of hindsight: The 2000 election will always be remembered for the 537 votes that divided Al Gore and George W. Bush in Florida, which were worth 25 electoral votes at the time. The final score was 271 to 267 after Bush v. Gore came down. (There was one faithless elector, so sometimes Gore’s tally is listed as 266.) Gore also lost New Hampshire, which went to Bill Clinton twice, by a mere 7,211 votes. Had he ceded Florida to Bush ahead of time and redirected his substantial Florida campaign to the Granite State, those four electoral votes would have made the difference — and saved us a lot of bellyaching over American democracy.

Sure, that’s easy to say 19 years later. But in fact, we can quantify how potentially “winnable” an election was for the losing candidate if they had followed this strategy. Doing so involves computing the minimum number of votes he or she would have needed to flip enough states to get 270 electoral votes, the margin of victory in the Electoral College, without going any further past the finish line than necessary. I call this the Deficit Factor. Determining it involves what’s known as the Knapsack Problem in computer science, where you pack as much weight as you can in a finite about of space — with the variation that we’re packing as few votes as necessary into a knapsack that narrowly flips the election. This is accompanied by a Surplus Factor, which is the number of extra votes the losing candidate won in states he or she captured if you reduce the margin of victory to one vote.

While a one-vote victory is facetious, the hypothetical ratio of deficit to surplus is an excellent way to see how badly the distribution of voters-by-state punished the losing candidate. Here’s how the voter distribution problem has affected the results in the past seven presidential elections, which offers some insight into how 2020 may play out.

1992: Deficit Factor of 1,398,196 votes

Incumbent President George H. W. Bush badly lost the Electoral College by 370 to 168 to then-Arkansas Gov. Bill Clinton. It would have taken fewer than 1.4 million extra votes for him to narrowly snag six states by one vote: Georgia, Illinois, Kentucky, New Jersey, Ohio and Pennsylvania, good for precisely the 102 electoral votes Bush needed to get to 270. There are several similar combinations that don’t involve both Pennsylvania and Ohio, with Deficit Factors under 1.5 million votes.

That said, Bush’s Surplus Vote was only 1,466,886. So while the imaginary Phillies method, which allows one to reapportion votes between states after the fact, could have eked out an almost comically close election demanding a 50-state recount, it wasn’t practical.

Verdict: Not flippable.

1996: Deficit Factor of 1,724,700 votes

Four years later, President Clinton widened his Electoral College victory to 379-159, meaning Bob Dole needed to capture 111 of Clinton’s total to win. (That’s half the margin of victory plus 1, since those 111 would also be subtracted from Clinton.) His best avenue was to take Arizona, California, Florida, Kentucky, New Mexico and Tennessee.

Seeing as his surplus was only 1,587,522 votes, even the bank couldn’t save him.

Verdict: Not flippable.

2000: Deficit Factor of 538 Votes

The number of extra votes Al Gore needed in Florida is conveniently the same as the number of electors, so it’s easy to remember — even when many Democrats of sufficient age would prefer not to. In addition to the previously mentioned New Hampshire Alternative, Gore could have won the election by turning Nevada with 21,598 votes or West Virginia with 40,979.

Given that his surplus was 6,477,548 extra votes in states he won, Gore had hundreds of alternate paths.

Verdict: Extremely flippable.

2004: Deficit Factor of 115,573 Votes

Sen. John Kerry lost the electoral vote by 286 to 252, meaning he needed to pick up 18 electoral votes to win. Curiously, the minimum viable alternative does not include Ohio, the archetypal swing state. Instead, if Kerry could have snagged Colorado, Iowa and Nevada — at the expense of never setting foot in Ohio — he would have won.

Winning Ohio by itself would have worked as well. He lost there by 118,599 votes, just a bit more than the combined extra votes he needed in the aforementioned three states. Kerry’s surplus was 5,758,792 in states he won, so again, there were plenty of Democratic votes to go around.

Verdict: Very flippable.

2008: Deficit Factor of 1,230,018 Votes

While John McCain had a surplus of 4,932,922 votes in red states, all of the smallest deficits involve a capture of six blue states. (There’s a modestly better solution involving seven states, which requires the algorithm to test 1,560,780 septets, but six already seems like an outside number of states to turn.)

The minimum viable alternative scenario requires McCain to capture Florida, Indiana, New Hampshire, North Carolina, Ohio and Pennsylvania, netting him 98 electoral votes, one more than he needed to pick up.

Verdict: Minimally flippable.

2012: Deficit Factor of 429,468 Votes

Mitt Romney had both a larger surplus than McCain — 6,800,601 votes — and less ground to make up, with 64 extra electoral votes required to win. His best path was to capture Florida, New Hampshire, Ohio and Virginia, though there are many alternative routes under 600,000 votes that don’t involve Ohio. For example, 595,271 votes could have very, very narrowly flipped Florida, Iowa, New Hampshire, New Mexico and Pennsylvania.

Still, that’s a lot of travel time.

Verdict: Slightly Flippable.

2016: Deficit Factor of 77,747 Votes

With a surplus of 11,226,150 votes, there are many alternatives other than Pennsylconsigan. A combination of Florida and Arizona would have cost 204,147 extra votes — which isn’t nothing, or even politically realistic. That’s also the 30th lowest deficit, though some of the smaller ones include Alaska, which, while it doesn’t require a lot of votes to flip since the population is so small, last went Democratic in 1964.

Verdict: Very Flippable.

What all this means for 2020

I’m not in the prediction game, but I’ll make an exception here: If President Trump wins reelection, the Democratic candidate’s surplus will likely far, far exceed his or her deficit. The same will probably be true for Trump if he loses.

But generally speaking, in our fairly small sample of seven presidential elections, the contests that Republicans won are much more fungible than those Democrats won. Both parties are too clustered. But if there’s one advantage for Democrats, it’s that they’re so clustered that their candidate can afford to be overly attentive just where it matters most.

This is the millstone of the Electoral College, but there is a way to shed it: Choose the map that looks most winnable and is closest to 270 electoral votes and go for broke. That strategy has worked quite recently.


Historical presidential results come from the MIT Election Data + Science Lab and were carefully spot-checked against other sources. All scenarios use contemporary state allocations of Electoral College votes accessed from the National Archives website, which required modest manual corrections. Nebraska and Maine, which divide their small number of electoral votes by Congressional District, were treated as winner-take-all states since reliable historical data on the per-district returns is difficult to assemble and are unlikely to change the scenarios.