By Chris Wilson
The 2012 election is rapidly approaching the point where prayer is a legitimate campaign strategy. After well over a year of campaigning, President Barack Obama and former Massachusetts Gov. Mitt Romney have done everything in their power to sway the ad-saturated electorate in their favor. Now they need to get lucky.
Luck isn’t always relevant to elections. In 1984, Ronald Reagan could have personally shattered every mirror in America, and he still would have won by a landslide against Walter Mondale. You may recall that the 2000 election was a bit closer than that. If Al Gore had amassed 538 more votes in Florida—that’s regular people votes, not electoral votes—he would have won the presidency.
We can imagine all sorts of scenarios beyond the campaigns’ control that might have altered that election. Heavy rain along the Florida panhandle might have suppressed turnout in the Republican-rich region, or a few traffic jams in Miami could have deterred enough Democratic voters to make Bush’s lead indisputable.
It’s actually quite difficult to distill precisely how close an election was after the fact, given that the electoral combinations rapidly multiply as you accumulate “what-ifs.” The best way to explore the question is simply to “rerun” the election as a computer simulation, introducing a small random variance and watching how often the results change.
That’s what the following tool does. You can pick any election going back to 1944, set the maximum amount of luck you want to introduce—1 percent of the popular vote, by default—and “replay” the election.
When you click “Run Once,” this interactive takes every state and asks, What if things had been a little different?
The tool uses your browser’s random number generator to shake things up a little. If you go with the default of 1 percent random variation, for example, any state that was decided by less than 1 percent of the total vote in that state has a chance of flipping to the opposite party. If a state flips due to this random variance, it will change colors and its electoral votes will be awarded to the other side. (For simplicity, votes for third-party candidates are held constant.)
Because it’s random, the results will be a little different every time. Running this homemade experiment a few times will quickly give you a sense of how often a little luck can alter the results. For a more mathematical picture, you can run hundreds or thousands of trials—up to 5,000—at once and see a distribution of the results on the chart below the map. (Running thousands of trials at once may take a few seconds, depending on the device you’re using.) If you run enough trials, you’ll get every possible electoral combination of close states, as well as which particular combinations were the most likely to occur.
In 2000, an across-the-board variation of just 0.25 percent of the vote in each state flips the election to Gore about 47 percent of the time—but not always with the 291 electoral votes he would have amassed by winning Florida. In absolute votes, though with less than a tenth of the total ballots cast, New Mexico was even closer than Florida in 2000, going to Gore by 366 votes. About a third of the time, this tiny 0.25 percent variation flips New Mexico to Bush’s column, but its five electoral votes are not enough to counterbalance any change in Florida.
The 1960 election, by contrast, was not as close as popular history might suggest. Even a 3 percentage point variation in the results turns the election over to Nixon only 12 percent of the time, though it produces many different scenarios. And in 2004, even though the exit polls initially suggested that John Kerry might win Ohio, even a 5 percent “luck factor” aligns the presidency in Kerry’s favor only 5 percent of the time.
At the moment, it appears entirely possible that the 2012 election will be the second closest in modern history after 2000. At this rate, the candidate who loses appears doomed to spend the rest of his days wondering what might happened if the dice had fallen just a little differently.