Influences on Presidential Elections


Modern Presidential elections are decided by narrow margins - a few hundred thousand votes one way or the other can determine the winner. The more motivated party, that is the one who turns out more voters, usually wins. Motivating your own voters while discouraging your rival's is a big part of the election game.

There are three main influences on the outcome of a Presidential election.

1.) The first is run-of-the-mill politics - all of the ways politicians and influence peddlers try to convince voters to support their candidate over the other guy. Nearly all of this is legal, protected by the First Amendment, and accomplished through various media outlets, debates, rallies, advertisements, etc. The result is generally a better-informed public.

2.) The second is direct, illegal manipulation of ballots at voting stations, mailed-in ballots, and electronic voting machines. This rarely happens in large part because of the broadly distributed, mostly analog nature of voting. Instances of sophisticated digital vote tampering may increase if voting moves online.

3.) The third is the emergence of disruptive technology. One example is high-resolution (household to neighborhood level resolution) vote-prediction maps built from trillions of datapoints gleaned from social media sites, consumer databases, publicly-available archives, and statistically-rigorous derivatives. We have yet to see robust probability maps splashed across the evening news prior to election day, but that day is coming.

Consider the consequence of such maps. How motivated would you be to go out and vote knowing that the outcome of the election has already been precisely predicted? Just seeing such a map two weeks before an election - even if the prediction were wrong - would have tremendous influence on voter turn-out. And recall, all you need to do is influence a small percentage of likely voters to swing the outcome. And what if the predictive map is right once? Twice? Three times? Once high-resolution voter preference can be reliably predicted and mapped, the election game is up.

There are, however, mitigating factors to a disruptive technology. Voters learn. Voter preferences are well known to change over time with or without disruptive technologies. Politicians come and go, and new topics distract the media every couple weeks. Nate Silver of fivethirtyeight.com, prior to the 2016 Presidential election, predicted Hilary Clinton victory based on results of an state-of-the-art voter preference model. His model was valid until about 9pm PST on election night, a which time it collapsed catastrophically. Vote-prediction maps, in a similar way, have a shelf life. They will remain influential for a short period of time and must be recreated for each new election. Once commonplace, predictive maps are unlikely to change voter behavior.

Are high-resolution vote prediction maps (or other disruptive technologies) a danger to our election system? Maybe. They certainly are coming to a TV near you. Some will turn out to be correct. Others will not. Likely most of us will quickly grow accustomed to them rapidly, once their fatal flaws are discovered. At which point, they become just another tool of politic influence, like advertisements we all ignore today.

Last 50 Posts
All Posts by Month