Election 2020

Election Models Predict Trump Defeats Clinton

Publicly predicting presidential election winners can be tricky. If you go with conventional wisdom and are wrong, you have to lay low for a while and hope people forget your blunder. Just ask the Chicago Tribune.

If you go with the candidate who is down in the polls, has high negatives, and is not supported by the political class and you end up right, you may get 15 minutes of fame and all the glory that goes with it. Several players are taking that risk and predicting a Trump victory using both traditional and social media based computer models.

Allan Lichtman is a distinguished professor of history at American University in Washington, D.C. He has correctly predicted 30 years of presidential election outcomes.

Lichtman’s model analyzes the answers to 13 true/false questions. If six or more of the answers are false — that is, they go against the party in power — they lose. If fewer than six are false, the party in power gets four more years. For 2016, he is putting his money on Trump.

Helmut Norpoth is a political science professor at Stony Brook University in New York state. He has correctly predicted the outcomes of the last five presidential elections.

Norpoth’s model uses a candidate’s primary performance and the tendency for voters to switch parties during election cycles to determine the winner. For 2016, he too is putting his money on Trump.

Lichtman and Norpoth developed their models using extensive historical data. However, a number of initiatives are underway at developing computer algorithms based on real time trending models based wholly or partially on social media activity. All this is possible because as our mathematics has gotten more sophisticated and our computational tools more powerful, we have become better at amassing huge amounts of data in the hopes of seeing trends.

Influencing behavior and predicting the future may be the opposite sides of the same coin, because if you influence the desired behavior you might get the desired future. It’s no coincidence that when you shop online for a Brand X 50-inch HDTV that every website you visit after that has an ad for a Brand X 50-inch HDTV. Or the grocery store you frequent knows exactly what coupons you may need and stores them in your account or sends them to you in an email.

Success at predicting the future took a giant leap in the 1960s with better weather models and faster computers. Today, 10-day weather forecasts are routine. But what’s predicting the weather have to do with election results or politics in general? Plenty.

Knowing what the weather will be goes beyond helping you decide what to wear that day. Today, computer models are being exploited to argue for a complete transformation of our economic system in order to prevent the average temperature of the earth from rising 1.4 degrees by 2100. Or is it dropping by 0.9 degrees by 2150? I’ve lost track.

Weather and climate are not the same thing but their predictions are based largely on physical and mathematical deductive models. However, predicting presidential elections involves seeing patterns, something humans do very well. The one advantage computers have over humans is the capacity to handle large amounts of information. Can machine brute force win over human finesse?

The MogIA artificial intelligence system correctly predicted the last three U.S. presidential elections and is picking Trump to be the next commander in chief. The MogIA crunches massive amounts of Internet data including social media. It shows that Trump has overtaken Barack Obama’s peak in 2008 by 25 percent.

Another company, Cognovi Labs, an Ohio-based analytics startup, has a tool built on technology developed at Wright State University by Dr. Amit Sheth called Twitris. Cognovi correctly predicted the Brexit outcome hours ahead of the final vote, going against what traditional polling was indicating. James Mainord, Cognovi CEO, says at this point in the election there is still too much movement in the public mood to say with certainty who will win.

“If I had answered this question on Oct. 28 at 12:30 p.m., my answer would have likely been Clinton,” Mainord told PJ Media. “Even though her support had been harmed by a 10/24 HHS report on Obamacare and continued Wikileaks drops, it appeared that she was holding on to support. The FBI announcement by James Comey was very damaging and we’re following closely to see how much harm it has done.”

So that puts three models for Trump and one on the edge.

There are some adults within the Democratic Party who understand that no matter what happens, even if Clinton wins, she won’t be able to govern (that is, pass liberal legislation), and some are even asking for divine intervention. Proving once again that there are no atheists in foxholes or when in legal/political trouble.

Future campaigns will spend big bucks for software that can measure, react to, and then influence voting behavior. Commercial marketing departments will do the same for their products and services.

Computer models are only going to get better and more sophisticated. Forget about measuring data in terabytes; petabyte storage drives are next, and exabytes, zettabytes, and yottabytes will eventually follow. Each designation adds three zeros and a comma to the previous number.

In 1814, the scientist/mathematician Pierre-Simon Laplace famously suggested a “sufficiently powerful intellect” could, if it knew certain things at a given time, calculate the evolution of the system at any other time. In laymen’s talk, in a way he was saying the future is theoretically predictable.

It didn’t take long for later scientists to conclude that the only thing predicable about the future is its unpredictability. But then, they didn’t have access to our modern data handling and processing capabilities.

With increasing data handling capacities and accompanying faster computing speeds, are we are rapidly approaching Laplace’s concept of “sufficiently powerful intellect”? If you scoff at that idea, and many purists rightly will, just keep in mind that if you are willing to take a little liberty with definitions and accept a little fuzziness in the predictions, we may get close enough for all practical purposes.