Small Data: Why an entire US Election comes down to Virginia..
Everyone likes predictions. I’m no expert, but I do love a good election prediction. Unfortunately as far as Clinton Vs Trump goes, it is too close to call. All I can predict is this: The state of Virginia will be the decider.
So, how did I come to this outrageous prediction? Lots of big data crunching and analytics? Nope. I did it through Small Data. What is that, I hear you ask? Let me provide some history…
Back in 1999 I earned my Masters in Knowledge Based Systems. It was essentially an Artificial Intelligence course with an emphasis on Systems. My final project was able to demonstrate that it is possible to optimize the input horizon into an Artificial Neural Network (think of this as a simulated brain good at generalizing and pattern matching) to provide much better and accurate predictions than if you force the ANN to use all of the input data. And that is because this is a question of efficiency. It is less efficient to have your judgment clouded by irrelevant data. The less noise, the better the prediction. The more emphasis only on relevant data, the better the prediction.
Training the ANN was pretty easy. I used something called a Genetic Algorithm which essentially codes which inputs will be considered and which ones won’t … and does it using a system similar to genes. The idea is you randomly create a bunch of these genes, train the ANNs based on their inputs and then log the results. You then take the ‘strongest’ genes (i.e. the genes that produced the ANNs capable of better predictions) and use them to reproduce with one another in randomly selected pairs. As time goes on you move towards an optimal list of inputs.
So that’s the history. It’s what I call Small Data. Those hidden, key pieces of information that really make the difference. For a less techy explanation of this incredible concept, see the book Small Data: The Tiny Clues That Uncover Huge Trends.
“So, what are the “tiny clues” that you found,” I hear you ask. Well, there’s only one. It is something called BREXIT. If you aren’t gasping now, then I’ll assume this isn’t a sore point and explain what this is and why it is relevant. I’ll also include a super condensed survey of one aspect of the current political climate which will help put everything in context.
In June this year, the United Kingdom held a vote to determine if it should leave the EU. I did not follow the vote very closely at all. So instead of watching lots of YouTube videos with all of the detailed debates in (something I did after the result was in, for reasons I will go into in a bit), all I got was what filtered across social media. And it was pretty nasty stuff. The main theme was that if you voted to leave you were a racist or you were ignorant. And that was pretty much it. The side who campaigned to leave were painted as openly racist and as such I felt compelled to go back and watch the debates to see if this was true when the result was in. It was complete nonsense, of course. But then, in today’s world, the tactic of calling someone who disagrees with you a nasty label, like a racist or sexist, etc seems to be becoming more and more popular. Thankfully people are getting sick of this BS.
My friend groups on Facebook fell into two distinct groups when it came to BREXIT. There were fellow white-collar professionals, well educated and working at large, successful companies. And then there were my friends who did not graduate college, worked blue-collar jobs and were in less of a financially strong position. What bothered me were the way group 1 attacked group 2 using the aforementioned tactic.
When I asked group 2 why they were voting the way they did, they would always reference the UK having more control. In one case a female friend of mine was driven from her apartment as every other flat in the block was given to immigrants on welfare who ran the place down with drugs and violence. Had my friend been better educated maybe she could have afforded to move out and into a better neighborhood with less trouble where she could pretend these kind of realities did not happen. But sadly that was not her experience. And group 1 refused to believe it — it was an inconvenient fly in their utopian ointment.
The polls the night before the BREXIT vote ended up being 8% wrong. By that, I mean the “Leave” campaign got 4% more than it was estimated to and “Remain” got 4% less. It ended Leave 52% — Remain 48%. One of the most credible theories put forward by many different sources were that people were not honest when polled. People are wary of being seen as a racist or xenophobe. And this, essentially, is what I am going to call the Brexit Effect.
And it’s completely relevant — just look at how the Clinton Campaign, backed by the mainstream media, are trying to paint Trump and anyone who votes for him as racists and xenophobes.
With the current RealClearPolitics electoral map, there are not many states that fall around a +7.5% Clinton lead; these should be considered the true swing states, in my opinion, as they would move to Trump if the Brexit Effect on the polls vs actual held true (Trump +4%, Clinton -4%). Of all of these, Virginia seems to be the one hovering consistently near that number. As of today, Michigan shows Clinton only with a +3% lead but even the most ardent Republican would have to concede that Michigan is not likely to go to Trump.
When the final polls are released the night before the election, keep a close eye on Virginia. My prediction is as follows:
Ohio, Florida, Iowa, Georgia: Trump
Pennsylvania, Colorado, Michigan: Clinton
Virginia: The next president of the United States of America.
And if I’m wrong, I promise not to call you a racist for laughing at my awful prediction.