On Donald Trump, Superforecasting, and Granularity

I’ve been reading an enormous amount of punditry lately from smart folks trying to make sense of all the political weirdness going on. Donald Trump is like Hitler. Donald Trump is like George Wallace. Donald Trump is the ideological heir to George Bush.

We’re all trying to make sense of what feels like a very different type of election cycle. But whenever you compare one historical figure to another, the answer is always going to be, well, sort of yes and sort of no.

If you cherry pick the right data points, you can compare Donald Trump with anyone. For example, I could say Donald Trump is really like JFK. He’s from a wealthy east coast family, looks slightly jaundiced, has an Ivy-league education, and is famous for seeking out and sleeping with lots of attractive women. See, just like JFK!

But, if we can play this game to compare Donald Trump to anyone, it shows us that the question, “who is Donald Trump like?” is just a very Poorly Worded Question.

We all do it. It’s hard to avoid given the inherent vagueness of natural language. But if we wish to acquire a better understanding of our world, we have to ask better questions. Questions we can answer precisely, so that we can determine whether our answers are accurate.

Instead of asking, “will there be a political realignment at some stage in the not-so-distant future?” or “Who is Donald Trump like?” let’s try to get more precise. (Philip Tetlock’s book Superforecasting is great training exercise in thinking in this way). Rather than asking a question whose answer cannot be measured, let’s ask one that can.

For example, here’s one: “in the next five years, will at least 30% of those who currently self-identify with either the Republican or Democratic Party self-identify with a different political party?”

I think that’s a more precise articulation of the realignment question, and it’s something that could be measured with decent accuracy through polling.

I think the answer to that question is no. I’d put the odds at less than 20% of the above-statement coming true. So, if I’m right, perhaps a full realignment is unlikely.

And since I’m playing this game, I’ll ask another question that I think is more interesting, and more reflective of the weirdness we’re likely to see over the next 8 months.

“Will the person who achieves the third-highest vote tally in this fall’s Presidential election receive more than 5% of the total votes cast in the election?”

To that question, I answer, “yes.” And I would put the odds at 62%. It’s happened five times in the last 100 years, or about once in every five election cycles. Given the lack of success that Trump has had in heavily Republican states such as Utah, Wyoming, Idaho, Texas, and Oklahoma, and the likelihood that he wins the Republican nomination, I think that someone will run to fill that political void. I think a more traditional conservative will oppose him in the general election.

I may be right about that prediction. I may be wrong. But either way, the accuracy of my prediction can be measured. And that’s the first step toward coming up with any sort of meaningful answer to any question.