Even a Broken Clock is Right Twice a Day


Nate Silver was supposed to be some sort of modern-day prophet. He predicted Barack Obama’s win, after all.

But he is just as mortal as any of us, and elections remain an impossibly difficult thing to predict.

When you’re playing the game of predicting the future, it’s important to note that even people who are right for the wrong reasons are still right. Until they are wrong. Then they’re wrong. And if they were right for the wrong reasons, the chance they’ll get the next one right is really, really small.

This applies in science much more than people realize. See, when you do an experiment, you introduce errors that will vary the result one way or the other. After all, we’re only human. The more complicated the experiment, the more that can go wrong. Just because you see a particular result once or twice doesn’t mean that those results will happen all the time. Even after running countless experiments, you still cannot predict that something will or will not occur. No one knows what the future holds. Given that, every scientific prediction is really footnoted with some kind of uncertainty. You would do well to investigate the uncertainty before selling your soul to a scientific theory.

I’m studying machine learning right now, and the most important thing I’ve learned is that you always have two options for making predictions, and those two options are likely pretty good predictors. The first is to just randomly guess. If there are 4 possibilities, and you randomly choose one, then there’s a 25% chance you got it right. That’s much better than many algorithms will give you with certain kinds of data. But you also have another option: choose the winner. That is, look at what is most likely to happen, and just choose that. If the most likely scenario is only 1% likely (because there are a lot of possible scenarios), then choosing that as your prediction will make you 1% accurate, which is often better than many algorithms will give you with certain kinds of data.

It’s important that you look closely at the algorithm you use for predictions, and measure it with real-world data. If it doesn’t do better than random choice or majority, then it’s not good at all.

Nate Silver was never a prophet. He had no supernatural powers. His predictions seemed to work, but so far, he’s really not doing much better than random choice among the most popular candidates. I feel comfortable ignoring Nate Silver and others like him. I may pay attention the same way I’d pay attention to a really good chess player or some other spectacle worth watching.

In the upcoming election cycle, do yourself a favor. Ignore the polls. Just vote your conscience. Vote in a way that you will feel comfortable telling your grandkids and feel comfortable knowing you voted for the guy if he gets elected. What’s more, by ignoring the polls and people like Nate Silver, you’ll free up valuable time and save some money. You won’t need as much heartburn and headache medication, after all.


Tags: , , , , , , , , , , , ,

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: