Margins and discipline

Humans are notoriously bad at probability. It's much easier to stereotype, or make decisions based only in emotion. Probability is hard, and doesn't always follow our wishes.

A recent tweet about a handful of candy caught my attention. There's an image of a small bowl of Skittles, along with the following caption:

If I had a bowl of skittles and I told you just three would kill you. Would you take a handful?

That's our Syrian refugee problem.

I decided to turn this tweet into a probability exercise.

A quick note before I get started: This is an incredibly heart-breaking topic to research. Labeling people and classing them with a spreadsheet goes against every moral fiber in my body. However. Candy is sweeter but illustrations like this still have real life or death implications for thousands of people. My primary data source is Wikipedia, which I realize is less than reliable. If you disagree with my data or assumptions, the framework should still hold with updated information.

Question 1

How many Syrian refugees will kill you?

As you might expect, I had a hard time finding concrete facts on Syrian refugees who will kill you. My life experience with refugees has been that they are bright, smart, and kind - just like my cousins. I haven't encountered any killers. But let's get back to the data.

Let's start with the assumption that a refugee from Syria who will kill me is a terrorist. Wikipedia has numbers on terrorist attacks, and Wikipedia also has a list of state sponsors of terrorism. For simplicity, let's assume that all terror attacks were carried out by terrorists who originated in one of the state sponsors.

Data from the last five years hold that 5,066 people were killed in 103 terror attacks. Enter another assumption that each attack was the result of five bad actors, or 515 bad actors in total. Based on the population of state sponsors of 138 million people, we could estimate that approximately one person in 268,000 is a bad actor. Assuming Syrian refugees fit this same probability profile:

One in 268,000 refugees will kill you.

Question 2

How does this compare to the bowl of Skittles?

Fortunately, data on Skittles is readily available. A one pound bag contains approximately 427 Skittles. The bowl in the picture is pretty small, but we'll assume that it holds an even pound.

We'll assume the bowl holds three handfuls, or 142 Skittles per handful.

We are told that three Skittles in the bowl are bad. There are three handfuls in the bowl. Which works out to approximately one killer Skittle per handful. Assuming this ratio of killer Skittles to regular Skittles:

One in 142 Skittles will kill you.

Question 3

Is this a statistically accurate illustration?


Question 4

What does this all mean?

This is how policy is made. We the people vote for lawmakers. The premise is that lawmakers will make laws that satisfy their constituents - or they won't be re-elected.

We live in a world with limited resources. With these resources we try to maximize our progress in eliminating sources of discomfort and death, and build a better world for us. Do we spend our dollar killing mosquitoes, or feeding a hungry child, or researching cancer, or adding soldiers, or building bridges, or funding libraries?

In 2010, one in 704 Syrians was a medical physicians. Assuming every physician saves five lives, one in 140 Syrian refugees saves a life.

Using the numbers above, for every killing we prevent by closing the border, we lose out on saving 1,903 lives.

Is it worth it?

Question 5

What would this actually look like in Skittles?

First we'd need to increase the bowl size to hold 1,882 pounds. Imagine a pickup truck bed heaping full of Skittles.

In this giant mound of Skittles would be three lives lost and 5,710 lives saved. You can take one handful. Choose wisely.