It’s All About the ppm

On a cool Shanghai afternoon in March, 2015, it was time for an afternoon coffee break. I start the European espresso machine and remove a coffee mug from the heated cupboard. My heart rate spikes as I notice a beetle floating in the espresso machine’s water tank. I began prodding the machine in a desperate attempt to detach the water tank. I wanted to throw out the water and refill it from our stack of Nestle 6-oz bottles. Benny stops me and says “just take the guy out and put the water back!”. I stare at him in disbelief.

Benny is a highly-intelligent Danish engineer who has been living in China for a decade. He continued:

Noah, look at how tiny this bug’s legs are! Even though the sewage line is exposed over there, how much crud could he have picked up? Then, how much would be transferred into this water tank? Finally, there’s a lot of water in this tank and it’s going to be boiled. What’s the ppm of dangerous dirt particles in that’s going to be in your coffee?”

An Insect Standing on Water

This memorable experience occurred at a contract manufacturing site in Pudong. Pudong is a special economic zone on the edge of Shanghai. Each day, my colleagues and I would triage hundreds of iPhones that had come off the assembly line. These units were prototypes that failed to meet the company’s high quality standards. Each of us sat at our place in the failure-analysis room with our stack of units grouped by failure symptom. We would investigate to determine the root-cause of different failures.

There will always be failures in manufacturing, but the important question is: “what is the fallout?” Fallout tells you how often a failure occurs and is measured in dppm (defective parts per million). It tells you whether a majority of the units are defective, or a very rare few. In other words, I have this unit that isn’t working right and cannot be sold to customers. If this failure only happens once per million units made, we can mark it up as a loss and forget about it. But, If this failure is commonplace, there is a real problem that we have to spend time to fix.

Taking Benny’s advice went against my instincts, but his reasoning is correct. I followed his advice. Ppm (parts per million) is a measure of the extent to which something is adulterated, defective, or different. Discussions involving ppm are everywhere in statistics, science, engineering, and manufacturing. We want to know how many dirt particles per million particles of water solution.

In the real world, circumstances are rarely binary, yet we treat them as if they are.

We ask questions like “is it safe for me to go to the grocery store?”, “is this water safe to drink?”, “is this software free of bugs?”, “will I get coronavirus by spending time with my friends?” or “is my car safe to drive?”. These questions are non-binary; they do not have a yes or no answer. Instead, you must assign a risk to the action. Risk is the outcome’s severity multiplied by the chance of a occurring. We then weigh the risk against the benefits of an incident-free outcome. A risk matrix is often employed to categorize such multiple risks. Time should be devoted to the items with highest risk.

A Risk Matrix

Most of the human population is not hyper-rational. People have vague notions of risky behavior, but they fall prey to dozens of biases. Individuals increase their risky behavior when negative outcomes aren’t observed. This leads to thinking that behavior is safer than it is. But, if someone has a negative first experience, it will be difficult to convince them to try again.

Learning can help correct this default behavior. Many people are more afraid of flying on a commercial airliner than driving in their car to the airport. Statistics show that you are much more likely to die in a car accident than in a commercial plane accident. With a basic understanding of data like the chart below, you can grow and adapt to live a better life.

dp100pm (Deaths per 100 million passenger miles)

For more information on risk, I recommend Nassim Nicholas Taleb who is the world’s leading expert on risk. He is not the kind of “risk expert” that lead to incidents like the 2008 housing crisis. He has authored many books and specializes in high-risk, low probability events.

Traveler, musician, and engineer.