Thumbstoppers

A few years ago, I spent five days in a glass conference room in Palo Alto learning how data scientists could predict what people would do.

I was consulting for a company building a credit algorithm. Outside the windows were eucalyptus trees and perfect California weather, sun, low humidity, a gentle breeze.

Inside were coffee cups, notepads, pencils and a lesson I have never forgotten.

The instructors from UC Berkeley described how seemingly unrelated details of a person’s online behavior predict whether that person would repay a loan. One example I remember was based on if someone let their phone battery die.

People who did were much more likely to default or be late in their payments, they said this one variable was almost as predictive as a normal credit score.

What stayed with me was the method. The machine did not need to know why you did something. If a pattern helped predict what you would do next, it had value.

Their focus was on thumb-speed, using the camera to evaluate your visual focus on the screen and how to interpret that in real time to adjust the next piece of content to be the most likely to hold your attention.

I left that room thinking about credit. I should have been thinking about the rest of human life.

The capacity to observe behavior and predict the next choice is the foundation of the worlds largest businesses.

A platform learns what makes you pause, what makes you angry, what makes you feel recognized and what makes you reach for the phone again. It ranks the next piece of information accordingly. You supply the attention. They sell access to it.

The product is not the video, the post or the photograph. The product is your continued presence through optimized design to trigger dopamine responses as your time vanishes and their bank accounts swell.

Information that leaves you wiser may take ten minutes and send you back to your life.

A provocation machine takes ten seconds at a time and consumes half of your life.

What happens when the easiest money goes to capturing attention rather than improving a life? What is the opportunity cost of all that lost time?

A person who cannot direct his own attention has less capacity to think, work, love, learn or decide what matters. The platform gets better at predicting you while you get worse at hearing yourself.

This is the difference between technological power and human advancement. A tool that helps a scientist understand a disease creates value. A tool that helps a student learn creates value. A tool that helps a person find a friend creates value.

But when the financial reward comes from keeping all three on a screen for as long as possible, usefulness becomes the bait for extraction.

We should judge these businesses by what happens to the people who use them, not by how many hours they manage to take. Do we leave more capable? More informed? More connected to one another and to our own lives?

Or do we leave agitated, dependent and ready to come back for another hit?

The people in that Palo Alto room taught me that a machine can know what I am likely to do without understanding me at all.

We are building an economy that calls that intelligence.

I call it a warning.