INSIDE THE BOX: How Constraints Make Us Better

Guest post by David Epstein, whose new book is out this week!

I’m so excited to share this guest post from David Epstein. David is the author of the new book INSIDE THE BOX: How Constraints Make Us Better, and of the best sellers Range and The Sports Gene. You can learn more about his work at davidepstein.com.

David is a fabulous writer and I am always excited by his work. INSIDE THE BOX is, like his other books, an absolute must read and is already on my list of books that will make you smarter.

The Signs That Were Supposed to Save Lives

In 2012, Texas started displaying the year’s running traffic-death count on its electronic highway signs. The intuition was obvious: remind drivers how many people had already died on the state’s roads, and they’d be more careful.

Instead, the signs caused crashes.

Texas ran the program for only one week each month, which gave a pair of researchers a near-perfect natural experiment. Same roads, same season, same time of day, same day of the week — the only difference was whether the death count was up. They studied eight years of data. The signs were associated with a measurable increase in accidents — a few thousand additional crashes a year, with the worst effects on the most demanding stretches of road.

The signs worked exactly as designed. They grabbed drivers’ attention, and that’s what got people killed.

Working memory is the bottleneck of human cognition. You only have so much of it, and once it’s full, something has to give. We all know this implicitly. It’s why you turn the music down in the car when you’re looking for an address. The music isn’t impairing your vision, it’s taking up working memory. The Texas signs were demanding capacity the drivers needed in order to make decisions. The drivers weren’t reckless. They were full.

By the time I came across this study while researching my new book, Inside the Box, I wasn’t surprised, because this kind of design failure had become so familiar. We do this all the time. We build organizations, policies, products, and our own lives for an imaginary user: someone with infinite attention and perfect self-control. Then we’re surprised when actual humans don’t behave the way the imaginary perfect one does.

A group of researchers in Brazil and the U.S. coined a useful pair of terms for this. They distinguish between the “ideal citizen” — who weighs costs and benefits perfectly and acts on the result — and the “real citizen,” who has limited attention, inconsistent preferences, and the cheerful but inaccurate belief that future-self will make sacrifices present-self refuses to make. The ideal citizen does not exist. The real citizen is who you’re designing for, whether you know it or not. And the most useful thing you can give a real citizen is rarely more information, choice, or motivation. It’s the right constraint — a boundary that makes the good decision the easy one.

A fascinating illustration comes from a tea company in Malawi. The company offered employees a “pay me later” program: workers could elect to have part of each paycheck withheld and paid out in a lump sum after the harvest season. Employees who said they wanted to participate were then randomly assigned to either be enrolled in “pay me later” or kept on their normal schedule. Months after the experiment ended, the workers who had been in pay-me-later owned about 10 percent more assets than the control group, and were more likely to have made improvements to their homes.

For an ideal citizen, the program shouldn’t have mattered at all. They could have transferred the same amount to savings on their own. But ideal citizens don’t exist. Real citizens needed the constraint.

Most of the policies and habits that actually work look like this. They don’t ask the user to be a better version of themselves. They take the user as they are and put the desired behavior on the path of least resistance. A 401(k) with auto-enrollment. A grocery list made before you’re hungry. The phone left in another room while you write. The running shoes laid out the night before. Or, as I’ve seen in action with pro sports teams during the draft, asking each member of the group to write down their pick before anyone in the room has spoken.

The instinct, when something isn’t working, is usually to demand more discipline — from employees, from citizens, from ourselves. The evidence almost always points the other way. The trouble isn’t that humans are too weak for the system. It’s that the system was designed for someone who isn’t there.

If you want better decisions from real citizens, build the road for the drivers you actually have. Not the ones in the brochure.