It was easy to predict mass car ownership but hard to predict Walmart. -Carl Sagan

The human brain is a wonderful, powerful thing – but it has some blind spots. One of these, I believe, is the way we compare things. Our go-to comparison method is “This versus that”, and it has a close cousin: “Before – After”. We’re good at comparing things that are similar – that’s why the scientific method, for instance, relies on controlled experimentation. Hold as many variables still as possible; test one at a time. This method can be very good for revealing to us the way things are. But it’s usually quite poor at hinting to us the way things might change.

Why is that so? Because AB testing (or Before-After testing, which is really the same thing) requires a key assumption: that you’re testing the right variables in the first place. And when trying to predict the future, and guess at ways that the future might be different from the present, the whole point is that we don’t know which variables will be important. To be tautological about it, we don’t know what the future looks like because we don’t know what the future looks like.

But what if we’d like to guess? How could someone, having just seen the Model T, be able to predict Walmart?

The first thing we can say for sure about the future is that it will be different. What does that mean? Well, one thing it means is that future you will not have the same constraints as present-day you.

The second thing we can say confidently is that business models are built around constraints. So if future constraints around X are different, then future business models around X will be different as well.

The third thing we can say is that constraints exist for a reason: they’re hard to break. It usually takes a lot of innovation, effort, money and time to invent and ship some new technology that fundamentally breaks a constraint. But when they do, it’s big news.

Let’s come back to our Model T for a minute. Before cars, a few miles distance was a significant constraint. Cars were difficult to invent and build. But once you had one, that distance was significantly less difficult to travel. The constraint went away (or at least changed significantly). Clearly, everybody wants a car now.

So, looking at a Model T for the first time, I should be able to grasp the notion of ‘now I can travel over distance faster, and in less time.’ That’s an A-B test; a ‘before-after’ comparison. But what if I want to understand the second-order consequences? How I know where to look?

To begin: what constraint went away? The time required for an individual traveler to traverse many miles, at will, decreased dramatically. So what businesses might be built with the old constraint in mind?

In our case, there’s a big one: selling things. Before cars, if you sold some sort of consumer good, your customers were people within reasonable traveling distance of you. In cities that meant one thing – your business might look like a corner grocery. In the countryside another – it might look like the town general store. Either way, your businesses were built around a specific constraint – that all of a sudden goes away once you own a car.

The first order consequence of cars was that you could move places faster. But the second order consequence of cars is that new business models became possible: now you could create a store that sold to everybody within 50 miles of you, not one mile. That store probably looks very different than the ones that came before it. It looks like Walmart.

So what does this mean for making predictions today? Start by looking for constraints. There are still a lot of them: food, water, energy; homes, jobs, transportation; safety, education, happiness; constraints are everywhere. What happens if they go away, or change significantly? What current business models, built around those constraints, will suddenly be very out of date? How would brand new business models, free from those constraints, be different? How will the future be different from the present?

It was easy to predict online ratings; it was hard to predict Airbnb.

It was easy to predict a taxi-hailing app; it was hard to predict Uber.

It was easy to predict hot-or-not; it was hard to predict Facebook.

Until we’re in the future.