Ideas Are Bred, Not Born.

At Affinnova I sat in windowless rooms with executives who ran brands most of the world has held in its hands. On the screen was a grid of package designs. They assumed an agency had made them. None had been chosen by a committee. They had been evolved. We ran a live round. The design one executive’s team had championed for months held on for a while, then lost to a combination nobody on her team would have picked. “We would never have made that,” she said. She said it with something close to wonder.

The method was not ours alone. In the 1970s John Holland formalized genetic algorithms at the University of Michigan. Nature solves hard design problems without a designer: a population of candidates, a way to measure which do better, and a method for combining and varying what works. We brought that logic to a human question. What will people actually choose? Over time it changed how I treat my own thinking.

Four rules for breeding an idea

Variation comes first

Produce many versions before you commit. The first version is usually the least original.

Selection must be honest

Set up a test that can tell you something you do not want to hear. Then listen to it.

Recombination is where breakthroughs live

The winner that day was two ordinary ideas, a color and a claim, that nobody had put together. The creator’s job is often less to invent than to connect.

Mutation keeps you from settling

Let in the unexpected: a new field, a strange conversation, a book you would not normally read. Without it, your thinking converges on a good answer and never finds the great one.

What the research adds

Adam Grant, in Originals (2016), draws on the psychologist Dean Simonton’s research on creative productivity. The most celebrated creators did not have a better hit rate than their peers. They produced far more work, which gave them far more chances at a hit. Your idea is not you. It is a hypothesis you are carrying for a while.

Recombination has data behind it too. Brian Uzzi and his colleagues at Northwestern studied millions of scientific papers for a 2013 study in Science. The highest-impact work was grounded in conventional combinations of prior work, with an intrusion of unusual pairings. Mostly familiar, with one surprising join. That is the color and the claim.

Breed your next idea

1. Write ten variantsTake the idea you are most committed to. Change the customer, the price, the form, the channel, the core promise.
2. Pick three and recombineLook for features from different variants that could join. Notice whether your original still wins. Often it does not.
3. Design an honest testChoose a test whose result could embarrass your favorite. Decide in advance what result would make you drop it.
4. Schedule a mutationSpend one hour a week outside your field. Each month, write down one idea from it that might apply to your work.