Northwestern University physicist Adilson Motter has spent years watching power grids, flocks of birds, and brains do something remarkable: they work better when their parts are a little different from one another. Now, in a new study published in the journal Science, he and his team have built the mathematical framework that proves it—and explains why scientists missed it for so long.
For decades, the default assumption in network science was simple: perfectly similar parts make the most stable systems. A power grid with identical generators. A brain with identical neurons. A food web with identical species. Real-world networks, though, are rarely uniform. Generators differ, neurons differ in shape, birds differ in personality, and even our social relationships can be asymmetrical. Scientists traditionally treated these differences as flaws to be ignored or smoothed away.
Motter and his colleagues turned that idea on its head. They developed a framework that identifies exactly when these differences—a form of variation called "disorder"—can make a system more stable. The finding is broad: many physical, engineered, and biological systems become more robust when their components, or the interactions among them, are intentionally varied.
The work builds on hints that had been mounting for years. In a 2020 Nature Physics study, Motter's team showed that power generators could synchronize more effectively when they operated slightly differently from one another. And in a 2025 Nature Communications study, co-author Arthur Montanari found similar effects in models of flocking birds and drone swarms. The big question was whether these were isolated quirks or the sign of a deeper rule.
The answer, the new framework reveals, is that disorder can stabilize networks—but only when the node dynamics are rich enough. Simplified mathematical models, like the widely used Kuramoto model, describe each node with just one variable. Those models, which powered generations of network science, can strip away the very stabilizing effect researchers were trying to capture.
"Simplified models can inadvertently strip away the very stabilizing effect we want to capture," Motter said. "We now know which kinds of systems can benefit from differences, and why."
The practical promise is substantial. Instead of treating variation as a flaw, engineers could design differences deliberately into power grids, metamaterials, and other interconnected systems to make them stronger. The insights may also explain why disorder is so common in natural networks—from neural connections to entire ecosystems.
The team also built a public website where anyone can tweak parameters and watch network components interact, synchronize, and organize into patterns in real time. Northwestern postdoctoral researcher Arthur Montanari and graduate student Pietro Zanin are the study's co-first authors. The research was led from the Center for Network Dynamics at Northwestern's Weinberg College of Arts and Sciences in Evanston, Illinois.
It is a reminder that perfection is overrated—and that sometimes, our flaws are what hold everything together.