Perfection is overrated—at least when it comes to complex systems like the power grid, food webs and advanced materials. For decades, scientists generally assumed that networks function most reliably when their individual components are as similar as possible. But real-world networks are rarely uniform.

Generators in a power grid, neurons in a brain, animals in a food web and components in a material all differ in ways that scientists traditionally treated as imperfections.

Now, Northwestern University physicists are overturning that long-held assumption.