There is no way to run an experiment on the universe. You cannot rewind it, change a single ingredient, and watch a different cosmos unfold to see what mattered. The universe is one run of one experiment, and we arrived in the middle of it. So to test how it came to look the way it does, physicists do the only thing left to them: they grow another one inside a computer.
The method is called an N-body simulation. You fill a virtual volume with particles — billions of them, each standing in for an immense clump of matter — and give the computer a single instruction: let gravity act, step by tiny step, for billions of simulated years. Every particle pulls on every other. From that one rule, structure grows.
And what grows is uncanny. Begin with the near-uniform haze of the early universe, let the simulation run, and the matter slowly gathers — into filaments, into sheets, into a vast cobweb of clusters strung around empty voids. It ends up looking remarkably like the real arrangement of galaxies that telescopes actually map across the sky. One landmark run, two decades ago, followed ten billion particles for exactly this purpose.
That resemblance is the experiment. If you build a model universe out of ordinary matter alone and it comes out looking wrong, the recipe is missing something. It was simulations like these, stubbornly refusing to match the real sky without it, that helped make the case for dark matter — an ingredient no one can see, inferred in part because the cosmos simply won’t come out right without it in the code.
This is a genuinely new way of knowing. For most of history, science rested on two pillars: theory, which proposes, and experiment, which tests. Simulation is a third — a way to test ideas about systems you can never set on a workbench, by building them out of rules and watching what they do.
It rests on a quiet act of faith in software: that if you get the rules right and keep the arithmetic honest, the thing you grow inside the machine will tell you something true about the thing you can never touch. That faith has paid off often enough that a great deal of modern physics now runs this way.
We’re drawn to that — the idea that getting the small rules exactly right is what lets something trustworthy emerge at scale. More about how we work →