The Simulation Is Spending Its Money In The Wrong Place

ITHACA, N.Y. — The simulation argument has a corollary that its proponents rarely follow, and following it is the only interesting thing this piece does.

If the universe is computed, the computation is bounded. Bounded computations economise. A simulation that economises would put its detail where it is observed and skimp where it is not.

So: where is the universe expensive?

The question we asked

We put one question to four physicists, in these words: ignoring whether the simulation argument is true, where in our current description of the universe does the arithmetic get most expensive to carry out?

Three answered the question. One refused it, at length, and we are printing his refusal because it is the strongest thing in the piece.

The three answers

All three named the same domain: the strong nuclear force at low energy.

Not cosmology, which is cheap — vast scales, simple rules, very few interacting bodies per unit volume. Not general relativity, which is geometry and is expensive but tractable. Not quantum mechanics in the ordinary sense.

Quantum chromodynamics at the scale where quarks bind into protons.

Prof. Nkechi Bergström-Vance put the numbers on it. A lattice QCD calculation of the proton mass — one number, for the commonest object in the visible universe — took the field about thirty years and required national-scale supercomputers, and it is still not the kind of calculation you can simply run.

"The universe does that computation, correctly, ten to the eighty times over, continuously, for free," she said. "If somebody is paying for this, that is where the bill is."

Dr. Corinne Yamamoto-Fairbairn made the same point from lattice theory and added the part that our correspondent has not stopped thinking about: nobody observes it.

"You cannot see a quark," she said. "Confinement means you never will. The most computationally brutal part of the description is the part that is, by construction, unobservable — and it is being carried out to full precision inside every proton in your hand."

What that does to the argument

The economising simulation should render the observed and approximate the unobserved. What we appear to have is the reverse: cheap where we look, ruinous where we cannot.

This does not refute anything. An argument this loose can absorb the result — perhaps the expense is the point, perhaps our notion of expense is parochial, perhaps the substrate finds QCD easy and orbital mechanics hard.

But it is a specific, checkable shape, and it points the other way from the intuition the argument is usually sold on.

The refusal

Prof. Sandor Alameddine declined to answer and asked for room to say why. He gets it.

The question contains its conclusion. "Computationally expensive" is not a property of physical law. It is a property of our particular way of approximating physical law with silicon, and it changes every time somebody invents a better algorithm.

Lattice QCD is expensive for us because we discretise a continuum onto a grid and fight the sign problem. That is a fact about our numerical methods in 2026. It is not a fact about protons. A proton is not doing a lattice calculation. A proton is being a proton.

Your piece will imply that the universe is spending resources somewhere. The universe is not spending anything. We are. Every sentence about the "cost" of a physical law is a sentence about the state of our own numerical analysis wearing a costume, and I would like your readers to know that at least one of the people you interviewed thinks the whole exercise is a category error.

This newspaper has printed that in full, before its own conclusion, without a rejoinder, because he asked and because he may be right.