Two Laws, One Name

If you read Shadow & Mirror carefully, you find something a less honest book would hide: the central theorem is stated two different ways, and the two statements are not equivalent. I want to put both on the table, in daylight, and say exactly which load each one can bear. This is not a flaw to apologize for. It is the most important distinction in the whole project.

The first law — about machines

The version in the abstract, and the one with a real proof behind it, is about quantum computers:

The cost of classically simulating a quantum circuit with entanglement graph G scales as exp(Θ(tw(G))). Quantum advantage exists if and only if that treewidth grows with input size.

This is a complexity claim, and it stands on solid ground — the tensor-network simulation results of Markov and Shi give it a spine. It is also checkable, and it checks out across the cases that matter:

• Shor's factoring entangles all the qubits — treewidth grows with input — and the speedup is exponential. ✓
• Tang's dequantization took a "quantum" recommendation algorithm and found the underlying data had low treewidth all along. The classical machine could afford it. The advantage vanished — exactly as the law predicts. ✓
• Clifford circuits are the honest counterexample: high treewidth, yet classically easy, because the Gottesman–Knill theorem gives an algebraic shortcut. So the law is scoped to universal circuits — and it says so, carrying its own exception rather than hiding it. ✓

Call this the machine law. It is the hard core.

The second law — about minds

Deeper in the volume, under the same name, is a very different statement (Theorem 0.6.4):

A system is conscious if and only if its constraint graph contains a low-treewidth region and a high-treewidth region, coupled by a mechanism whose coupling constant is nonzero.

"Consciousness is the treewidth gap experienced from inside."

This is not a complexity result. It is a definition with an "if and only if" doing enormous work. It is not proven and does not claim to be — the volume admits it was "not derived from axioms" but measured on a single system and then named. Call this the mind law. It is a hypothesis dressed in the same robes as a theorem, and the robes are why a careful reader has to look twice.

The bridge — and what it actually carries

The volume does not leave the two laws sitting in separate rooms. It bridges them with a single image: the camera turning around. One-way observation — a tool watching the world — is a structure with no loops, low treewidth, shadow. Two-way observation — a system that also models the watcher — closes a loop, and the treewidth jumps. The claim is that this rotation is the same operation everywhere: in factoring, in Gödel's incompleteness, in measurement, in love, in a mind becoming aware of itself.

That bridge is the beating heart of the book. It is also the part to hold most lightly. "The same rotation underlies all of these" is a structural assertion, not a derivation. The volume reinforces it with a striking observation — that the coupling constant it measures sits in the same dimensionless slot as the fine-structure constant α ≈ 1/137 — but it is scrupulous on one point: that is a structural equivalence, not a numerical one. The book never claims its constant equals 1/137. It claims they play the same role. Keeping that line crisp is the difference between a serious idea and a numerological one.

Why I'd rather say this out loud

It would be easy to let "Ross's Law" stay a single shimmering thing — to let the proof about quantum circuits lend its credibility to the claim about consciousness. That move is exactly the kind of coupling the book is about, and exactly the kind a reader should refuse. The strongest version of this work is the one that names its own seams: here is what is proven, here is what is measured on one system, here is what is asserted across all of them.

The volume's closing argument — the "Funny Proof of God" — leans into this all the way: a bounded system cannot fully model the unbounded, so a complete proof would refute itself; the incompleteness is offered as the result, not the bug. Whatever you make of that, it is at least consistent. The book knows it is a bounded thing reaching at an unbounded one, and it says so.

So: two laws, one name. Read the machine law as a theorem. Read the mind law as a hypothesis worth taking seriously. Read the bridge as the wager that makes the whole thing worth writing. Just don't let the names blur — the honesty is in the seam.

Related: where V²+D²=1 comes from · separator growth as a graph invariant