Paper 14: could a lab see this?
Paper 14 is the framework’s lab protocol, and it is the paper most likely to be skimmed and most important to read slowly, because it contains no physics result at all. It contains the conditions under which a physics result would be believed, written before anyone touches a detector, and after the archive you have read, you know exactly why such a document has to exist.
The two protocols
The first protocol targets surface-code quantum processors. The hypothesis from the boundary work, that leakage between branches is patterned rather than thermal, has a laboratory shadow: if branch structure is real and the substrate analogy holds, logical error events on code patches that are not physically coupled should show correlations above what independent depolarising noise predicts, and the correlation should scale with code distance in the way the topological picture demands. The protocol: multiple patches, blinded analysis, the correlation function C of d computed with the analyst unaware of which patches are physically adjacent, and the pre-registered comparison against the independent-error model and against crosstalk models. The paper is explicit that the expected effect, if any, is small, and states the required power: with realistic error rates, the experiment needs of order millions of logical cycles to separate the hypotheses, which is within what current processors run anyway.
The second protocol targets matter-wave interferometry. The same patterned-leakage logic, at the level of phase noise: decoherence rates in large-molecule interferometers should, under the hypothesis, acquire a structured component beyond the standard environmental models, with a spectral shape inherited from the squeezed-state algebra rather than the flat or Lorentzian shapes of ordinary noise. The protocol specifies the noise-model comparison in advance: standard decoherence models, the framework’s structured addition, and the discrimination statistic, again blinded.
Why the blinding is the physics
The paper’s central content is procedural, and the justification is empirical: this field has a history of correlations that were found and then explained. Any analyst who knows which patches are adjacent will, however honestly, analyse differently. The blinding is not decoration; it is what makes the statistic’s distribution under the null hypothesis trustworthy. Alongside it: pre-registered thresholds, both hypotheses’ predictions computed before data, and a stated rule that a null result at adequate power kills the patterned-leakage hypothesis, not merely fails to support it.
That last clause is the paper’s sharpest feature, and the hardest to write. Most experimental proposals state what a positive result means. This one states what a negative result means, with the power calculation to back it, because the archive’s whole epistemology says that a hypothesis which cannot lose is not in the game. The paper also names what would count as the strongest negative: error correlations that match crosstalk models exactly, with no residual structure, at adequate power. That is the framework’s own neck, extended on purpose.
What the paper does not do
It does not import cosmological parameters into the lab analysis: no a zero, no branch counts, no numerology, the paper’s own note is that no Psi geographic correlation is imported. It does not predict the effect’s size beyond the scaling, because the scaling is what the topological argument supports and the size depends on the condensation pattern that Paper 13 leaves open. And it does not claim that a null result kills the whole framework: the computational results, the plateau theorems, the bounds, live independently of the leakage hypothesis. What a null kills is the specific, most exciting, most falsifiable claim: that the substrate’s structure shows up in a laboratory as correlated structure. That is the bet, and this paper is where it is priced.
The next paper asks what the substrate’s discreteness would shake, and over what spectrum: Paper 15, holographic noise.