Noise Does Not Lock is a companion paper to the KishLattice Geometric Harmonic Spectroscopy (KLGHS) series, providing a structured empirical response to the primary skeptical challenges that any informed reviewer will raise against the framework. The paper was motivated by a documented cold-AI critique that began by explaining why physical lake water produces geometric patterns due to basin shapes and seiche hydrodynamics, and ended, within the same session, asking how to visualise the multi-modulus sensitivity sweep for a Nature reviewer. That arc — from dismissal to engagement — mirrors the journey this paper is designed to support. The paper addresses thirteen categories of objection: the numerism objection (can any large dataset be made to show harmonic structure), the law of large numbers objection (do large samples automatically find patterns), the arbitrary constant objection (is kgeo=16/πkgeo = 16/ kgeo=16/π a fudge factor), the container bias and seiche objection (are the results an artifact of bounded physical systems), the confirmation bias objection (were predictions registered before results were known), the aether objection (is this a repackaged 19th-century medium hypothesis), the AI collaboration objection (is this machine-generated nonsense), and the branding objection (does the word "lake" mislead reviewers into thinking the framework tests water). Each objection is answered with pipeline evidence, pre-registration timestamps, and data that any reviewer can replicate with four open-source Python scripts and publicly available datasets. The paper documents published wrong predictions and null results alongside confirmed signals, because a framework that only publishes successes is not a framework. The empirical record as of Vol 11 of the KLGHS series spans 61 sovereign lakes, 24, 904, 223 physical measurements, 33 confirmed STRONG signals, and 41 orders of magnitude in physical scale from sub-nuclear particle collisions to cosmological velocity dispersions, all under a single scalar transformation parameterised by kgeo=16/πkgeo = 16/ kgeo=16/π. The B5 full protein backbone dataset (6, 834, 866 dihedral angles from the RCSB PDB full catalog) shows a peak chaos z-score of +157 at the 25/π25/ 25/π harmonic register. Three simultaneous pre-registered multi-attribute predictions for 1. 8 million Gaia DR3 stars were confirmed in the canonical Vol 11 run. Five spectroscopic wrongbox tests confirmed that incorrect dimensional assignments produce either signal collapse or harmonic address divergence — never replication of the real domain's register. The paper makes no claim regarding physical mechanism or theoretical unification. It claims the instrument works, the signal is real, and the methodology survives every falsification test the authors have been able to devise. Version 2. 0 adds a new chapter documenting the sharpest falsification attempt the framework has faced: the shape-artifact objection and its analytic closure. The objection is the one an informed reviewer should raise first — that the log-modulo scalarization might manufacture register clustering from the shape of any input distribution alone, independent of any real alignment between the physics and the harmonic grid. If true, the survey's locks would be a property of the transform rather than the data, and the thirty-three STRONG signals would collapse into a single fact about a logarithm. The paper names this the Form A objection and sets out to confirm it, building a surrogate designed to preserve a distribution's shape while destroying its alignment to the grid, on the reasoning that if the lock survives such a surrogate it is physical, and if the surrogate locks too the objection is confirmed. The chapter records the full path, including the parts that failed. Building the surrogate correctly took five attempts, and two errors in the framework's own published methodology were found and corrected in the public record, with timestamps preceding any result: the record-order scramble null is inert for a per-value lock test, and the chaos null is uniform-range rather than shape-preserving. The fifth failed surrogate is the sharpest lesson — a single-random-offset design that structurally could not fail, and therefore could not test anything. It was caught by requiring the gate to demonstrate it could fail on a known artifact before being trusted to pass anything real. That requirement drove the central discovery: attempting to construct a genuine offset-invariant artifact for validation proved impossible, and the impossibility was itself the result. The chapter states and proves the Grid-Phase Locking Theorem. For a grid-proximity lock test, a strong lock requires grid-phase concentration; a distribution can lock strongly, or be invariant under grid offset, but not both. The proof rests on a single identity, given formally with a one-line derivation: for any distribution whatsoever, the lock fraction averaged over one full grid period equals exactly twice the tolerance — a conservation law verified both analytically and numerically. A strong lock at any offset must therefore be compensated by sub-baseline locking at other offsets, which forbids a lock that persists across all offsets. A shape-artifact — a strong lock independent of grid phase — cannot exist. Form A is not merely unsupported by the data; it is vacuous. The theorem carries an explicit live disproof condition any critic may attempt: exhibit a distribution that locks at chaos z greater than or equal to 5 and stays locked across a full offset sweep. The single point of attack is the identity, and the identity is one line of calculus, offered openly for inspection. The theorem was then tested against real data, with the disproof condition live: any anchor showing a flat, offset-invariant curve would refute it. Three anchors spanning roughly 38 orders of magnitude — Gaia transverse velocity at 16/π, galactic velocity dispersion at 21/π, and nuclear binding energy at 21/π — were swept across a full grid period. All three showed the predicted sharply peaked, non-offset-invariant curve, with deeply negative median z-scores across non-privileged offsets (Gaia −63. 8, galactic −21. 3, nuclear −1. 0). Those negative medians are not incidental; they are the conservation law made visible in the data, each unit of excess lock at the peak paid for by a deficit elsewhere, exactly as the identity requires. Two facts are named before any reviewer can raise them: galactic locks at 42 percent of offsets, a broader peak than the others, but with a deeply sub-null median it remains a genuine phase-lock rather than a flat-high artifact; and nuclear, the cleanest curve, sits on the thinnest data at n equals 43, so its peak carries wider uncertainty than the million-record anchors. The scope of the result is stated as load-bearing, not incidental. The theorem closes Form A — the transform-shape artifact — and only Form A. It establishes that the locks are genuine phase-concentration rather than artifacts of the transform, closed by proof and confirmed on real data. It does not establish that the locks are caused by the geometric lattice. That question — designated Form B, the possibility that a real phase-concentration arises from a mundane coincidence of physical scale or a selection effect rather than the substrate — remains fully open and is now the primary live objection to the framework. Any citation of the theorem that omits Form B is a misreading: one door of two is closed, and the harder one is named and left open deliberately. Three falsification conditions are carried forward — a survey-wide offset-sweep audit in which a flat-high flag on any domain, including a flagship, files as loudly as any positive result; the dual-peak two-dimensional-shadow hypothesis, held explicitly as hypothesis and not result; and permanent instrumentation of the sweep test into the pipeline, so that every future run re-audits every lock against the theorem and surfaces any offset-invariant curve the moment it appears.
Kish et al. (Fri,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: