# n47 — pre-registration: local clocks in the drainage well (written before any code)

**Weird effect targeted:** gravitational time dilation. GR says clocks
deeper in the potential run slower, by the same fraction whatever the
clock is made of. That universality is what makes gravity a METRIC and
not a force. The drainage mechanism has a candidate explanation: near a
mass the vacuum's links are depleted; every local process runs on those
links; so every local process slows — universally, because it is the
substrate that slowed, not the clock.

**Prediction under test:** in the n46 vacuum (information-metric
drainage), the fractional frequency shift of a local clock at site r is
  delta-omega/omega (r) = kappa * phi(r),
with ONE constant kappa for clocks of different construction, where
phi is the theory's own potential: the staggered-gauge-transformed
Lagrange-multiplier field of the sum rule (lambda = -W^+ c — "the
depression of the water table"), unit-cell averaged.

## Model
Ring N=200, mu^2=1e-6, information-metric drainage (n46 machinery),
opposite-parity consumer pair at separation d=99 (max clean), F=0.3.
Lattice-scale staggering is declared sub-metric structure: all gated
quantities are UNIT-CELL AVERAGES (adjacent site/bond pairs). The
metric, if it emerges, is the long-wavelength envelope.

## Three clocks (declared constructions)
- C1 site clock: local zero-point frequency omega1(r) = Omega_rr,
  Omega = sqrt(K(dk)).
- C2 bond pendulum: omega2(e) = sqrt(2 k_e), the two-site relative-mode
  frequency of link e.
- C3 correlation clock: omega3(r) = 1/(Omega^{-1})_rr — the inverse
  local position-fluctuation scale (a completely different functional
  of the spectrum than C1).
Fractional shifts are computed against the undrained vacuum, then
unit-cell averaged.

## Gates
- G1 SIGN: cell-averaged shifts NEGATIVE (clocks slow) at the cells
  adjacent to each consumer, for all three clocks.
- G2 PROPORTIONALITY: affine fit of each clock's cell-averaged shift
  against cell-averaged phi over all cells farther than 2 cells from a
  consumer: fracRMS < 5% per clock.
- G3 UNIVERSALITY (the metric gate): the three fitted slopes kappa_i
  agree: max_i |kappa_i/kappa_mean - 1| < 10% -> PASS ("time dilation,
  universal at the toy's resolution"). Spread > 50% -> the effect is
  clock-dependent, i.e. a force not a metric: the time-dilation claim
  DIES (recorded as such). Between 10% and 50%: partial, reported
  without spin, universality unproven.
- G4 LINEAR RESPONSE: kappa doubles F=0.15 -> 0.3 invariant
  (|kappa(0.3)/kappa(0.15) - 1| < 2%): the shift is set by the field,
  not by higher-order back-reaction, at these amplitudes.
- Report (ungated): the redshift table — frequency ratio of two C1
  clocks at different depths vs 1 + kappa*delta-phi (Pound–Rebka analog).

## Declared limitations
- Static clocks only: no dynamics, no cone, no moving clocks (special-
  relativistic dilation untested — that is the n34 wall, not this gate).
- 1D: no bending, no PPN gamma; that is n48 territory (2D wave
  propagation through the well) and is NOT claimed here.
- Unit-cell averaging is declared, not derived: the claim is about the
  envelope metric, and the staggered microstructure is reported openly.
- If G3 lands in the partial band, the honest statement is "the substrate
  slows processes near mass but not yet universally" — no metric language
  on the site until a pre-registered pass.

## RESULTS (same day; script n47_clocks.py, results n47_results.npy)
- G1 SIGN: PASS, all three clocks, both F values — every clock slows
  near the mass. The qualitative effect is real and universal in sign.
- G2/G3/G4: FAIL as registered. Diagnosis (visible in the redshift
  table): the registered potential — the signed water table lambda — is
  ANTISYMMETRIC between the two opposite-parity consumers, but the
  measured clock shifts are SYMMETRIC. Clocks are parity-blind; the
  registered gates were anchored to the wrong field. kappa ~ 1e-14
  (no correlation), so G3/G4 numbers are void, recorded as fired.
- POST-HOC (flagged, not gated): the clocks track the SQUARED drainage
  envelope. C2 (bond pendulum) is exactly -<dk^2>/8: fit fracRMS 0.000%,
  F-exponent 2.008, coefficient -0.127 ~ -1/8 (analytic: cell-average of
  sqrt(1+dk) over the +-E staggering). C1/C3 (spectral clocks) carry an
  ADDITIONAL non-local channel: F-exponent 1.64-1.67, residuals 4.8% and
  21% against pure |dk|^2 — a mixed linear + quadratic response.
- STRUCTURAL FINDING, stated plainly: in this toy the force potential
  and the clock potential are DIFFERENT FACES of one field — the force
  rides the SIGNED staggered component (parity-charged), the clocks ride
  its SQUARE (parity-blind). In GR both are one metric and redshift is
  LINEAR in the source mass; here the dominant clock response is
  QUADRATIC. If no isolable linear channel survives n47b, the mechanism
  does not reproduce GR-like redshift — a lethal constraint, named now.
- n47b (owed): two-channel decomposition registered in advance — shift =
  a * (linear envelope functional) + b * <dk^2>, universality gated on
  the LINEAR coefficients across clock constructions; parity-restricted
  clocks (sublattice-built) to probe whether matter-like clocks couple
  linearly with parity sign, which would be a sharp anti-GR signature
  (opposite-parity clocks speeding up) — either outcome is decisive.