# n171 — the prediction program's first harvest: G1 killed at the
#         gates, prediction 30 born from the kill
# (2026-07-11. Generator G's head candidate, processed through the
#  six promotion gates. It died at G5 - and the death was fertile.)

## The candidate
G1: "charged superpositions carry an extra decoherence channel (the
staggered audit), scaling q^2, with its own knee."

## Gate G5 (registry consistency) - FAILED, twice over
1. STRUCTURAL: the monitor reads ENERGY-budget violations; charge is
   exactly conserved (topologically, n158) and therefore offers no
   violations to watch. A channel needs something to see. There is
   no staggered audit - the EM "sector" of the monitor was always
   the energy channel wearing the staggered statics.
2. REDUCTIO: had the channel existed at the monitor's own zeta-Q
   strength, the bridge transfer (G -> e^2/4pi-eps0-m_e^2, a 4.2e42
   enhancement) gives Gamma ~ 4e16 s^-1 for a benchmark electron
   biprism - coherence time 2e-17 s. Electron interferometry has
   existed since 1955. The candidate would have falsified the
   framework seventy years before it was written.

## The harvest — PREDICTION 30: the audit is charge-blind
Charged and neutral superpositions decohere IDENTICALLY at equal
energy modulation: no q^2 excess, at any precision, ever. Coulomb
energy modulations enter only through the one energy channel at
gravitational normalization - which is exactly why charged-particle
interferometry is possible at all.
KILL: any confirmed charge-dependent vacuum decoherence at fixed
dE/dt. DISCRIMINATES against charged-collapse (CSL-charged)
variants. TEST: comparative ion-vs-atom interferometry at matched
modulation - cheap, existing technology.
Gates passed: mechanism (the monitor's definition + exact charge
conservation), shape (exact equality), kill (named), prior art
(charge-dependent collapse models exist and are the discriminated
class), registry consistency (it IS the registry's own structure),
class [A-structural].

## The program's first lesson
The generator produced a candidate; the gates executed it; the
execution produced a better prediction than the candidate - a null
with a seventy-year retrodictive pass already in hand. The machine
works exactly as designed: procedure, gates, harvest.