# n170 — THE PREDICTION PROGRAM: seven generators for new falsifiables
# (2026-07-11. The registry holds 29; this is the machine for finding
#  the next ones. Each generator = a procedure, not a brainstorm.)

## PROMOTION GATES (what a candidate needs before entering the registry)
G1 mechanism (which organ produces it) · G2 shape or number ·
G3 kill condition · G4 prior-art check (the n167 discipline) ·
G5 consistency sweep against the standing 29 · G6 class label.

---------------------------------------------------------------------
GENERATOR A — THE SECOND AUDIT CHANNEL (momentum spectroscopy)
The momentum budgets (n164-166) are new organs; they must have their
own violation channel. Procedure: enumerate lab observables that
modulate MOMENTUM FLOW at fixed energy flow, and vice versa.
CANDIDATES:
A1. Fixed-power discrimination: internal-state modulation (energy
    channel) vs center-of-mass shaking (momentum channel) at equal
    dE/dt produce decoherence in a FIXED dimensionless ratio set by
    the constraint structure - a new lab number, no free parameter.
A2. Co-location: both channels share ONE knee frequency (one
    monitor). Two knees at different frequencies = the one-monitor
    axiom dies. A clean second-generation Auditometer test.

GENERATOR B — THE DARK-STATE CENSUS
Every audit-invisible configuration is a zero-decoherence null.
Procedure: enumerate steady constraint-satisfying configurations not
yet registered.
CANDIDATES:
B1. Persistent supercurrents are stationary and divergence-free:
    audit-dark. NO vacuum-origin flux-noise floor in SQUIDs, at any
    future precision. Kill: a confirmed irreducible vacuum flux noise.
B2. Inertial (uniform) motion is exactly dark: decoherence couples
    to modulation only, never to velocity. Discriminates against
    velocity-coupled noise models. Kill: confirmed v-dependent
    vacuum decoherence at constant v.

GENERATOR C — SECTOR TRANSFER
Every mechanism proven in one sector implies an analog in the other
(trace/shear <-> longitudinal/transverse; thermostat; Alice).
CANDIDATE:
C1. Alice-defect interferometry: matter encircling an odd-cycle
    relic returns C-conjugated - a matter/antimatter conversion
    holonomy. An exotic-search signature for topological relics
    [C]: no flux prediction yet, but a unique smoking gun.

GENERATOR D — EXECUTIONER INVERSION (downstream of forced organs)
Each data-forced structure must re-audit the standing registry.
FIRST RUN, DONE TODAY: does prediction 15's dark rotation survive
the four-row audit? YES - stationary rotation in mechanical
equilibrium satisfies FULL momentum conservation (internal stresses
balance centripetal flow; total div T = 0 in steady state), so the
momentum rows see nothing. THE ROTATION THEOREM IS UPGRADED: dark
under both audits. (Had it failed, molecular rotors would have
gained a computable decoherence rate - the generator can revise as
well as create.)

GENERATOR E — ANCHOR-BAND CROSSINGS
Everything pinned to hbar-gamma = 3.7 keV (knee, m_DM band) crosses
somewhere. Procedure: enumerate keV-band coincidences and ask which
are structural. (Low yield expected; F10 discipline mandatory.)

GENERATOR F — RELIC SPECIES
The crystallization epoch made more than windings.
CANDIDATE:
F1. Alice defects (odd cycles) are REQUIRED relics: charge
    quantization needs at least one in the causal past (n158). A
    MINIMUM-abundance statement - the first lower bound the
    framework owes on any relic. Kill: a proof of exactly-zero
    Alice density with charge still quantized.

GENERATOR G — THE TWO-CHANNEL FAMILY
The zeta-Q law relates gravity's channels; the EM sector has its own
pair (Coulomb vs radiation; staggered audit vs photon transparency).
CANDIDATE (the program's sharpest new item):
G1. CHARGED SUPERPOSITIONS HAVE AN EXTRA DECOHERENCE CHANNEL: a
    charged pattern is a staggered budget - WATCHED by the EM-sector
    audit - while a neutral one is not. Prediction shape: excess
    motion-gated decoherence for charged matter waves, scaling q^2,
    with its own knee; neutral-vs-charged interferometry (atoms vs
    ions/electrons) is the discriminating pair. Known electron/ion
    interferometry already bounds it: the FIRST TASK is to check
    whether existing data kills it on arrival (the n110 reflex).

---------------------------------------------------------------------
PRIORITY QUEUE
1. G1 (charged-vs-neutral) - confrontable with existing data NOW.
2. A1/A2 (the second channel ratio + co-location) - needs the
   momentum-channel rate computation; feeds the Auditometer design.
3. B1/B2 (SQUID + inertial nulls) - cheap, registrable after a
   prior-art pass.
4. F1 (minimum Alice abundance) - needs the surgery-rate estimate.
5. C1, E - parked until their sectors mature.
Every candidate passes gates G1-G6 or stays a candidate.