The substrate-noise window is OPEN and experimentally live: GRW's collapse rate λ = 10⁻¹⁶ s⁻¹ at r_c = 10⁻⁷ m survives every published mass-proportional CSL bound — the tightest (Majorana Ge X-ray, coherent-nuclear channel) leaves ×49 headroom — while Adler's λ ≈ 10⁻⁸·⁵ is excluded by ×7 to ×6×10⁵ and GRW's original 1986 non-mass-proportional coupling is dead. 36–57% of the historically proposed GRW-to-Adler parameter box survives. MAQRO-class space interferometry reaches 1900× past GRW's point on a ~10–15 year horizon.

0e-10   ALIVE x4.4e+06      DEAD x7.2e+00
KDTL interferometry (compiled)      [A4]       1.00e-07   ALIVE x1.0e+09     alive x3.2e+01
X-ray, NON-mass-proportional CSL    [A1]       5.15e-20    DEAD x1.9e+03      DEAD x6.1e+10

GRW benchmark (lambda = 1e-16, r_c = 1e-7 m):
  mass-proportional CSL: ALIVE under EVERY published anchor; the
  strongest (coherent-nuclear X-ray) sits a factor 49 above it.
  non-mass-proportional (GRW's ORIGINAL 1986 coupling): DEAD -- 
  excluded by a factor 1942 by [A1].
Adler benchmark (lambda = 1e-8.5): DEAD by 7.2x (cantilever, weakest applicable) to 6.4e+05x (X-ray aggressive).
Adler's FULL band 1e-8+-2 at r_c = 1e-7 m: even its low edge 1e-10 is excluded
  by the conservative X-ray anchor (575x above it); at his r_c = 1e-6 m the low edge 1e-08
  is likewise 575x above the conservative X-ray bound 1.74e-11. The Adler
  proposal is dead at both his quoted r_c values, by orders.

============================================================================
SECTION 6 -- the (lambda, r_c) plane: curves and surviving fractions
============================================================================
Stated rectangle: r_c in [1e-9, 1e-3] m (6 dec), lambda in [1e-20, 1e-6] s^-1 (14 dec).
Macroscopicity floor (OUR convention, stated): a 1 um-diameter,
rho = 2000 kg/m^3 grain superposed over d = 1 um must decohere in
t <= 10 ms; lambda_floor(r_c) = 1/(t * Gamma1_grain). Literature
floors vary by orders with the chosen macroscopicity criterion.

Full-rectangle log-area surviving:   48.8% (conservative anchors) /  39.7% (aggressive)
MOTIVATED window (above floor)   :   41.5% (conservative) /  32.4% (aggressive)
At r_c = 1e-7 m the surviving lambda band spans 7.2 dec (cons.) / 5.7 dec (aggr.)
  within the rectangle (floor 4.5e-21 lies below its 1e-20 bottom there);
  bounds 1.7e-13 / 4.9e-15 s^-1; GRW 1e-16 is INSIDE both bands.
Motivated window fully CLOSED at: r_c < 1e-8.6 m and r_c > 1e-3.1 m (cons.) ; r_c < 1e-8.3 m and r_c > 1e-3.1 m (aggr.)

Historically-proposed box (GRW-to-Adler: lambda 1e-16..1e-8.5, r_c 1e-7..1e-6 m):
  surviving log-area fraction = 57% (conservative) / 36% (aggressive).
  The Adler end is gone; the GRW end stands untouched.

Non-mass-proportional CSL at r_c = 1e-7 m: floor 4.5e-21 vs bound 5.15e-20 -> 1.06 decades left above our
  floor: NEARLY CLOSED, and its GRW point is dead by 1942x. The surviving
  theory space is the mass-proportional family.

============================================================================
SECTION 7 -- forward look: the closing experiments
============================================================================
[F1] MAQRO-class space interferometry (arXiv:2202.01535): SiO2
     sphere, d = 100 nm, t = 100 s. Computed reach at r_c = 1e-7 m:
       1e9  amu: lambda ~ 5.3e-20 s^-1  (1905x BELOW the GRW point)
       1e10 amu: lambda ~ 8.7e-22 s^-1
     Post-MAQRO(1e9 amu) motivated window: 14.0% survives (from 41.5%);
     at r_c = 1e-7 m the in-rectangle band shrinks to 0.72 dec -- the reach sits a
     factor ~11.7 above the unclipped floor: the canonical column is
     essentially closed. A band at r_c ~ 1e-6..1e-3 m
     survives even MAQRO under these assumptions (heavier masses /
     longer times / LPF-successor force sensors needed there).
     Timescale per whitepaper: ground+microgravity campaign this
     decade, implementation 'within the subsequent decade' (~2030s).

[F2] LEGEND-1000-scale Ge (arXiv:2107.11462; 1e4 kg yr, x10 lower
     background ASSUMED; sqrt(B/exposure) scaling -> x52 tighter):
       m-p electrons-only channel: lambda ~ 3.4e-15 s^-1 at r_c = 1e-7 m
       m-p +coherent-nuclear     : lambda ~ 9.6e-17 s^-1 -- REACHES the GRW point
     on a ~2040 horizon (late-2020s construction + ~10 yr run).
     This projection is OURS, not the collaboration's; spectral-fit
     systematics and 2vbb backgrounds could limit it.

============================================================================
SECTION 8 -- SEAM VERDICT (with its caveat, stated plainly)
============================================================================
STATUS: OPEN AND EXPERIMENTALLY LIVE.

  * GRW's original rate lambda = 1e-16 s^-1 at r_c = 1e-7 m is
    ALIVE in the standard mass-proportional CSL: the strongest
    published bound (Majorana, coherent-nuclear channel) sits a
    factor 49 above it, and 5.7-7.2 decades of lambda survive at
    the canonical r_c. The substrate-noise question is not closed.
  * The historically-proposed (GRW-to-Adler) box retains 36-57% of its
    log-area; the broader motivated window retains 32-42%. Adler's
    parameter island is dead by orders; the original
    non-mass-proportional coupling is effectively closed.
  * Named closers: MAQRO-class space interferometry (2030s) closes
    the canonical column to the macroscopicity floor; LEGEND-scale
    Ge (~2040, our scaling) reaches the GRW point independently.
    This seam gets ANSWERED within ~10-15 years either way.

THE CAVEAT (constitutive, not a footnote): a detected universal
collapse noise would be a noise-floor-shaped fact about reality --
CONSISTENT with a computational substrate but NOT proof of one.
Collapse models are standalone fundamental physics; an uncomputed
universe can house them just as comfortably. Detection would open
the substrate question, not settle it. Conversely, a fully closed
window would kill the noise-floor tell entirely.

DISCRIMINATOR TEST (the project's knife, applied): would a computed
and an uncomputed universe answer differently? Not cleanly -- both
outcomes are available to both metaphysics. This seam is therefore
NOT a discriminator; it is a one-way falsifiable TELL: closure
removes it forever, detection makes it the most interesting number
in physics. Unlike the null frame-dependent seams (grid, clock,
dispersion, floats), this one is alive, funded, and scheduled.

============================================================================
HEADLINES
============================================================================
H1 anchors: [A1] arXiv:2202.01343 (X-ray Ge: 5.15e-20 nmp / 1.74e-13 mp-e / 4.94e-15 mp-nuc at 1e-7 m);
           [A2] arXiv:1606.03637 (LPF 2.96e-8; rot. update arXiv:2501.08971); [A3] arXiv:2002.09782 (4.4e-10);
           [A4] Fein 2019 + arXiv:2203.04231 (~1e-7); benchmarks GRW 1986, Adler quant-ph/0605072.
H2 GRW 1e-16 at r_c 1e-7: ALIVE under all mass-proportional anchors (margin 49x at tightest); DEAD only
   in the non-mass-proportional variant (1942x). Adler 1e-8.5: DEAD by 7x-6e+05x; his full
   band is excluded at both his r_c values.
H3 surviving window: GRW-Adler box 36-57%; motivated window 32-42%; full rectangle 40-49%.
H4 closers: MAQRO (2030s; reach 5e-20 at 1e-7 m, 1905x past GRW) + LEGEND-scale Ge (~2040; 1e-16).
H5 verdict: substrate-noise seam OPEN and experimentally live; a
   detection would be consistent-with-substrate, not proof; closure
   kills the tell. One of the few live falsifiable seams in the hunt.

This script needs data files, a heavy dependency,
or more time than a browser tab should take.
Download it from /data and run it locally.