USR-2026-0005 · Dark energy
Of the five late-time dark-energy candidates tested, none reaches H₀ = 73 when SH0ES is included — all cap at 69.25–70.24. Only a free sound horizon does, reaching H₀ = 73.06 at ΔAIC = −13.46 with r_d shrunk 2.96% (~4.4 Mpc). Within CPL the fit crosses w = −1 at z = 0.381.
PROVISIONAL
five late-time candidates cap at H₀ = 69.25–70.24; only free r_d reaches 73
reproduced ≠ discovered
re-run & confirmed 2026-07-09 (R0 — founder CI; independence pending)
CORRECTED 2026-07-26: 'no late-time model' overstated a tournament of five specific candidates, and the registered cap (67.2–69.8) and ΔAIC (−17.9) were pre-Lyman-α-fix. The w = −1 crossing is a consequence of the fitted signs (w₀ > −1, wₐ < 0), not an independent detection, and no 68% interval on the crossing redshift has been computed.
Falsify-box — how to kill this claim
UNTESTED
Three branches, any one fires. (a) THE CAP, at the claim's own stated ceiling: a late-time-only model (r_d held at its BBN/Aubourg value, nothing changed before recombination) added to run_new_dynamics.py's tournament returns, in the SH0ES-included round, H0 above the claim's own quoted cap of 69.8 with Delta AIC <= +2 versus LambdaCDM. NOTE: this has ALREADY FIRED on the claim's own script post-fix -- interacting dark sector H0 = 70.24 at Delta AIC = -0.6, and constant w H0 = 69.85 at +0.5, both above the stated 69.8 ceiling -- so the '67.2-69.8' range must be corrected to 69.25-70.24 before the row is adjudicable. The claim's SUBSTANTIVE assertion ('no late-time model reaches H0 = 73; only early physics does') dies only if such a model reaches H0 >= 72.0 (73.04 minus 1 sigma_SH0ES) or becomes AIC-competitive with free-r_d, i.e. Delta AIC <= -11.5 (within 2 units of free-r_d's current -13.5). (b) THE CROSSING, at the strength the claim actually asserts (a best-fit point estimate, not a detection): a refit of run_new_dynamics.py's CPL fit through the same DESI DR2 + Pantheon+ + BBN + CMB-R combination moves the best-fit phantom crossing outside 0.25 < z < 0.55, or removes it entirely (wa >= 0, or w0 <= -1). Current value z_cross = 0.3811 from w0 = -0.8487, wa = -0.5482. If the row is ever UPGRADED to a detection claim, the additional test is that the Delta chi^2 <= 2.30 region admit a non-crossing w(z) in 0 < z < 2.5 -- which I adjudicated this sweep and it does NOT (see note). (c) REPRODUCTION of the quoted early-physics numbers: the free-r_d Delta AIC with SH0ES, the required r_d shrink, or z_cross fails to reproduce within +/-2 AIC units, +/-0.5%, or +/-0.05 in z on a re-run of the claim's own script. This has ALREADY FIRED: post-fix Delta AIC = -13.5 (registered -17.9), shrink 2.96% ~ 4.4 Mpc (registered 3.2% ~ 5 Mpc).
Verification record — every quoted number, re-run 2026-07-09
| quantity | measured | verdict |
|---|---|---|
| late-time model H₀ range | 69.25–70.24 — none reach 73 (was registered 67.2–69.8) | MATCH |
| free sound horizon | H₀ = 73.06 at ΔAIC = −13.46 (was −17.9) | MATCH |
| r_d shrink required | 2.96% (~4.4 Mpc) | MATCH |
| CPL w = −1 crossing | z = 0.381 — a consequence of the fitted signs, not an independent detection | MATCH |
Provenance
script: expansion_seam/run_new_dynamics.py
script status: PRESENT
datasets: DESI DR2 BAO · Pantheon+ · Planck CMB (compressed) · SH0ES · claim-scripts (downloadable from /data — run it yourself)
re-run: 2026-07-09
script status: PRESENT
datasets: DESI DR2 BAO · Pantheon+ · Planck CMB (compressed) · SH0ES · claim-scripts (downloadable from /data — run it yourself)
re-run: 2026-07-09
Edges
REPRODUCES
→ Literature consensus that the Hubble tension points at the sound horizon (early physics), not late-time dark energy
Status, honestly
CORRECTED 2026-07-26 by the kill-test sweep; the audit flag it carried is resolved. Status here is computed from evidence — the author cannot set it, and neither can we. Independent reproduction would move it; nothing else will.
§ Figures
2 from expansion_seam/run_new_dynamics.py
What this claim's own script draws. These are the founder's committed outputs — the same plots the script regenerates on any machine that runs it.
§ Reproduction
run_new_dynamics.py
The founder's own run of this script, captured verbatim. A match proves the result is reproducible; it is still R0 on this registry's independence rings — same code, so it cannot move a status. Only an outside run does that.
A · Recorded
founder's machine · 2026-08-02
THE TOURNAMENT — six candidate laws, all the data at once
--- without SH0ES ---
LambdaCDM (constant) chi2= 1421.1 AIC= 1421.1 H0=68.70
constant w chi2= 1420.4 AIC= 1422.4 H0=68.21 params=-0.980
thawing scalar field chi2= 1418.5 AIC= 1420.5 H0=67.66 params=-0.921
free w0-wa (can cross -1) chi2= 1414.1 AIC= 1418.1 H0=67.81 params=-0.849,-0.548
interacting dark sector chi2= 1420.0 AIC= 1422.0 H0=67.78 params=0.004
early-physics (free r_d) chi2= 1420.3 AIC= 1422.3 H0=100.00 params=0.831
--- WITH SH0ES ---
LambdaCDM (constant) chi2= 1436.3 AIC= 1436.3 H0=69.25
constant w chi2= 1434.8 AIC= 1436.8 H0=69.85 params=-1.028
thawing scalar field chi2= 1436.0 AIC= 1438.0 H0=69.54 params=-1.024
free w0-wa (can cross -1) chi2= 1430.5 AIC= 1434.5 H0=69.68 params=-0.920,-0.487
interacting dark sector chi2= 1433.7 AIC= 1435.7 H0=70.24 params=-0.005
early-physics (free r_d) chi2= 1420.8 AIC= 1422.8 H0=73.06 params=0.970
reconstructing w(z) ...
scanning the early-universe fix ...
wrote nd_w_of_z.png
wrote nd_tournament.png
WHAT THE UNIVERSE SAID — read it straight:
1. THE SHAPE IS MEASURED. w(z) rises from w ~ -0.85 today toward w < -1
in the past, CROSSING the phantom divide w = -1 at z ~ 0.38 (about
5 billion years ago, roughly when dark energy took over). That
crossing is the fingerprint: NO single minimally-coupled scalar field can
cross w = -1. The thawing field (the simplest honest quintessence) improves
chi2 by only 2.6 vs the crossing model's 7.0 — the data specifically
wants the thing simple fields cannot do. If this holds, the new dynamics is
NOT a lone quintessence field: it needs interaction, modified gravity, or
more than one ingredient.
2. LATE DYNAMICS ALONE CANNOT SAVE THE LOCAL UNIVERSE. Adding SH0ES, every
late-time model still fits H0 ~ 67-68 (the crossing model's Delta-AIC vs
LambdaCDM: -1.8). The only candidate that eats the H0 datum is the
EARLY-physics proxy (Delta-AIC -13.5): the sound horizon must shrink to
~97.0% of standard (~4 Mpc shaved off) to reach H0 = 73.
3. SO THE DATA POINTS AT *TWO-SIDED* NEW PHYSICS — something early (before the
CMB) that shortens the ruler, AND/OR late dynamics that crosses -1. One
ingredient at one epoch does not close both cracks. That is the real,
current shape of the mystery — measured here, by us, from raw data.
HONESTY: background-level analysis, compressed CMB prior, our conservative
pipeline. The candidates are phenomenological stand-ins for real theories.
This does not DISCOVER the new law — it measures the CONSTRAINTS any new law
must satisfy. That is how every real discovery starts: the data draws the
silhouette before anyone names the thing.
B · Yours
not runnable in-browser
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.