This paper looks at a simple number that sits in supernova data and helps separate measurement problems from new physics. The number is an “intercept” a_B that appears in the basic relation between a supernova’s apparent brightness and its distance. In formulas the paper uses: m_B = 5 log10 d_L(z) − 5 a_B, and −5 a_B combines the supernova absolute brightness M_B and the Hubble constant H0. Because a_B can be built directly from observed brightness m_B and redshift z for any assumed late-time distance model, checking whether a_B is constant across samples and redshift ranges gives a diagnostic for whether tensions come from local measurement systematics or from new late-time cosmology.

The authors reconstruct a_B from current supernova compilations and from combinations of other probes. For the PantheonPlus sample they split the data into a local set (redshift z < 0.0233, 336 type Ia supernovae) and a Hubble‑flow set (0.0233 < z < 0.15, 490 supernovae). They also build an “inverse distance ladder” using Hubble‑flow supernovae (HFSN), two-dimensional baryon acoustic oscillation data (2DBAO), and cosmic chronometers (CC) to set the late‑time distance scale independently. The Hubble‑flow plus inverse‑distance‑ladder gives H0 = 68.5 ± 3.5 km/s/Mpc and M_B = −19.40 ± 0.11. But the directly reconstructed intercepts differ: a_IDL^B = −4.7612 ± 0.0018 versus a_local,direct^B ≃ −4.7739, a shift that corresponds to about 0.06 magnitudes. Depending on details of the analysis and calibration, that displacement is reported at the 3–7σ level. Removing this local a_B discrepancy yields a higher H0 of 73.4 ± 1.0 km/s/Mpc, a value consistent with the local distance‑ladder (SH0ES) measurements.

They also study a later redshift mismatch. In the Dark Energy Survey Year‑5 sample (DES‑Y5) they find a_B moves near z ≃ 0.1. When that late‑time intercept shift is removed, the data show much less preference for a dynamically changing dark energy. An updated comparison that adds the DES‑Dovekie sample indicates the z ≃ 0.1 effect is largely driven by tension between DES supernovae on one side and DESI plus Planck constraints on the other. In that situation, fits that allow dark energy to change with time can appear preferred only as a compromise between inconsistent datasets.