First-hand information for everyone
This paper asks whether the standard brightness calibration of Type Ia supernovae (SNIa) might change with cosmic time. Such a change could shift distance measurements and affect key puzzles in cosmology, including the disagreement in the Hubble constant measured from the early Universe and from local distance ladders. The authors test a simple, phenomenological correction that makes supernova magnitudes vary in proportion to cosmic look-back time — the time elapsed between now and when the light was emitted — and ask how that would change cosmological inferences.
The correction is written as Δm(z) = S_SN × Δt(z) in magnitudes per gigayear, where Δt(z) is the look-back time at redshift z and S_SN is a free amplitude. They fit this one-parameter model using standard tools (Markov-chain Monte Carlo with Boltzmann-code calculations) to a combination of datasets: cosmic microwave background (CMB) data from Planck and ACT, baryon acoustic oscillation (BAO) distances from DESI Data Release 2, and the Pantheon+ collection of 1,550 SNIa spanning redshifts 0.001 to 2.26. They consider both the usual Λ Cold Dark Matter (ΛCDM) model and a dynamical dark energy model (called w0wa) whose equation of state can change with time. They run fits both without and with the SH0ES prior on the SNIa absolute magnitude (Mb), which represents the local distance-ladder calibration.
When they do not impose the SH0ES prior on Mb — in other words, when SNIa are treated as uncalibrated — the data show no evidence for a redshift-dependent calibration. The bias parameter S_SN is constrained to be of order a few 10^-3 and consistent with zero (for example S_SN ≈ −0.005 ± 0.003 in ΛCDM). Allowing this extra degree of freedom has a negligible effect on the main cosmological parameters in ΛCDM and does not qualitatively change the conclusions about dark energy in the w0wa model. The analysis also disfavors a previously proposed larger amplitude (about 0.03) for such a trend.
The story changes when the authors include the SH0ES prior on Mb. In that case a nonzero S_SN is weakly preferred in ΛCDM. With the dynamical dark energy model the preference for a nonzero calibration grows strongly (reported at 4.3σ). In that combined case the fit can accommodate both the distance-ladder Mb and the early-Universe CMB+BAO constraints, and the inferred Hubble tension is reduced to about 1.5σ. The best-fit solution in this scenario behaves like a constant dark-energy equation of state with w < −1 (often called “phantom” dark energy), and the fitted sign of S_SN implies intrinsically dimmer SNIa at higher redshift, opposite to an earlier claim in the literature.
There are important caveats. The correction studied is purely phenomenological: it is not a physical model and is not claimed to represent supernova progenitor ages. The authors explicitly note that current observations find no significant age-related brightness bias at low redshift. The apparent resolution of the Hubble tension only appears when two things happen together: the SH0ES prior on Mb is imposed, and the late-time expansion history is allowed to deviate from ΛCDM. In plain terms, the data do not support a redshift-dependent calibration within the standard ΛCDM model, although such a calibration can help reconcile datasets if one also allows nonstandard late-time dark energy behavior. More work and independent tests would be needed to show whether any such effect is real and physically motivated.