The Hubble Tension Is Real

The Hubble constant is a measurement of the expansion of the universe, sometimes attributed to a cosmological constant in General Relativity (and the source of more than two-thirds of the mass-energy of the universe in conventional cosmology). Except, it appears that the Hubble constant isn't quite constant. So the explanation must be more complicated than a simple cosmological constant.

The Hubble tension isn't huge in relative terms, 10% over measurements more than ten billion years removed from each other. 

But it is highly statistically significant at the five sigma plus level, and isn't a simple methodological artifact of late time Hubble constant measurements (although it could be a methodological artifact of model dependent cosmic microwave background radiation measurements).

Context. The direct empirical determination of the local value of the Hubble constant (H(0)) has markedly advanced thanks to improved instrumentation, measurement techniques, and distance estimators. However, combining determinations from different estimators is nontrivial due to their correlated calibrations and different analysis methodologies.

Aims. Using covariance weighting and leveraging community expertise, we have constructed a rigorous and transparent “Distance Network” to find a consensus value and uncertainty for the locally measured Hubble constant.

Methods. Experts across all relevant distance measurement domains were invited to critically review the available datasets spanning parallaxes, detached eclipsing binaries, masers, Cepheids, the tip of the red giant branch, Miras, carbon-rich asymptotic giant branch stars, Type Ia (SNe Ia) and Type II supernovae, surface brightness fluctuations, the fundamental plane, and Tully–Fisher relations. Before any calculations, the group voted for first-rank indicators to define a “baseline” Distance Network. Other indicators were included to assess the robustness and sensitivity of the results. We provide open-source software and data products to support full transparency and future extensions of this effort.

Results. Our key findings are as follows: (1) The local H(0) is robustly determined, with first-rank indicators internally consistent within their uncertainties. (2) A covariance-weighted combination yields a relative uncertainty of 1.1% (baseline) or 0.9% (all estimators). (3) The contribution from SNe Ia is consistent across compilations of optical or NIR magnitudes. (4) Removing either Cepheids or the tip of the red giant branch has a minimal effect on the central value of H0. (5) Replacing SNe Ia with galaxy-based indicators changes H(0) by less than 0.1 km s^−1 Mpc^−1 while doubling its uncertainty. (6) The baseline result is H(0) = 73.50 ± 0.81 km s^−1 Mpc^−1, 7.1σ from the early Universe plus ΛCDM result 67.24 ± 0.35 km s^−1 Mpc^−1 and 5.0σ from BBN+BAO within a flat ΛCDM DESI DR2 (68.51 ± 0.58 km s^−1 Mpc^−1).

Conclusions. A networked approach, such as the one presented here, is invaluable for enabling further progress in Hubble constant measurements, as it provides the much needed advances in accuracy and precision without overreliance on any single method, sample, or group.

H0DN Collaboration, "The Local Distance Network: A community consensus report on the measurement of the Hubble constant at ∼1% precision" 708 A&A, A166 (April 10, 2026).

arXiv Is Moving

A few weeks ago, arXiv.org announced that it will be leaving Cornell, the university that currently manages it, and establishing its own nonprofit.

From 4 gravitons.

Calculating Light Meson Masses From First Principles In QCD

How good are current Standard Model calculations at predicting the experimental values of the light meson masses?

A new paper that makes that attempt for most light mesons under 1.5 GeVs of mass (except scalar mesons). And, physicists are finally starting to do a pretty good job of describing the meson mass spectrum which has been an elusive target for decades, even for axial vector mesons, which had long been challenging.

As explained in the introduction:

In the present work we employ the procedure described above to compute the masses of relatively light mesons, namely mesonic states no heavier than about 1.5 GeV. Specifically, for mesons composed of u and ¯d quarks, we compute the masses of π±, ρ(770), b1(1235), a1(1260), π±(1300), and ρ±(1450). For the strange sector, we calculate the masses of the states K±, K∗(890), K1A, K1B, and K±(1460). 

In general, the computed masses are in good agreement with the experimental values. In fact, our findings represent a definite improvement over the results obtained within the standard rainbow-ladder truncation [84], where the masses of axial-vector mesons and radially excited states tend to deviate considerably from the observed values.

Notably, this omits the f(0)(500) scalar meson a.k.a. the sigma meson and seven other true scalar mesons with masses under 1.5 Gev. The other omitted scalar mesons are the f(0)(980), f(2)(1270), f(1)(1285), f(0)(1370), f(1)(1420), f(2)(1430) and f(0)(1500). This may be because their internal structures are less well understood.

The actual procedure used is too technical to discuss at this blog, which is aimed at an education layman readership.

The money chart is as follows:

With the exception of spin-1 kaons (where the relationship is inverted for some reason), the experimental values (in red) tend to be at the very high end of the theoretically predicted values using their methods (in blue), and their predictions, in turn, tend to be more massive than those made using a previous "rainbow ladder" truncation method (in green).

The predictions (and measurements) of excited state light meson masses are much less precise than the predictions (and measurements) of ground state light meson masses.

Is The Newtonian Expectation For Galaxy Rotation Curves Modeled Incorrectly?

The conclusion of this paper is a very big deal if true, and I don't dismiss it out of hand.

But given how well established and widely used the models it claims are grossly wrong are, this needs peer review and time for commentary papers in response to it in order to be taken seriously. I wouldn't be surprised if it contains some significant conceptual flaw.

The approximately flat outer parts of spiral galaxy rotation curves are commonly interpreted as evidence for a discrepancy between the observed baryonic mass and the dynamical mass inferred from the measured orbital velocities. In most standard analyses, this discrepancy is quantified using v2(R)=GM(<R)/R, which is exact only under spherical symmetry. However, spiral galaxies are flattened disk systems, for which mass exterior to the galactocentric radius under consideration can contribute non-negligibly to the gravitational field. 

We introduce the Lost and Found (LF) model, a geometrically consistent Newtonian framework based on direct full-disk gravitational integration and a parametrized representation of the disk surface density. In this approach, the gravitational field is computed without imposing spherical symmetry, and the disk mass distribution is represented by two exponential components with a smooth outer truncation. 

We apply the LF model to a heterogeneous sample of disk galaxies spanning a broad range of masses and radial extents. The model reproduces the main observed features of the rotation curves, including the inner rise and the approximately flat outer behavior, without explicitly invoking a dark matter halo or modifying Newtonian gravity. Across the sample, the LF-inferred mass scales nearly linearly with the conventional dynamical mass, with a characteristic reduction factor ηLF ~ 0.67. 

These results indicate that part of the inferred mass discrepancy may arise from the geometric treatment of gravitation in disk galaxies, and motivate a reassessment of mass inference in non-spherical systems.

Adolfo Santa Fe Dueñas, "Galactic Rotation Curves from Full-Disk Newtonian Gravity: The Lost and Found Model" arXiv:2604.06917 (April 8, 2026) (submitted to MNRAS).