The expansion rate of the universe — the Hubble constant, H0 — can be measured in two independent ways that ought to converge on the same number. They do not. The gap between them has hardened, over more than a decade of increasingly careful work, from a curiosity into one of the most consequential unresolved problems in physics. Cosmologists call it the Hubble tension, and the more it is scrutinised, the more stubborn it becomes.
Two roads to one number
The first road starts in the early universe. The cosmic microwave background — the relic radiation released about 380,000 years after the Big Bang — carries a pattern of acoustic peaks whose geometry encodes the contents and expansion history of the cosmos. Feeding the Planck satellite's exquisite measurement of those peaks through the standard ΛCDM model yields H0 ≈ 67.4 kilometres per second per megaparsec, with sub-percent precision.
The second road is built in the nearby universe, rung by rung, as a cosmic distance ladder. Geometric parallax calibrates the intrinsic brightness of Cepheid variable stars; Cepheids in turn calibrate Type Ia supernovae, the standardisable explosions bright enough to be seen across cosmological distances; and those supernovae are used to measure the expansion directly. The leading effort of this kind, the SH0ES program, arrives at H0 ≈ 73. The two numbers differ by roughly 9 percent, and as both methods have tightened their error bars, the discrepancy between them has grown past five sigma — the threshold physicists conventionally treat as a discovery rather than a fluctuation.
Closing the loopholes
The obvious resolution would be a hidden systematic error in one method, and for years the distance ladder was the prime suspect. Calibrating Cepheids is delicate work: crowding of stars in photometric images, the dependence of the period-luminosity relation on a star's metallicity, and the standardisation of supernovae all introduce potential biases. Each has been interrogated in turn. Most pointedly, the James Webb Space Telescope has now re-observed many of the same Cepheid host galaxies at infrared wavelengths, where crowding is far less severe than in the optical, and the results have broadly confirmed the earlier measurements rather than erasing them. Independent calibrators, such as the tip of the red giant branch, tend to land between the two camps without collapsing the tension.
If the discrepancy is real
The harder possibility is that the measurements are right and the model is incomplete. Because the early-universe value depends on ΛCDM to extrapolate forward, a genuine tension would most likely indicate new physics in the early universe, and most serious proposals modify the sound horizon — the standard ruler that the cosmic-microwave-background route relies on. Early dark energy, a transient component that would have acted briefly around the epoch of recombination, is the most discussed candidate: it can raise the inferred H0, but it tends to degrade the fit to other cosmological data, trading one problem for another. No model yet resolves the tension cleanly.
That is precisely what makes the Hubble tension serious rather than merely irritating. It is not one shaky result that better data will quietly retire. It is two mature, independently scrutinised programs, each internally consistent, that disagree at high significance. Resolving it will require either exposing a subtle systematic that has eluded a decade of concentrated effort, or revising the cosmological model that underpins essentially everything else we believe about the universe's composition and history. Either outcome would be a milestone — and for now, the most honest summary is that one of the foundations of modern cosmology has a crack in it that no one has been able to seal.
The independent referees
What keeps the debate from collapsing into a standoff between two camps is the steady arrival of independent cross-checks that owe nothing to the Cepheid ladder or the cosmic microwave background. The tip of the red giant branch offers an alternative distance calibrator; strongly lensed quasars, whose images flicker on measurable time delays, yield H0 from pure geometry; and gravitational-wave "standard sirens" like the 2017 neutron-star merger provide distances directly from the physics of the waveform. None has yet been precise enough to settle the question on its own, but collectively they are closing off the escape routes — making it progressively harder to blame any single technique and progressively likelier that the tension is telling us something real about the universe rather than about our instruments.
