The Hubble constant — H₀, the present-day rate of cosmic expansion — has been measured for nearly a century. Hubble's 1929 estimate was about 500 km/s/Mpc, an order of magnitude too high. Allan Sandage and Gustav Tammann pulled the value down toward 50 over decades; Gérard de Vaucouleurs argued for 100. By the early 2000s the Hubble Key Project had narrowed it to about 72, ending what was called the Hubble Wars. The new fight, since 2013, is structurally different: two independent, methodologically rigorous methods of measuring H₀ have converged on different values. The cosmic microwave background gives about 67 km/s/Mpc; the local distance ladder gives about 73. The discrepancy is now ~5σ, and it has not gone away under increasingly careful measurement. This is the Hubble tension.
The CMB-based measurement runs through ΛCDM cosmology: the Planck satellite (ESA, 2018 final release) measures the CMB angular power spectrum to extreme precision, and the spectrum constrains the parameters of the standard cosmological model — including H₀ — given that ΛCDM is the right framework. The Planck value is 67.4 ± 0.5 km/s/Mpc. The local distance ladder runs through Cepheid-calibrated Type Ia supernovae: parallaxes anchor Cepheid luminosities, Cepheids in nearby galaxies anchor Type Ia supernova luminosities, and supernovae out to z ~ 0.1 give a direct H₀. The SH0ES collaboration led by Adam Riess gives 73.0 ± 1.0 km/s/Mpc. The two values disagree by about 8% with non-overlapping error bars — a 5σ discrepancy that has persisted under independent recalibration and cross-check.
Three classes of resolution are possible. A systematic error in one of the measurement chains — Cepheid metallicity, supernova progenitor populations, dust extinction, parallax zero-points — has not been found, and the 2024 Riess JWST result confirming the SH0ES Cepheid distances has weakened the case for a local-ladder systematic. New physics: an additional component in the cosmological model that affects the CMB inference — early dark energy, modified neutrino properties, time-varying dark-energy equation of state, modified gravity — each fits some subset of the data and creates new tensions elsewhere. A statistical fluke is the third option, becoming less plausible as precision improves and the discrepancy holds. The 2024 DESI BAO + supernova combination shows modest preference for evolving dark energy, which if confirmed would be the new-physics resolution.
Gravitational-wave standard sirens — the kilonova counterpart to GW170817 was the first, providing a model-independent H₀ measurement consistent with both Planck and SH0ES at the current large error bar — should constrain H₀ at percent-level precision once tens of multi-messenger events have been observed. The third-generation detectors planned for the 2030s (Cosmic Explorer, Einstein Telescope) are designed in part for this. DESI's full five-year run (through ~2026) will tighten the BAO constraint and test the evolving-dark-energy hypothesis. Roman Space Telescope (NASA, 2027) and Euclid (ESA, science 2024+) bring independent constraints from weak-lensing and large-scale structure. The next ten years should resolve whether the answer is a missed systematic, new physics, or a statistical fluctuation that finally regresses to the mean.