how we know dark energy might be changing

The Mechanism

The *Dark Energy Spectroscopic Instrument* (DESI) is a 5,020-fibre multi-object spectrograph mounted on the *Nicholas U. Mayall 4-metre Telescope* at *Kitt Peak National Observatory* in Arizona, operated by the *Lawrence Berkeley National Laboratory* in collaboration with a 70-institution international consortium. DESI began full survey operations in May 2021; in *March 2025* the collaboration released its *Data Release 2 (DR2)* cosmology results, based on the first *three years of operation*, covering nearly *15 million galaxies and quasars* — more than double the dataset of the earlier 2024 Year-1 release. The DR2 paper, *"DESI DR2 Results II: Measurements of Baryon Acoustic Oscillations and Cosmological Constraints"* (arXiv:2503.14738, submitted 18 March 2025, published in *Physical Review D* 112: 083515, 2025), reports the most precise measurements ever made of *baryon acoustic oscillations* (BAO) — a characteristic spatial scale of about 150 megaparsecs imprinted in the distribution of galaxies, frozen in at the time of recombination (380,000 years after the Big Bang), and visible as a small bump in the two-point correlation function of galaxy positions. By measuring the angular size of this *standard ruler* at multiple redshifts between *z = 0.1* (recent universe) and *z = 4.2* (distant quasars), DESI constructs the *expansion history of the universe* over the last twelve billion years to better than one percent precision. The result is the central probe of the equation of state of *dark energy*. Dark energy is the name given to the substance — or property of the vacuum, or geometric correction to general relativity, depending on whom one asks — that drives the accelerating expansion of the universe, discovered in 1998 by the *Adam Riess* and *Saul Perlmutter* supernova teams. In the simplest case, dark energy is a *cosmological constant* — *Einstein's Λ*, an unchanging energy density of the vacuum that produces exactly the negative pressure required to drive the observed acceleration. In the language of the *equation of state*, a cosmological constant has *w = -1* exactly, at all redshifts. The *standard cosmological model* (*ΛCDM*) assumes this. The DESI DR2 analysis instead fits the more general *w₀w_a-CDM* parametrisation, in which the equation of state can evolve with cosmic time as *w(a) = w₀ + w_a (1 - a)*, where *a* is the cosmic scale factor (*a = 1* today, *a < 1* in the past). Standard ΛCDM corresponds to *w₀ = -1, w_a = 0*. The DESI DR2 BAO data alone are *consistent with ΛCDM*. But when DESI BAO are combined with the *Planck 2018* cosmic microwave background data and with one of the three major Type Ia supernova compilations (*Pantheon+*, *Union3*, or *DES-SN5YR*), the joint fit *prefers an evolving dark energy* over the cosmological constant at *2.5 σ, 3.5 σ, or 3.9 σ significance depending on the supernova dataset*, with a preference for *w₀ > -1* (less negative than -1 today, dark energy currently weakening) and *w_a < 0* (dark energy was stronger in the past). The DR2 preferred best-fit point is approximately *w₀ ≈ -0.84, w_a ≈ -0.61* — a substantial departure from the textbook *w = -1*. Put plainly: the data are starting to favour a model in which dark energy *is not constant*. It may have been a stronger driver of cosmic expansion in the past and is *currently weakening*. The cosmological constant — Einstein's lambda, the simplest case that has been the cosmology-textbook default for twenty-five years — may be wrong. The result has not crossed the *5 σ discovery threshold* and the DESI collaboration is careful not to claim discovery: *3.9 σ* corresponds to a *p-value of about 1 in 14,000* (impressive, but not the *1 in 3.5 million* that the 5σ convention requires); systematic uncertainties in the supernova samples remain debated; and an alternative explanation that does not invalidate ΛCDM — that the *neutrino mass sum* is closer to zero than the lower bound implied by oscillation experiments — would partially soften the tension. But the tension itself is real, robust to the specific BAO analysis, increases over Year-1 DESI tension (which was at 2.6 σ in early 2024), and points toward the same corner of parameter space as several earlier independent indications. Two further DESI data releases are expected (the full five-year survey ends in late 2026), and the *Euclid* satellite, the *Vera C. Rubin Observatory*'s Legacy Survey of Space and Time, and the *Nancy Grace Roman Space Telescope* will each provide independent cross-checks within the next several years. If the hint holds, the cosmological constant is dead — or, more precisely, dark energy is *not* the cosmological constant but a *dynamical field* of some kind (the favourites in theoretical cosmology are *quintessence*, in which a slowly-rolling scalar field plays the role of Λ; *coupled-dark-energy* models, in which dark energy and dark matter exchange energy; or modified-gravity models in which the apparent acceleration is a deviation from general relativity rather than a vacuum energy at all). As of June 2026, the DESI DR2 result is the most precise probe of cosmic acceleration ever made, and the answer it is starting to give is the one that the 1998 supernova teams, the 2003 first-year WMAP, and the 2013 Planck satellite all assumed could not be the case. *The dark energy is not Einstein's lambda. It may be something that changes.*