The universe is expanding, and the pace of that expansion is set by dark energy, the mysterious force accelerating the recession of galaxies. In the standard cosmological model, its density is constant: a cosmological constant, an unwavering metronome since the birth of the cosmos. But new data from the spectroscopic instrument DESI, which has mapped millions of galaxies, make us listen more closely: the rhythm seems to waver slightly.

Astronomers compared distances to distant galaxies, marked by baryon acoustic oscillations (ripples in matter density left over from the early universe), with the brightness of Type Ia supernovae, explosions of white dwarfs that serve as standard candles. Eight different models, in which the equation of state of dark energy (the relation between its pressure and density) can vary over time, were tested against DESI DR2 data and three supernova catalogs. It turns out that nearly all the signal for evolution comes from two types of luminous red galaxies, LRG1 and LRG2, at a redshift of about 0.5, where light is stretched by a factor of 1.5 due to cosmic expansion. Remove them, and the parameters revert to a cosmological constant.

The statistical significance of the deviations does not exceed 2.3 sigma, which is at the level of a random fluctuation rather than a discovery. Bayesian factors, a numerical measure of preference for one model over another, indicate only weak evidence in favor of dynamics. Interestingly, in many models the parameter w crosses the boundary of −1 around a redshift of 0.5, the so-called phantom crossing. But this could be an artifact of fitting a parameterization over a limited range, not a sign of exotic physics.

This work changes the picture: previously it seemed that DESI data confidently pointed to evolving dark energy, but now it is clear that this conclusion rests on two tracers and falls apart without them. Such a conservative approach is not a disappointment but a safeguard against false discoveries. Future surveys, such as Euclid and the Roman Space Telescope, will provide tens of times more data and allow us to distinguish a real rhythm from noise. If dark energy truly varies, it will rewrite the textbooks; if not, we will better understand the limits of our precision.