37 pairs of galaxies of different ages served as clocks. Their hands, the difference in stellar ages, told how fast the Universe expanded at different epochs. Based on these readings, cosmologists built a lie detector for dark energy. The suspect is the cosmological constant itself, the simplest form of dark energy, whose density is by definition unchanging. It is accused of being too simple: too constant. But new data from the DESI instrument hinted that dark energy could vary over time. The lie detector put these claims to the test.
The lie detector is a hierarchical Gaussian process (GP), a mathematical method that does not impose a shape on history but listens to the data. Unlike conventional approaches where dark energy is assigned a specific equation, GP searches for any deviations from a constant, relying on direct measurements of the expansion rate: those same 37 points from galaxy chronometers. The scheme resembles a polygraph interrogation: the device registers the slightest fluctuations in the readings but does not suggest what to say to the witness.
In addition to the chronometers, the case included data on 1624 supernovae (their brightness serves as a distance measure) and imprints of sound waves from the early Universe in the galaxy distribution — baryon acoustic oscillations. Also compressed hints from Planck's cosmic microwave background. GP ran through thousands of possibilities for how dark energy could behave and delivered its verdict: nearby, at redshift z ≈ 0, its parameter w = −0.80, and this is only 0.8 sigma away from −1—the value for an absolute constant. Statistically, it is almost indistinguishable.
A previous analysis by the DESI collaboration using the full Planck map hinted at dynamic energy with a confidence of 2.8–4.2σ — a strong argument in favor of a 'non-constant' dark energy. However, when GP tested the same combination of data with a compressed version of Planck, the signal dropped to 0.8σ. The authors checked whether sensitivity was hidden in the details: they turned off supernovae, removed key BAO ranges, changed the 'instrument' settings. The result remained within one sigma.
The lie detector does not claim that DESI was wrong. It shows: without the full power of the cosmic microwave background, the evidence for evolution is insufficient. Possibly, compressing the Planck map into a few numbers, like translating a complex painting into a brief description, misses details. For now, the verdict is cautious: the Universe is most likely governed by the good old cosmological constant. But the galaxy clocks have not stopped: their ticking may still change the picture.
The 37 cosmic‑chronometer points are direct measurements of H(z), obtained without any model assumption — simply from the age differences of passively evolving galaxies. It is like gauging a child’s growth rate from a series of school photos taken over the years.