Quasar variability briefly threatened the expansion of the universe itself. Time dilation requires every distant clock to run slow by a factor of (1+z), and supernova light curves obligingly stretch with redshift. Quasar light curves, in a celebrated analysis of 28 years of monitoring data (Hawkins 2010), did not: the characteristic variability timescales of nearly 900 quasars showed no redshift dependence at all, a result so at odds with expectation that it spawned explanations from microlensing-dominated variability to nonstandard cosmologies. The anomaly stood for over a decade as a quiet embarrassment, too strange to integrate, too careful to dismiss.
The resolution arrived in 2023: Lewis and Brewer (Nature Astronomy) analyzed 190 quasars with two decades of multi-band monitoring, modeling the variability as a damped random walk, and detected exactly the (1+z) dilation cosmology requires. The reconciling diagnosis is that earlier null results conflated the intrinsic luminosity and wavelength dependence of variability timescales with the cosmological stretching, and lacked the cadence to separate them. Some tension residue remains, the Hawkins dataset's null has not been fully re-derived under the new methodology, and quasar timescale systematics (black hole mass, accretion state, observed band) still entangle with redshift in ways that demand care, but expansion's clock test now reads as passed.
The standing is a resolved anomaly with instructive residue: time dilation is confirmed in quasars, the decade of null results stands as a warning about population systematics in cosmological clocks, and Rubin-LSST's dense long-baseline light curves will turn quasar dilation into a precision measurement rather than a consistency check.