Astronomy detects things by what they emit. Stars shine. Galaxies glow in radio, infrared, X-ray. Gas clouds fluoresce. Even black holes — which emit nothing directly — are found through the radiation of infalling matter, the warping of background starlight, or the gravitational waves of their collisions. The object is dark, but the detection is bright.
Note (2026-11-15; read against the paper's full text): The paper's constraint is a back-of-envelope argument, in a section it titles naive estimates, not an analysis of existing CMB maps. It computes a shadow's area and negative flux against the blackbody background and argues that a black hole of about 2 times 10 to the 17 solar masses at the last scattering surface would have a shadow roughly the angular size of the full Moon, 'obviously' not seen; I found no matched-filter or catalog search of microwave data in it, so 'Lacki uses existing CMB data to set constraints' overstates it. The reason shadows get easier to detect past redshift 1.6 is also different from the essay's flashlight-and-wall picture: in the paper the angular diameter distance falls beyond that redshift (to about 13 Mpc at last scattering, from a peak near 1.79 Gpc), and the CMB is much brighter at high redshift, so the negative luminosity grows. The paragraph about a cold spot having the wrong shape for noise and the wrong spectrum for dust, and the examples from epidemiology, ecology and linguistics, are the essay's own.
Lacki (2026) inverts this. If black holes more massive than a trillion suns exist — “stupendously large black holes,” or SLABs — they would cast shadows on the cosmic microwave background. The CMB is the most uniform light source in the universe, a wall of ancient radiation filling every direction. A SLAB between us and the CMB would block some of that radiation, creating a cold spot: a negative source.
The term is precise. In radio astronomy, a source is a localized excess of flux above the background. A negative source is a localized deficit. Not noise, not a gap in the instrument — an actual reduction in signal caused by something physical. The absence is the data.
The geometry produces a counterintuitive result: at a fixed mass, a SLAB's shadow becomes easier to detect at higher redshifts past z = 1.6. More distant objects cast more detectable shadows. This happens because the CMB lies behind everything — the farther away the SLAB, the more of the CMB screen it can block from our perspective. The flashlight-and-wall effect: push the wall back, and the shadow grows.
Lacki uses existing CMB data to set constraints. No SLABs above 10^17 solar masses exist within the observable universe's last scattering surface. The cosmological density of SLABs between 10^15 and 10^18 solar masses is less than 10^-5 of the critical density. These are among the strongest constraints on objects this massive, derived entirely from what isn't there.
The method only works because the background is known. The CMB has been mapped to extraordinary precision. Any deviation from its expected spectrum is either noise (characterized), foreground contamination (modeled), or signal. A cold spot consistent with a gravitational shadow — wrong shape for instrumental noise, wrong spectrum for foreground dust — would be evidence for something that emitted nothing.
This is a specific case of a general method: detecting things by what they prevent. In epidemiology, a disease can be identified by the absence of expected cases in a vaccinated population (the “prevented fraction”). In ecology, a keystone predator is sometimes identified by what happens to the ecosystem when it's removed — the detection is retrospective, through absence. In linguistics, a phoneme can be defined by the words that would become ambiguous without it (minimal pairs are tests of what distinguishing work a sound does).
The negative source is the cleanest version because the background is universal and well-characterized. The CMB doesn't cooperate, doesn't adapt, doesn't evolve. It simply radiates, uniformly, in every direction. Against that uniformity, absence becomes geometrically precise. You can measure the shape, depth, and redshift of what isn't there, and from those measurements, constrain what must be.
The largest objects in the universe — if they exist — would be found not by what they do, but by what they don't allow. The signal is the shadow. The evidence is the absence. The detection is negative.