Merging astrophysical ("real world") black holes are quite different from the idealization in the Schwarzschild setting.<p>You need a gravitational quadrupole to emit gravitational radiation; in essence you need a "barbell" with an infinitesimally small "bar" connecting two large weights. (The weights can differ; the bar's length can change). Schwarzschild is characterized by a central mass, and so can emit no gravitational radiation. (One can do perturbative methods on Schwarzschild to introduce gravitatonal radiation; and in reverse the effective one-body formalism basically (over-simplified) turns a binary black hole with a large mass ratio into a test particle in a Schwarzschild background).<p>Additionally, there is the Vaiyda null dust absorbing metric, where "null dust" is very much like light. In this metric, the central mass grows, and the apparent horizon (where outgoing null rays turn inwards) is r = 2m(v), where v is the advanced time. That locally-defined dynamical surface moves outwards. There is <i>also</i> an event horizon in Vaiyda, but it is neither locally detectable nor coincident with the apparent horizon (that global null surface lies well outside during accretion).<p>That's the closest starting point to an answer for an isolated non-rotating black hole that grows by accretion; one can perturb Vaiyda with something more realistic than a constant "rain" of noninteracting massless (but not momentumless) particles.<p>That's just for monopole growth though. Like with Schwarzschild, Vaiyda is non-radiating (and in V's case this is by construction). An astrophysical isolated single black hole will see infalling material that has angular momentum; anisotropic stresses (beyond the null direction) (radiation from matter-matter interactions); a gravitational quadrupole (emitting gravitational radiation); and for big lumps of infalling material, some self-gravity.<p>In these cases, the black hole "shadow", the topic of the article at the top, reveals the details of the black hole, including growth. The shadow is fundamentally a gravitational lensing effect, and depends on the BH's mass, angular momentum, and charges. For a growing Vaidya black hole, the shadow is not static. The distant observer will see the shadow evolve in various ways. This is because it's a dark region whose boundary tracks the changing spacetime geometry. In the pure Schwarzschild case, the shadow is static, eternal, empty, and set by the boundary of the photon sphere at r = 3M rather (and rather bigger) than the event horizon at r = 2M.<p>For an astrophysical black hole, the observer's view of an infaller will be largely washed away by various sources of noise and the gravitational redshift of the infaller. However projects like the GRAVITY collaboration (at ESO's Very Large Telescope) look for infrared flares around Sgr A*, and have spotted really interesting things like the "G2" gas cloud. This is a busy area of current research. TDECat is a catalogue of more than a hundred tidal disruption events, mostly from X-ray flares and delayed infrared emissions from various space instruments. A lot of this is stuff actually falling into the black hole, but "covered" in a huge burst of high-energy radiation from real matter colliding together faster and faster as it inspirals.<p>Finally, we've seen the gravitational waves emitted by merging black holes of various mass-ratios. The result is a single bigger black hole and an enormous amount of gravitational radiation. The behaviour of matter near the merger and between the merger and us can corroborate that result. So we can see growth in that case, as well as where we can image a BH's shadow. These BHs are not very close to us. So clearly the idea that nobody can watch growth or consumption is wrong, although one might get nitpicky about which of these observables are indirect, but that's like saying that because Miss Marple sees <i>only</i> a bled-out corpse with lots of stab wounds there is no evidence that a murder has been committed.