Adherents of the probe reading tell 1991 VG as a visit. A small, bright object found in November 1991 approached Earth on an orbit almost identical to ours, seemed to slow and loop around the planet like a satellite, and flickered in a regular rhythm that looked like flashing panels. Some astronomers, they note, openly floated an artificial origin, one wrote about an alien probe, and then, in their telling, the military classified the data. The object was found again in 2017, and its orbit and the debate around it are published. The question is whether the early oddities point to a machine or to a small natural body on a rare orbit.
The fullest account of the debate is by Carlos and Raúl de la Fuente Marcos in Monthly Notices of the Royal Astronomical Society in 2018 (https://arxiv.org/abs/1709.09533). They write that 1991 VG, discovered on 1991 November 6, "was the first near-Earth object (NEO) ever found on an Earth-like orbit," and that "others argued that an artificial (terrestrial or extraterrestrial) origin was more likely." The paper lists the candidates: a returning spacecraft or rocket stage, lunar ejecta, the asteroid belt, and "an extraterrestrial artefact (e.g. Steel 1995a)." It notes that the object "was not detected by radar at NASA's Goldstone Deep Space Network on December 12." It concludes that "there is no compelling reason to believe that 1991 VG is not natural," and that almost certainly it "was never launched from the Earth."
The flickering came from the first light curve, and its original reading is recorded. R. Brasser and P. Wiegert, in "Asteroids on Earth-like orbits and their origin," Monthly Notices in 2008 (https://physics.uwo.ca/~pwiegert/papers/2008MNRAS.386.2031.pdf), recount that J. V. Scotti reported the object with the Spacewatch telescope and that Brian Marsden's orbit was "strikingly similar to that of the Earth." Short-period brightness changes were observed, and "the light curve appeared as that of a rapidly rotating satellite with highly reflective panels." That is the rhythmic signal in its documented form: reflected sunlight varying as the object turned. Brasser and Wiegert judge a low-inclination Amor or Apollo asteroid "arguably the more likely source" and estimate that about 1 in 20,000 near-Earth asteroids ever reach such an orbit, typically for about 10,000 years. They put the impact probability for such objects at less than 1 in 200,000 over the next 5,000 years.
The recovery tested the artificial idea directly. The European Space Agency's Near-Earth Object Coordination Centre reported "The case for 1991 VG: A long sought recovery" (https://neo.ssa.esa.int/-/the-case-for-1991-vg-a-long-sought-recovery). With ESO's Very Large Telescope, the team detected the object on 30 May 2017 at magnitude 25 and confirmed it on 1 June. "If it had been an artificial object, its low bulk density would have resulted in a noticeable drift from a purely gravitational orbit during these 25 years, due to the effects of solar radiation pressure." The updated orbit "does not show apparent evidence of such drift," and is "compatible with the typical densities of a rocky or metallic asteroidal object." The team calls this a guess that "further detailed studies shall confirm."
The orbit is public in NASA's catalogue. The Jet Propulsion Laboratory's Small-Body Database entry for 1991 VG (https://ssd.jpl.nasa.gov/tools/sbdb_lookup.html#/?sstr=1991%20VG) classes it as an Apollo asteroid with a semi-major axis of 1.03 astronomical units, eccentricity 0.0525 and inclination 1.43 degrees. Its minimum orbit intersection distance with Earth is 0.00227 au. The solution uses 70 observations from 1991-11-06 to 2017-06-01, and the absolute magnitude is 28.3. ESA's account describes it as roughly a ten-metre object. The 2018 paper explains the long gap between sightings: the object "has spent over a quarter of a century at small solar elongation angles, out of reach of ground-based telescopes."
The looping around Earth is a known natural behaviour. Mikael Granvik, Jeremie Vaubaillon and Robert Jedicke calculated "The population of natural Earth satellites" (https://arxiv.org/abs/1112.3781), objects "temporarily captured from the near-Earth-object (NEO) population." They find that "at any given time there should be at least one NES of 1-meter diameter orbiting the Earth," and that a typical captured object makes about three revolutions during a capture that lasts about 286 days. Their results match "the single known natural TCO, 2006 RH120." The 2018 paper confirms that 1991 VG itself "was briefly captured by Earth's gravity as a minimoon during its previous fly-by in 1991-1992." Slowing and looping near Earth is what captured asteroids do.
Artificial objects do turn up among near-Earth discoveries, and the way they are recognised is documented. NASA's Jet Propulsion Laboratory reported on 2 December 2020 that "Near-Earth Object (NEO) 2020 SO is, in fact, a 1960's-Era Centaur rocket booster" (https://www.jpl.nasa.gov/news/new-data-confirm-2020-so-to-be-the-upper-centaur-rocket-booster-from-the-1960s/). Its orbit traced back to a 1966 approach suggesting an Earth origin, and Paul Chodas linked it to the Surveyor 2 mission. Vishnu Reddy's team compared its infrared spectrum with stainless steel and with another Centaur booster in orbit, and found them consistent. For 1991 VG, the orbital test points to a natural body; for 2020 SO, orbit history and spectra together pointed to hardware.
Read against the published record, 1991 VG is a roughly ten-metre near-Earth asteroid on a rare Earth-like orbit, briefly captured by Earth in 1991 and 1992, and recovered in 2017 without the drift a hollow object would show. Its early light curve suggested panels to some observers, and an artificial origin was openly debated in the literature. No source records a military classification of the data.
What Stays Contested
The light curve from 1991 has not been repeated in detail, and the recovery team called its density result a guess pending further study. Steel's artefact paper is part of the published debate, and it remains one of the proposals the 2018 review weighed. The secret-radar and classification claims appear without a source; the one recorded radar attempt, at Goldstone, did not detect the object.