The case
In 1960 the physicist Freeman Dyson proposed that an advanced civilization might be found by the infrared glow of the starlight it captures. Suazo and colleagues open their 2024 paper on Project Hephaistos by recounting that idea: structures built to absorb light around a star would re-emit it as infrared. Their pipeline screened about five million sources from the Gaia, 2MASS and WISE catalogues within 300 parsecs and named seven candidates, all M dwarfs, for which "astrophysical phenomena cannot easily account for the observed infrared excess emission." The same paper states the limit of the method: it is "essentially impossible to prove the existence of a Dyson spheres based on photometric data only."
A second line of work looks backward. Solano, Villarroel and Rodrigo (2022) list 298,165 sources that appear only on the red plates of the 1950s Palomar Observatory Sky Survey, with 5,399 still unidentified after cross-matching. They report that no point source seen on both the first and second surveys then vanished, which puts the Milky Way failed supernova rate over 70 years below one in a billion.
The best-known single case is a red plate taken on 12 April 1950. Villarroel and colleagues (2021) describe nine point sources that appeared within half an hour inside a region of about ten arcminutes, absent from the blue exposure taken thirty minutes earlier and from later images. The authors name two possibilities: an unknown plate contamination, or, if contamination can be excluded, very fast solar reflections from objects near geosynchronous orbit.
In 2025 Bruehl and Villarroel built a daily dataset of 2,718 days from 19 November 1949 to 28 April 1957. They report that transients were 45 percent more likely on dates within one day of a nuclear test (p = .008), and that each additional independent UAP report on a date went with an 8.5 percent increase in transients. The paper describes the hypotheses it tests as speculative.
The adherents' reading
Adherents read the infrared excesses and the vanishing lights together. To them the excesses are the first outlines of megastructures, and the lights that appeared and disappeared on 1950s plates are reflective objects circling Earth before the first human satellite flew.
The mainstream view
In 2026 Zackrisson and colleagues pointed the James Webb Space Telescope at two of the seven candidates, D and E. Their imaging and spectroscopy trace the infrared excess to background galaxies within about one arcsecond of the stars, one at a redshift of about 0.9 and the other at about 0.4, and not to Dysonian megastructures or other emission close to the stars. The other five candidates were not part of that study.
Scientific American (28 October 2025) collected named comments on the plate work. Michael Wiescher of Notre Dame said "Nuclear tests obviously have an impact on the atmosphere" and can leave "a lot of junk in the outer atmosphere." Sean Kirkpatrick, head of the Pentagon's AARO from July 2022 to December 2023, said the first thing that came to mind was "solar flare radiation or ionized particle radiation from nuclear testing," and also named high-altitude balloons. Nigel Hambly of Edinburgh urged studying the original plates rather than digital copies. Robert Lupton of Princeton was reported as saying that aligned patterns across thousands of plates "could easily be mere coincidence." The magazine also reports that arXiv did not accept the papers.
The adherents' answer
Adherents answer that the Webb study covered two of the seven Hephaistos candidates, so five remain untested. They add that the 2025 paper still describes the Palomar transients as of unknown origin.