Space & signals
Five million objects, seven unusual candidates: a search for stellar megastructures
Seven infrared-excess objects survived a large search. Follow-up on candidate G shows why a shortlist is the beginning of an investigation.
Searching for heat, not a photograph of a sphere
Project Hephaistos combined Gaia, 2MASS and WISE measurements to search for objects whose infrared emission might fit a model of stellar energy collection. The 2024 paper by Matías Suazo and colleagues narrowed a sample of about five million sources to seven candidates. All seven were associated with M-dwarf stars.
The underlying idea concerns energy: a structure absorbing starlight would have to release waste heat. An unusual infrared signature can therefore motivate a search. It is not unique to engineering. Dust, other astrophysical sources and imperfect separation of nearby objects can produce signals that need careful examination.
Different telescopes do not see identical dots
Combining catalogues requires deciding whether measurements at different wavelengths belong to the same physical source. Instruments have different resolution, and a foreground star can lie close on the sky to a much more distant galaxy. A measurement that blends them may look like one exceptionally strange object.
The original search included filters and image inspection to remove contaminants. Those steps were substantial, but a selection procedure cannot guarantee that every surviving case has the proposed origin. Candidate status means the object warrants additional scrutiny under the search's criteria. It is not a probability certificate or a confirmed discovery.
Candidate G received a closer look
A 2024 follow-up by Tongtian Ren, Michael Garrett and Andrew Siemion identified possible background contamination. A 2025 paper then reported higher-resolution radio observations associated with candidate G. The radio source resolved into components characteristic of an active galactic nucleus, while the radio and mid-infrared positions supported a background-galaxy explanation.
That follow-up is stronger than a generic statement that contamination is possible. It connects the proposed alternative to observations of this particular candidate. The authors argued that the background source explains the infrared anomaly attributed to G. Their discussion also considered other candidates, but the direct high-resolution result should not be silently expanded into identical proof for all seven.
A useful search can return false positives
A search designed for rare phenomena must work through ordinary phenomena that resemble its target. That is not inherently a failure. Its quality depends partly on whether the false positives can be identified and whether the procedure improves after follow-up.
Read the original candidate table alongside the later images and position comparisons. The story then becomes more informative than either seven discovered Dyson spheres or a search dismissed because a candidate changed status. It shows how catalogue selection becomes an observational question. The archive preserves both the initial anomaly and the evidence that revised its interpretation, so a 2024 headline does not become a permanent claim after the scientific record has moved on.
Sources and further reading
- Project Hephaistos – II. Dyson sphere candidates from Gaia DR3, 2MASS, and WISE ↗
Suazo et al., arXiv 2405.02927; selection pipeline, seven-candidate results and contamination discussion
- Ren, Garrett and Siemion, Background Contamination (2024) ↗
Abstract; radio associations with candidates A, B, G
- Ren, Garrett and Siemion, High-resolution imaging of candidate G (2025) ↗
§§3.1–3.2, 4–5; radio components and wavelength-dependent centroids