Somewhere among the roughly 300 billion stars of the Milky Way, could one or more be wrapped in an artificial shell, its starlight dimmed and its waste heat leaking out as an infrared glow?
That’s what Freeman Dyson wondered when he proposed the spherical structures surrounding stars we now call Dyson Spheres, and the question sits at the heart of Project Hephaistos, a real observational search that has already turned up candidates worth a second look!
That’s part of what Erik Zackrisson presented this year during IAUS404: Advancing the Search for Technosignatures.
About the Presenter
Erik Zackrisson is a Professor of Astrophysics at Uppsala University in Sweden, where his primary research concerns extragalactic astronomy: the first stars, first galaxies, and first black holes, much of it now pursued with the James Webb Space Telescope. Alongside that work, he leads Project Hephaistos, a small technosignature-search project named for the Greek god of blacksmiths, the craftsman behind the mechanical servants and animated devices of the other Olympian gods.
Zackrisson has led the project through several published stages, from early galaxy-scale searches for advanced civilizations to the current effort to find individual Dyson spheres among nearby stars.
His technosignature work runs alongside a substantial career in cosmology and high-redshift astrophysics, giving Project Hephaistos an unusually rigorous statistical and observational footing.
More About Zackrisson’s Work in Advancing the Search for Technosignatures
A Dyson sphere, as Zackrisson explained, is a hypothetical megastructure an advanced civilization might build around a star to capture its energy, an idea popularized by physicist Freeman Dyson in a 1960 paper in Science. Rather than a solid shell, most modern thinking favors a “Dyson swarm,” a large number of independent absorbing elements in orbit, or a “Dyson bubble,” which uses light sails to hover above the star without orbiting at all.
A 2013 study by Stuart Armstrong and Anders Sandberg estimated that self-replicating robots could dismantle Mercury and build the early stages of such a swarm on a timescale comparable to constructing the Great Pyramid of Giza.
The signature such a structure would leave behind is straightforward in principle: it blocks some of a star’s visible light while re-radiating that captured energy as heat in the mid-infrared, shifting the star’s spectral energy distribution. Project Hephaistos hunts for exactly that shift by cross-matching three surveys: Gaia (for optical brightness and crucial parallax distances), 2MASS (near-infrared), and WISE (mid-infrared).
The trouble, as Zackrisson detailed, is that plenty of ordinary astrophysics mimics this signal: young stellar objects, dusty debris disks, and catalog artifacts like blended sources or telescope-optics “ghosts.” Weeding these out required a multi-stage pipeline:
Keep only stars detected in the longer WISE mid-infrared bands
Fit each candidate’s spectrum to a model of a star plus a partial Dyson sphere
Use a convolutional neural network to flag WISE images with messy, dust-cluttered surroundings
Discard stars with hydrogen-alpha emission or optical variability, both signs of youth
Check Gaia’s astrometric solution to exclude binaries and bad distances
Have a human eyeball whatever survives
Applied to five million stars within 300 parsecs, this process (detailed in Suazo et al. 2024, MNRAS) surfaced seven candidate M dwarfs, labeled A through G, with covering fractions around 10 percent and Dyson-sphere temperatures of roughly 100 to 300 Kelvin. Follow-up optical spectra from the Nordic Optical Telescope confirmed they look like ordinary M dwarfs with no significant hydrogen-alpha emission.
The most compelling part of the talk concerned what comes next: JWST follow-up. Zackrisson’s team won time to image and take spectra of three candidates, and the results for two of them, published shortly after this symposium, resolved the mystery decisively. Using JWST’s Mid-Infrared Instrument, the team found that the infrared excess for candidates D and E did not come from Dyson spheres or debris disks at all, but from background galaxies lying within about an arcsecond of each star, close enough to have been blurred together with the star in WISE’s lower-resolution images. Candidate D’s contaminant is a galaxy at redshift 0.9 whose infrared spectrum matches a Hot Dust-Obscured Galaxy, a rare, extremely dusty, active-galaxy type. A third candidate, A, was still awaiting its JWST observation at the time of the talk.
Key Takeaways
Dyson spheres would be detectable through a specific fingerprint: reduced optical light paired with excess mid-infrared “waste heat.”
Distinguishing a real megastructure candidate from a young star, dusty disk, or catalog artifact requires layered filtering, including deep learning image analysis.
Project Hephaistos identified seven M dwarf candidates from five million stars, with a few more still working their way through peer review.
Two of the three candidates sent to JWST turned out to be background galaxies, not Dyson spheres, illustrating how survey-resolution limits can manufacture false positives.
Even a “negative” result advances the field, refining the search strategy for the next round of candidates.
Project Hephaistos illustrates something valuable about technosignature science even when a candidate doesn’t pan out: each false alarm sharpens the filters for the next search, and each null result narrows what remains hidden. As WISE-scale infrared surveys reach the limits of what ground- and space-based follow-up can resolve, the approach Zackrisson described, using JWST to directly interrogate ambiguous cases, may become a standard tool for testing megastructure candidates wherever they turn up.
This video was included in the proceedings of the International Astronomical Union (IAU) symposium #404, Advancing the Search for Technosignatures, hosted by Blue Marble Space. Stay tuned as we explore the boundaries of knowledge in technosignature science!
Thanks for joining us here with Planetary Perspectives, an editorial endeavor and science communication project from Blue Marble Space.
