For a project looking for the signature of an advanced extraterrestrial civilization, the name Hephaistos is an unusually apt choice. And indeed the leaders of Project Hephaistos, based at Uppsala University in Sweden, are quick to point out that the Greek god (known as Vulcan in Roman times) was a sort of preternatural blacksmith, thrown off Mt. Olympus for variously recounted transgressions and lame from the fall, a weapons maker and craftsman known for his artifice. Consider him the gods’ technologist.

Who better to choose for a project that pushes SETI not just throughout the Milky Way but to myriads of galaxies beyond? Going deep and far is a sensible move considering that we have absolutely no information about how common life is beyond our own Earth, if it exists at all. If the number of extraterrestrial civilizations in any given galaxy is scant, then a survey looking for evidence of Hephaistos-style engineering writ large will comb through existing observational data from our own galaxy but also consider what lies beyond. Which is why Project Hephaistos’s first paper (2015) searched for what the authors called ‘Dysonian astroengineering’ in over 1000 spiral galaxies.

More recent papers have stayed within the Milky Way to incorporate data from Gaia, the 2 Micron All Sky Survey (2MASS) and the accumulated offerings of the Wide-field Infrared Survey Explorer, which now operates as NEOWISE, analyzing the observational signatures of Dyson spheres in the process of construction and calling out upper limits on such spheres-in-the-making in the Milky Way. Such objects could present anomalously low optical brightness levels yet high mid-infrared flux. This is the basic method for searching for Dyson spheres, identifying the signature of waste heat while screening out young stellar objects and other factors that can mimic such parameters.

This article is occasioned by the release of a new paper, one that homes in on Dyson sphere candidates now identified. And it prompts reflection on the nature of the enterprise. Key to the concept is the idea that any flourishing (and highly advanced) extraterrestrial civilization will need to find sources of energy to meet its growing needs. An obvious source is a star, which can be harvested by a sphere of power-harvesting satellites. The notion, which Dyson presented in a paper in Science in 1960, explains how a search could be conducted in its title: “Search for Artificial Stellar Sources of Infrared Radiation.” In other words, comb the skies for infrared anomalies.

I strongly favor this ‘Dysonian’ approach to SETI, which makes no assumptions at all about any decision to communicate. As we have no possible idea of the values that would drive an alien culture to attempt to talk to us – or for that matter to any other civilizations – why not add to the search space the things that we can detect in other ways. However it is constructed, a Dyson sphere should produce waste heat as it obscures the light from the central star. Infrared searches could detect a star that is strangely dim but radiant at infrared wavelengths, and we might also find changes in brightness as such a ‘megastructure’ evolves that vary on relatively short timeframes.

Funding plays into our science in inescapable ways, so the fact that Dysonian SETI can be conducted using existing data is welcome. It’s also helpful that in-depth studies of particular Dyson sphere candidates may prove useful for nailing down astrophysical properties that interest the entire community, especially since there is the possibility of ‘feedback’ mechanisms on the star from any surrounding sphere of technology. We go looking for extraterrestrial megastructures but even if we don’t find them, we produce good science on unusual stellar properties and refine our observational technique. Not a bad way forward even as the traditional SETI effort in radio and optics continues.

The number of searches for individual Dyson spheres is surprisingly large, and to my knowledge extends back at least as far as 1985, when Russian radio astronomer Vyacheslav Ivanovich Slysh searched using data from the Infrared Astronomical Satellite (IRAS) mission, as did (at a later date) M. Y. Timofeev, collaborating with Nikolai Kardashev. Richard Carrigan, a scientist emeritus at the Fermi National Accel­era­tor Laboratory, looked for Dyson signatures out to 300 parsecs.

But we can go earlier still. Carl Sagan was pondering “The Infrared Detectability of Dyson Civilizations” (a paper in The Astrophysical Journal) back in the 1960s. In more recent times, the Glimpsing Heat from Alien Technologies effort at Pennsylvania State University (G-HAT) has been particularly prominent. What becomes staggering is the realization that the target list has grown so vast as our technologies have improved. Note this, from a Project Hephaistos paper in 2022 (citation below):

Most search efforts have aimed for individual complete Dyson spheres, employing far-infrared photometry (e.g., Slysh 1985; Jugaku & Nishimura 1991; Timofeev et al. 2000; Carrigan 2009) from the Infrared Astronomical Satellite (IRAS: Neugebauer et al. 1984), while a few considered partial Dyson spheres (e.g., Jugaku & Nishimura 2004). IRAS scanned the sky in the far infrared, providing data of ≈ 2.5 × 105 point sources. However, nowadays, we rely on photometric surveys covering optical, near-infrared, and mid-infrared wavelengths that reach object counts of up to ∼109 targets and allow for larger search programs.

The Project Hepaistos work in the 2022 paper homed in on producing upper limits for partial Dyson spheres in the Milky Way by searching Gaia DR2 data and WISE results that showed infrared excess, looking at more than 108 stars. We still have no Dyson sphere confirmations, but the new Hephaistos paper adds 2MASS data and moves to Gaia data release 3, which aids in the rejection of false positives. Gaia also adds to the mix its unique capabilities at parallax, which the authors describe thus:

…Gaia also provides parallax-based distances, which allow the spectral energy distributions of the targets to be converted to an absolute luminosity scale. The parallax data also make it possible to reject other pointlike sources of strong mid-infrared radiation such as quasars, but do not rule out stars with a quasar in the background.

Notable in the new 2024 paper is its description of the data pipeline focusing on separating Dyson sphere candidates from natural sources including circumstellar dust. The authors make the case that it is all but impossible to prove the existence of a Dyson sphere based solely on photometric data, so what is essentially happening is a search for sources showing excess infrared that are consistent with the Dyson sphere hypothesis. The data pipeline runs from data collection through a grid search methodology, image classification for filtering out young stars obscured by dust or associated with dusty nebulae, inspection of the signal to noise ratio, further analysis of the infrared excess and visual inspection from all the sources to reject possible contamination.

This gets tricky indeed. Have a look at some of the ‘confounders,’ as the authors call them. The figure shows three categories of confounders: blends, irregular structures and nebular features. In blends, the target is contaminated by external sources within the WISE coverage. The nebular category is a hazy and disordered false positive without a discernible source of infrared at the target’s location. Irregulars are sources without indication of nebulosity whose exact nature cannot be determined. All of these sources would be considered unreliable at the conclusion of the pipeline:

Image: This is Figure 5 from the paper. Caption: Examples of typical confounders in our search. The top row features a source from the blends category, the middle row a source embedded in a nebular region, and the bottom row a case from the irregular category. On these scales, the irregular and nebular cases cannot be distinguished, but the nebular nature can be established by inspecting the images at larger scales. Credit: Suazo et al.

In the next post, I want to take a look at the results, which involve seven interesting candidates, all of them around a type of star I wouldn’t normally think of in Dyson sphere terms. The papers are Suazo et al., “Project Hephaistos – I. Upper limits on partial Dyson spheres in the Milky Way,” Vol. 512, Issue 2 (May 2022), 2988-3000 (abstract / preprint) and Suazo et al., “Project Hephaistos – II. Dyson sphere candidates from Gaia DR3, 2MASS, and WISE,” MNRAS (6 May 2024), stae1186 (abstract / preprint).