How it works
A swarm, not a shell. Freeman Dyson suggested in 1960 that a technical civilisation might end up using most of its star's light, and that we could look for the waste heat. A rigid shell around a star is unstable and no known material could hold it. A swarm of independent collectors, each on its own orbit, has neither problem. Here they fly in a few hundred inclined rings plus scattered orbits, each at the Keplerian speed for its distance.
The power does not depend on the radius. At distance r each square metre catches L☉ / 4πr² of sunlight, 1,361 W/m² at Earth. Cover a fraction f of the sky around the Sun and you catch f L☉, close or far. The radius decides how much material you need, since the area grows as r², and how hot the collectors run.
Where the material comes from. A full sphere at 1 AU is 2.8 × 10²³ m². At 1 kg/m² that is 2.8 × 10²³ kg, about 85 % of Mercury. Armstrong and Sandberg (2013) sketched how: a 1 km² seed of collectors powers mining and mass drivers on Mercury, which build more collectors, which power more mining. The swarm doubles every τ, so it looks like nothing for decades and then finishes in a few doublings.
How we would see one. Energy is conserved, so every watt the swarm catches is radiated again as heat, at the collectors' temperature: about 330 K at 1 AU, glowing near 10 µm. From outside, the star dims in visible light and gains an infrared excess. Project Hephaistos searched about five million stars in Gaia, 2MASS and WISE data this way and in 2024 reported 7 candidates, all M dwarfs. A follow up study argued that at least some are likely background galaxies.
Assumptions and sources
Model. Collectors are flat, black on both faces (absorptivity and emissivity 1), face the Sun and radiate from both faces, so T = (L☉ / 8πσSBr²)¼. All captured power, including what the civilisation uses, ends up as heat radiated by the swarm (Fig. 3). Orbits are circular; radiation pressure, shading between collectors and collisions are ignored. Electricity is shown at 30 % conversion, an estimate close to today's best space solar cells.
Construction. Growth is pure doubling from a 1 km² seed at the swarm radius, capped by the chosen coverage or by the mass of Mercury, whichever comes first. We assume all of Mercury's mass is usable, which is optimistic. The doubling time is the key unknown and every timeline here is an estimate. Lifting Mercury's mass out of its own gravity takes about 1.8 × 10³⁰ J (uniform sphere estimate), roughly 80 minutes of the Sun's total output.
Drawing. Orbits and radii are to scale. The Sun is drawn 13 times larger and the planets far larger still, so you can see them. Each drawn collector stands for a vast number of real ones (see the caption of Fig. 1). The drawn count follows the coverage, so an empty sky means under one part in 200,000 is built.
Sources. IAU 2015 Resolution B3: L☉ = 3.828 × 10²⁶ W, R☉ = 6.957 × 10⁸ m, Teff = 5,772 K. IAU 2012: 1 AU = 1.495978707 × 10¹¹ m. NASA planetary fact sheet: Mercury 3.301 × 10²³ kg, radius 2,439.7 km, 0.387 AU. Energy Institute Statistical Review of World Energy 2024: about 620 EJ of primary energy in 2023, 19.6 TW. Kopp and Lean 2011: solar constant 1,361 W/m². Dyson 1960, Science 131, 1667. Kardashev 1964, Soviet Astronomy 8, 217. Sagan 1973, The Cosmic Connection. Armstrong and Sandberg 2013, Acta Astronautica 89, 1. Suazo et al. 2024, MNRAS 531, 695 (Project Hephaistos II). Ren, Garrett and Siemion 2024 (arXiv:2405.14921), on background contamination of those candidates.
Made by airsup. More simulations at airsup.ai/simulations.