The Dyson Swarm

The Swarm view is the star, encircled. A live-shader Sun convects inside the Milky Way's cube sky while rings of hexagonal collectors, each plate standing for some 10¹³ real ones, assemble out of Mercury: a mass stream arcs from the shrinking planet to a glowing construction front, and pressing ▶ Build compresses the 31-year exponential into fifty seconds, two decades of nothing and then everything, exactly the ambush exponentials are. An infrared toggle re-lights the whole scene at the swarm's honest 331 K: the star curtained, the cloud glowing, the reason the searches look where they look.

The Fingerprint view is the search strategy as one picture: the νL_ν spectrum with the photosphere's hump sinking as the capture fraction grows and a second hump rising in the mid-infrared carrying exactly the captured share, twin peaks at f = ½. WISE's four bands and Kepler's window are shaded in, the four real searches, 1960, 2009, 2015, 2024, sit in cards beneath, and a violet note records the case of Tabby's Star: enormous dips, no waste heat, reading retired.

The Ledger view prices ambition. A Kardashev ladder climbs from humanity's 18.5 TW (K 0.73) past all sunlight on Earth (K 1.12) to the full Sun (K 2.06) and a galaxy of suns (K 3.2), your swarm's dot ascending it live; beside it the build-out curve runs its 58 doublings with Earth's actual photovoltaic pace, a doubling every ~3 years, drawn shallow beside it for honesty. Mass and count ladders at the bottom weigh the finished cloud against the Moon, Mercury and every grain of sand on Earth.

This instrument is the terminus of this site's SETI arc. The Drake Equation asks how many civilizations are out there; the Wow! Signal shows what a possible hail looks like; Tabby's Star shows what possible engineering looks like, and how it dissolved into dust under testing. The Dyson swarm is what that arc was always pointing at: not another signal or another dip, but what physics says we must see if anyone, anywhere, ever got seriously organized. It flips the search from listening to looking: it waits for no one to transmit, and simply checks the sky for heat that cannot be hidden.

It also caps the site's energy scale. The Laser Sail spends 100 gigawatts, humanity's boldest design, throwing 3.6 grams at Proxima; this page's dial turned to the stop reads 3.83×10²⁶ watts, four quadrillion times more, and it breaks no known physics. Permitted by engineering, absent from observation: that combination is the Fermi question at its sharpest. The instrument puts both halves on the table, the arithmetic of building and the blank sheet of searching, and then, in this site's tradition, goes quiet.

One measured number and three lines of arithmetic run the whole instrument; nothing else is asserted.

P = f·L☉ · A(t) = A₀·2^(t/196 d) · T = (L☉ / 8πσR²)^¼ · λ_peak = 2898 µm·K / T · K = (log₁₀P − 6) / 10

The energy budget is exact. The Sun's 3.828×10²⁶ W is the IAU nominal luminosity; captured power is that number times the covered fraction of the sphere at your chosen radius, and every comparison, the 18.5 TW of humanity, the 656,000 years per second, the Kardashev index, is division. The instrument adds no physics beyond bookkeeping, which is the point: the project never needed new physics.

The build is a labelled model. Armstrong & Sandberg's scheme, reinvested energy mining and launching Mercury with exponential feedback, is pinned by this instrument to a clean doubling every 196 days: one 1 km² seed to a spent planet in 31 years. Their published assumptions (1/3 collector efficiency, five-year processing surges, half the planet usable) are the provenance; the exponential arithmetic between the endpoints is exact; every derived readout says MODELLED.

The waste heat is thermodynamics, not speculation. A collector absorbing sunlight at radius R and radiating from both faces settles at T = (L☉/8πσR²)^¼: 331 K at 1 AU, 532 K at Mercury's orbit, 148 K at 5 AU, Wien peaks from 5 to 20 µm. The infrared view and the Fingerprint's second hump use these temperatures unmodified. Energy in equals energy out is the one law no engineering evades.

The searches are measured, and drawn to scale. Carrigan's IRAS census, the Ĝ galaxy survey and Project Hephaistos are plotted with their real sample sizes and real results, including the seven candidates nobody has closed. The instrument neither rounds the nulls up to "proof of absence" nor the candidates up to "detection"; both readings are left standing exactly as the data leaves them.

The evidence is tagged as you go. The bottom-left table is the conscience of the sim: the Sun's output and the survey results are MEASURED, the build pace and every temperature are MODELLED, and what the silence means is a READING. Rows flash as you touch the controls that depend on them, so the boundary between fact and model stays visible at all times.

The luminosity and the nulls are measured. The build-out and the temperatures are modelled, exactly, from stated assumptions. And whether the silence means absence or patience is the reading this site never supplies. The instrument computes; you conclude.

Four stations along the same arithmetic, so the scale of the thing has somewhere to land.

• Year 0 · the seed. One square kilometre of factory in orbit, indistinguishable from nothing. Captured power: a rounding error on a rounding error. Everything that follows is already latent in this frame plus one rule: each doubling pays for the next.
• Year 28 · the knee. The swarm crosses about one percent coverage, some 10⁵ times humanity's entire consumption, and for the first time the star's dimming would be measurable from another system. Twenty-eight years of apparent failure were the exponential working exactly as designed.
• Year 31 · Mercury spent. The full sphere at 1 AU: 85% of Mercury's mass airborne, 3.8×10²⁶ W captured, K 2.06, the visible star curtained and a 331 K infrared beacon in its place. The state every infrared survey since 1983 has been checking the sky for.
• 0.39 AU · the hot swarm. The same capture packed at Mercury's own orbit, where Armstrong & Sandberg actually put it: a sixth of the collector area and mass, panels running at 532 K, the fingerprint shifted to 5.4 µm. The dial every real search must guess, which is why survey bands are wide.

Every serious way the sky has been checked for this object, with who ran it and where it landed. One row is still open.

1. Feel the ambush. Reset to Year 0, press ▶ Build, and do not touch anything for the first thirty seconds. Nothing happens. Watch the strip chart instead: the line was climbing the whole time, in the only way exponentials ever climb, invisibly and then all at once.
2. Watch a star go out. With the build past year 29, drag the epoch slider slowly through the final two years and keep your eyes on the Sun. That two-year dimming, from a distance, is a civilization's entire construction era compressed into an eyeblink of stellar history.
3. See what WISE sees. At Year 31, flip the Infrared view on and off. Visible: a star-shaped absence. Infrared: the brightest object in the neighbourhood. That flip is the entire logic of forty years of searches in one toggle.
4. Grow the second hump. Open the Fingerprint, set the epoch to year 25, and play from there: the mid-IR hump rises through the WISE bands as the visible hump sinks, equal heights at half coverage. Then click the Tabby card and read why the one famous candidate failed exactly this chart.
5. Price the real world. In the Ledger, find the shallow dashed line: humanity's actual solar build-out, doubling every three years. Follow it to where it meets a spent Mercury: about 170 years. The gap between that line and the amber one is the entire distance between an economy and a self-replicating one.
6. End at the shortlist. Click the 2024 Hephaistos card in the Fingerprint view and read it as a skeptic, then as an optimist. Seven faint M dwarfs, probably contamination, formally open. Both readings are currently true, and holding both at once is what this site means by the periphery.

The honest line between what has been measured, what is modelled here, and what is a reading:

In one line: the Sun's 3.828×10²⁶ W, Mercury's mass, and every survey result, IRAS's empty census, Ĝ's hundred thousand quiet galaxies, Hephaistos's seven open candidates, Tabby's missing infrared excess, are measured; the 1 kg/m² collector, the 196-day doubling and every temperature on the panel are modelled, exact arithmetic on stated assumptions; and what forty years of silence means is a reading, the one this site leaves to you. The engineering is permitted; the telescopes have looked; the instrument keeps the two facts from being confused with each other.

• Dyson, F. J. (1960). Search for Artificial Stellar Sources of Infrared Radiation. Science 131 (3414), 1667-1668. The three-page letter: the energy argument, the waste-heat signature, and the proposal to search for infrared point sources.
• Dyson, F. J. (1960). Letters in response, Science 132. The correction on record: "a loose collection or swarm of objects traveling on independent orbits," not a rigid shell.
• Stapledon, O. (1937). Star Maker. The novel Dyson credited as the source of the idea.
• Kardashev, N. S. (1964). Transmission of Information by Extraterrestrial Civilizations. Soviet Astronomy 8, 217. The Type I / II / III energy scale this instrument's ladder is built on.
• Sagan, C. (1973). The Cosmic Connection. The interpolated K-index formula, K = (log₁₀P − 6)/10, used by the readouts.
• Armstrong, S., & Sandberg, A. (2013). Eternity in Six Hours: Intergalactic spreading of intelligent life and sharpening the Fermi paradox. Acta Astronautica 89, 1-13; with Armstrong's 2012 Oxford presentation of the Mercury-disassembly scheme. The ≈ 31-year exponential feedback build the instrument's timeline is pinned to.
• Carrigan, R. A. (2009). IRAS-based Whole-Sky Upper Limit on Dyson Spheres. Astrophysical Journal 698, 2075-2086. The first serious census: ≈ 250,000 sources, no convincing sphere.
• Griffith, R. L., Wright, J. T., et al. (2015). The Ĝ Infrared Search for Extraterrestrial Civilizations with Large Energy Supplies. III. ApJS 217, 25. 100,000 resolved WISE galaxies; none consistent with ≥ 85% starlight reprocessing.
• Suazo, M., Zackrisson, E., et al. (2024). Project Hephaistos II. Dyson sphere candidates from Gaia DR3, 2MASS, and WISE. MNRAS 531, 695. Five million stars filtered to seven unresolved M-dwarf candidates; subsequent work argues for background-galaxy contamination.
• Boyajian, T. S., et al. (2016). Planet Hunters IX. KIC 8462852, Where's the Flux? MNRAS 457, 3988; and Meng, H. Y. A., et al. (2018), ApJ 853, 1. The dips, and the chromatic, dust-like extinction with no infrared excess that retired the swarm reading.