Dyson Swarms

A Dyson swarm is a large number of independent collectors in orbit around a star, intercepting a substantial fraction of its output. It is the survivable member of a family that also contains the solid shell of popular imagination, which does not survive at all. The framing under test is that stellar-scale energy capture is an engineering problem.

The first thing to say is that Freeman Dyson did not propose building one. His 1960 paper in Science — Search for Artificial Stellar Sources of Infrared Radiation — is an observing proposal. Its argument is thermodynamic: any civilisation intercepting starlight must re-radiate the energy as waste heat at a low temperature, so it would appear as an infrared source at a few hundred kelvin, and that is something a telescope can look for. The Dyson sphere enters the scientific literature as a search target, and almost everything that has actually happened in the subject since has happened on the search side.

That is why this page is organised around instruments. The construction literature is thin, largely negative, and rests on a premise — self-replicating industry — that nobody owns and nobody has demonstrated. The observational literature is substantial, current, quantitative and, unusually for a speculative subject, resolved: a candidate list published in 2024 was worked to a conclusion by radio interferometry and by JWST within two years, and the conclusion was background galaxies.

The flag distribution on this page is therefore bimodal on purpose. The nulls carry established; the construction scenarios carry speculative and handwave. That is not a rhetorical device, it is what the evidence looks like. Stellar Engineering owns anything that acts on the star itself; this brief owns collectors in orbit around it, and Planetary Scale Energy Systems owns the space-based-solar-power economics that this page deliberately does not restate.

Established Dyson's actual proposal, with its numbers, because they are more specific than the popular version. A mass equal to Jupiter's — 2 × 10³⁰ grams — distributed as a spherical shell at twice Earth's orbital distance gives a surface density of 200 g/cm². Such a structure radiates at roughly 200–300 K, peaking near 10 micrometres, which sits inside the 8–12 micrometre atmospheric transparency window — and that is the entire point of the paper. Dyson's own phrase is “an artificial biosphere which completely surrounds its parent star.”

Established His stated limitation was a search limitation, not an engineering one: “A scan of the entire sky for objects down to the 5th or 6th magnitude would be desirable, but is probably beyond the capability of existing techniques of detection.” That sentence is the brief's organising fact. Dyson identified the instrument gap in 1960 and the subject's whole subsequent progress consists of closing it — IRAS, WISE, Gaia, e-VLBI, JWST — while the construction question stood still.

Speculative The solid shell is dead, and the number that kills it is better than the argument usually given. Jason Wright's Dyson spheres (arXiv:2006.16734, 2020), the standard modern theoretical review, disposes of it three ways. Gravitationally, by Gauss's law the field inside a uniform shell is zero, so the star exerts no restoring force and the shell is only neutrally stable — it drifts and must be actively station-kept. Radiation pressure does not help, because “photons are contained within the Dyson sphere” and the net force is zero. And mechanically: to avoid catastrophic buckling the shell material must have an elastic modulus of at least 10¹³ GPa, which exceeds the modulus of the strongest known material, carbyne, by nine orders of magnitude. Nine orders is a citable refutation; “it would be unstable” is not.

Speculative Wright's own definitional move is worth adopting because it explains the shape of this page. He uses “Dyson sphere” for “any collection of artificial material around a star that produces significant amounts of waste heat, regardless of its specific geometry.” The swarm-versus-shell distinction matters enormously for engineering and not at all for detection. Since the evidence in this subject is detection evidence, the geometry debate is downstream of everything measurable.

Speculative The most fully worked construction estimate retrieved is a disassembly scenario, and its arithmetic contains a surprise. Armstrong and Sandberg, Eternity in six hours (Acta Astronautica 89:1–13, 2013, Future of Humanity Institute): take Mercury's mass of 3.3022 × 10²³ kg, assume 50% convertible to reflective surface, place the swarm at Mercury's semi-major axis of 5.79 × 10¹⁰ m for an area of 4.21 × 10²² m², at a reflector areal density of 3.92 kg/m² — a thickness of half a millimetre. Mercury's gravitational binding energy is 1.80 × 10³⁰ J. With exponentially self-replicating mining, “Mercury itself will be completely disassembled in 31 years and 85 days, with most of the mass loss happening in the last 4 years” — a figure that is the output of an assumed doubling model and should never be repeated as a timeline.

Speculative Here is the surprise: unbinding Mercury costs about 1.3 hours of the Sun's entire output. The binding energy is 1.80 × 10³⁰ J and the Sun emits 3.8 × 10²⁶ W, so the whole disassembly is roughly 4,700 seconds of solar luminosity. The energy is trivial once you have a swarm, and the swarm is what you were trying to build. The binding constraint in every construction scenario is therefore neither energy nor material quantity. It is the self-replicating industrial base, which is assumed rather than derived, and which no brief on this map owns because nothing has ever demonstrated a partial version of it.

Speculative A cross-source check that turns an architectural quibble into a real fork. Armstrong and Sandberg's collectors are 3.92 kg/m². McInnes's 2026 stability paper gives the areal density at which solar radiation pressure exactly balances solar gravity as about 1.5 × 10⁻³ kg/m². The swarm elements are therefore roughly 2,600 times too heavy to levitate and must be in genuine orbits. Swarm and statite-bubble are not two words for the same thing; they are separated by three and a half orders of magnitude in areal density.

Speculative Wright's independent scaling adds the design-space constraint that most treatments miss. At a critical surface density of 0.8 g/cm², one Jupiter mass gives a shell at 30 au radiating at 71 K; with metamaterials, one Jupiter mass reaches 9,000 au. And the thermodynamic warning: “for every additional 'nine' of efficiency, the Dyson sphere requires three orders of magnitude more radiating material.” His Carnot bound is efficiency below 1 minus the ratio of emission to stellar temperature, giving a conservative 99.9% ceiling for the solar case — but “unless a technological species somehow incurs negligible cost when adding material to its Dyson sphere, the optimum size will be significantly smaller than its theoretical limit.” Economics sets the radius, not thermodynamics.

Established Now the part of the subject that has actually produced results. The first whole-sky limit is Carrigan's, from IRAS. IRAS-based whole-sky upper limit on Dyson spheres (arXiv:0811.2376; ApJ, 2009) searched 250,000 IRAS sources covering 96% of the sky for blackbody spectra between 100 and 600 K, for both complete and partial spheres. Sensitivity reached solar-luminosity objects to 300 parsecs, a volume containing roughly a million solar-type stars. Very few candidates survived filtering for stellar mimics, and those that did were ambiguous.

Established The strongest single result in this cluster is at galaxy scale. Griffith, Wright, Maldonado, Povich, Sigurdsson and Mullan, The G-hat Infrared Search for Extraterrestrial Civilizations with Large Energy Supplies. III. The Reddest Extended Sources in WISE (ApJS 217:25, 15 April 2015), examined about 100,000 galaxies — “1,000 times more galaxies than the only previous such search” — with 563 extended red sources surviving visual vetting. The published limit is flat: “No galaxies in our sample host an alien civilization reprocessing more than 85% of its starlight into the MIR.” Only 50 galaxies show mid-infrared luminosities consistent with more than 50% reprocessing, and most of those have ordinary star-formation explanations. Whatever anyone has built anywhere in a hundred thousand galaxies, nobody has enclosed more than 85% of a galaxy's starlight.

Established At stellar scale the numbers come from Project Hephaistos, and they are the ones to quote. Suazo, Zackrisson and colleagues, Project Hephaistos I. Upper limits on partial Dyson spheres in the Milky Way (arXiv:2201.11123; MNRAS 512:2988, 2022), searched for structures absorbing 1–90% of bolometric luminosity with waste heat between 100 and 1000 K. For roughly 2.7 × 10⁵ stars within 100 parsecs at 90% completion and about 300 K, the fraction is below about 2 × 10⁻⁵. For about 2.9 × 10⁸ stars within 5 kiloparsecs, the limit is 8 × 10⁻⁴. The authors are explicit that constraints weaken sharply for incomplete structures, because a partial sphere is hard to distinguish from ordinary infrared-bright astrophysics, and that caveat travels with the numbers.

Established Then the candidate list, which is where most popular accounts of this subject stop. Suazo, Zackrisson, Mahto, Lundell, Nettelblad, Korn, Wright and Majumdar, Project Hephaistos II. Dyson sphere candidates from Gaia DR3, 2MASS and WISE (arXiv:2405.02927; MNRAS 531:695, 2024), ran a pipeline over five million objects, using a convolutional neural network to reject WISE source confusion, and produced seven candidates, all M dwarfs, all with infrared excess that “standard astrophysical explanations cannot readily account for.” The authors described them as requiring further investigation rather than as detections.

Established The further investigation happened, and it is the most complete worked example on this site of a technosignature candidate list being closed by instruments. Ren, Garrett and Siemion, Background Contamination of the Project Hephaistos Dyson Spheres Candidates (arXiv:2405.14921, 23 May 2024), found candidates A, B and G “misidentified as DS candidates” because of dust-obscured galaxies close to the line of sight, and concluded on sky-density statistics that “Hot DOGs can probably account for the contamination of all 7 DS candidates.”

Established Candidate G was then resolved directly. Ren, Garrett and Siemion, High-resolution imaging of the radio source associated with Project Hephaistos Dyson Sphere Candidate G (MNRAS Letters 538:L56–L61, 2025): European VLBI Network e-VLBI at 5 GHz detected a compact source at 13.69 ± 0.12 mJy in a beam of 12.7 × 6.6 milliarcseconds with a brightness temperature above 10⁸ K — an active galactic nucleus. e-MERLIN at 1.5 GHz resolved three components across about 0.8 arcseconds. There was no radio emission at the M dwarf's position at all. The conclusion: “the MIR excess previously attributed to a potential Dyson Sphere is instead the result of contamination by a background AGN.”

Established And in July 2026 the search team resolved two of its own candidates with JWST. Zackrisson and colleagues, Project Hephaistos IV. James Webb Space Telescope Observations of Two Dyson Sphere Candidates (arXiv:2607.09460, 10 July 2026), imaged and took spectra of candidates D and E with MIRI. Candidate D is an unresolved background galaxy at redshift about 0.9, spectrally a Hot Dust Obscured Galaxy. Candidate E is a background galaxy at redshift about 0.4 with extended morphology and bright knots — a dusty starburst. The mechanism is source confusion: WISE resolution is 6–12 arcseconds and the contaminating galaxies sit within about 1 arcsecond of the M dwarfs, which MIRI separates cleanly. The paper's conclusion: “The infrared excess does not originate from Dysonian megastructures… but from background galaxies.” The search team publishing the refutation of its own candidate list is interest against the finding in the strongest form available.

Frontier Boyajian's Star is the other case that was worked to a conclusion, and it is a methodological exhibit rather than a megastructure result. Boyajian and colleagues, Planet Hunters IX. KIC 8462852 — Where's the flux? (arXiv:1509.03622; MNRAS, 2016), reported irregular aperiodic dips of up to about 20% lasting 5 to 80 days, confirmed astrophysical rather than instrumental, in an ordinary F3 V main-sequence star with no significant infrared excess and no close stellar companion. The leading natural hypothesis was a family of exocomets or planetesimals from a breakup event, with a minimum total mass above 10⁻⁶ Earth masses, implying an original body over 100 km across.

Speculative The megastructure reading was then stated precisely enough to be testable, which is the important part. Wright, Cartier, Zhao, Jontof-Hutter and Ford, The G-hat Search… IV. The Signatures and Information Content of Transiting Megastructures (arXiv:1510.04606, 2015), enumerated ten ways anomalous silhouettes, orbits and transmission properties would distinguish artificial occulters from exoplanets, and said of KIC 8462852 that it was “an object with a bizarre light curve consistent with a 'swarm' of megastructures… an outstanding SETI target.” And then it was tested, and the answer was dust. Boyajian and 369 co-authors, The First Post-Kepler Brightness Dips of KIC 8462852 (ApJL, 2 January 2018), caught four dips — Elsie, Celeste, Skara Brae and Angkor — at 1–2.5% depth, and found that “multiband photometry of the dips shows differential reddening favoring non-grey extinction.” Wavelength-dependent dimming is inconsistent with an opaque macroscopic occulter and consistent with “an occulter consisting primarily of ordinary dust, where much of the material must be optically thin with a size scale much less than 1 micrometre.” The same paper is careful that this does not constrain the star's separate long-term secular dimming, which remains unexplained.

Frontier The one positive engineering theorem in the retrieved record is nine months old. Colin McInnes of the James Watt School of Engineering, Glasgow, Stellar engines and Dyson bubbles can be stable (MNRAS 546:1–18, accepted 13 January 2026; arXiv:2603.00203), is a proper stability analysis by a solar-sail dynamicist rather than a scaling argument. For the Dyson-bubble architecture — the swarm's levitating cousin — a single static reflector above a star is unstable, but a dense cloud of reflectors with optical depth is passively self-stabilising, because light pressure falls off faster than gravity as you move outward through the cloud. The critical areal density for levitation in the solar case is about 1.5 × 10⁻³ kg/m², scaling with stellar mass to a power near 2.5. McInnes states his own limitations plainly: rigid perfectly reflecting discs, internal mechanical stresses treated only in an appendix, a simplified model of diffuse scattered radiation, and gravitational interaction between reflectors neglected until self-gravity is added explicitly. This is the one place where the “engineering problem” framing gets genuine support — a peer-reviewed paper moving a specific architectural question from “presumably unstable” to “stable under stated conditions” — and it is one paper.

Established Position one: the radiative physics is settled and it is what makes the subject scientific at all. A structure intercepting starlight must re-radiate at low temperature; a Jupiter-mass shell at 2 au sits at 200–300 K peaking near 10 micrometres; that is inside the atmospheric window. Nothing about that argument depends on the structure being possible, which is why the search survives the engineering's collapse.

Established Position two: Dyson meant a biosphere or swarm, not a solid shell. His own text says “artificial biosphere”, and the annotators of the reproduction make the point explicitly. Position three: the solid shell is mechanically impossible, needing an elastic modulus nine orders of magnitude beyond carbyne. Held by Wright (2020), and it is an engineering statement about a hypothetical rather than a measurement, which is why it carries the weaker flag despite being the strongest argument in the construction literature.

Speculative Position four: the optimal swarm radius is set by economics rather than by thermodynamics, and each additional “nine” of efficiency costs a thousandfold more radiating material. Wright again. It is the most useful design-space constraint in the subject and it has never been tested against a worked design. Position five: half of Mercury gives 4.21 × 10²² m² of half-millimetre reflector, from Armstrong and Sandberg — a mass budget that is arithmetically fine and rests on an industrial premise that is not.

Speculative Position six, derived here: swarm elements at about 3.9 kg/m² are roughly 2,600 times too heavy to levitate and must be in genuine orbits. Combining two published figures from different authors. Position seven: a dense Dyson bubble is passively self-stabilising while a single statite is not — McInnes, MNRAS 546 (2026), with the author's own limitations attached. These two positions describe genuinely different machines and the literature routinely conflates them.

Established Position eight is a measurement: no galaxy in about 100,000 reprocesses more than 85% of its starlight into the mid-infrared. Position nine is another: fewer than about 2 × 10⁻⁵ of stars within 100 parsecs host a 90%-complete structure near 300 K, and fewer than 8 × 10⁻⁴ within 5 kiloparsecs. Position ten is the whole-sky version: IRAS constrains 250,000 sources to 300 parsecs, and the survivors are ambiguous. Three independent instruments, three published numbers, three nulls.

Established Position eleven: seven M-dwarf candidates from five million objects had unexplained infrared excess. True as stated in 2024 and no longer the current state. Position twelve: all seven are plausibly Hot DOG contamination, from an independent group on sky-density statistics. Position thirteen: candidate G is a background AGN, resolved at milliarcsecond scale with nothing at the star's position. Position fourteen: candidates D and E are background galaxies at redshift 0.9 and 0.4, resolved by JWST/MIRI in July 2026 by the original search team. Reporting the seven as live candidates in 2026 is out of date, and this is the single most common error about the subject.

Speculative Position fifteen: KIC 8462852's dips were consistent with a swarm of megastructures and it was an outstanding SETI target. Stated by Wright and colleagues in 2015 with ten specific discriminants — and therefore falsifiable, which is the point. Position sixteen: the dips are wavelength-dependent, ruling out an opaque occulter and favouring sub-micron dust. Boyajian and 369 co-authors, 2018. A speculative hypothesis was stated precisely, tested and defeated inside three years. That is what the subject looks like when it is done properly.

Speculative Position seventeen: a Starlink-scale planetary megastructure at 280 K would be detectable by VLTI to about 260 light years. Osmanov (arXiv:2103.07227, 2021) extrapolates a construction mass near 10¹⁷ g of graphene — seven orders below Earth's total carbon inventory — radiating at 280 K with a 10 micrometre peak, potentially resolvable within a volume containing about a thousand solar-type stars, and by FAST at radio to about 160 parsecs. The same paper asserts a current human energy index of 0.7 on the Kardashev scale and derives a thousand-year Type-I transition from a 2.1% growth rate. Planetary Scale Energy Systems established that no obtainable source supports any specific Kardashev figure for humanity. Carry the detectability numbers; do not carry the 0.7.

Speculative Position eighteen: a helical satellite orbit around a Dyson ring is stable under the combined gravity of ring and star. Raval (arXiv:2407.10624, 2024), with a successful simulation reported. The abstract's numerical results could not be retrieved for this brief, so the existence of the claim is carried and no figure from it appears. Position nineteen, and the most interesting unworked question in the whole cluster: a Dyson swarm would change the star it encloses, and nobody has studied it. Wright calls it “an area of study ripe for investment”. Stars have negative heat capacity, so returning energy to one makes it expand, cool and dim. This brief names it and hands it to Stellar Engineering, which owns anything acting on the star.

Speculative The binding bottleneck is not energy, not material quantity and not orbital mechanics. It is self-replicating industry, and it is smuggled in as a premise. Unbinding Mercury costs about 1.3 hours of the Sun's total output. Half of Mercury supplies 4.21 × 10²² m² of reflector. Every construction scenario retrieved for this brief closes the gap between “we have a planet” and “we have 10²² square metres of half-millimetre film in orbit” by assuming exponentially doubling autonomous manufacturing. Nothing demonstrates it, nothing partially demonstrates it, and no brief on this map claims to own it — not Space-Based Manufacturing, whose product comes home, and not Orbital Shipyards, whose largest manifested orbital assembly demonstration is 1.4 metres across.

Speculative The second is the areal-density fork, and it decides which machine you are building. At 3.9 kg/m² the elements orbit; at 1.5 g/m² they levitate. Those differ by a factor of about 2,600 and imply completely different structures, station-keeping regimes and failure modes. McInnes's stability result applies to the light case and Armstrong and Sandberg's mass budget to the heavy one, and no retrieved source reconciles them.

Speculative The third is thermodynamic and it caps ambition rather than blocking it. Efficiency is Carnot-limited by the ratio of emission temperature to stellar temperature, with a conservative ceiling near 99.9%; but each additional nine of efficiency costs three orders of magnitude more radiating material. The practical consequence is that a real swarm would be much smaller than its theoretical limit, which in turn makes it harder to detect than the idealised object the searches are tuned for.

Frontier The fourth is a detection bottleneck and it is the one that actually binds current work. Hephaistos I says explicitly that constraints weaken sharply for incomplete structures, because partial reprocessing is confusable with ordinary infrared-bright astrophysics. The Hephaistos II candidates failed for exactly that reason at a coarser scale: WISE's 6–12 arcsecond resolution against contaminants within 1 arcsecond. The upper limits are strong for complete structures and weak for partial ones, and a real swarm built to an economic optimum would be partial.

Handwave And a fifth that is a gap rather than an obstacle: nobody knows what a swarm does to its star. Wright names the problem and no literature addresses it. If the radiative feedback is significant then every luminosity assumption in every construction scenario is wrong in an unknown direction. This brief states it, flags it at the weakest level, and hands it to Stellar Engineering.

Established The adjudicated dependency is Planetary Scale Energy Systems, and the seam is drawn tightly. That brief owns terrestrial and near-Earth energy including all space-based-solar-power economics — the cost studies, the levelised-cost figures, the launch share of lifecycle cost, the beaming efficiencies, and the critique of the Kardashev scale itself. None of that is restated here. It hands this brief swarm architecture, construction, material budgets, orbital mechanics and stability, and that is exactly what sections 2 and 4 cover. In one sentence: FR-I-23 treats the Dyson swarm as the asymptote of an economic argument and a detectability programme and then stops; this brief starts where the structure itself becomes the subject.

Speculative The second dependency is unowned, and saying so is more useful than assigning it. Self-replicating manufacturing is the premise every construction scenario rests on. It is not owned by Space-Based Manufacturing, which covers making things in orbit to sell on Earth, nor by Orbital Shipyards, which covers assembling structures from launched components, nor by Asteroid Mining, which covers extraction. It is an unowned premise, and a brief that assigned it to a neighbour would be laundering a hand-wave through a hyperlink.

Established The third is instrumental, and unusually it has already been supplied. Everything measurable in this subject came from general-purpose astronomy: IRAS, WISE, Gaia DR3, 2MASS, the European VLBI Network, e-MERLIN and JWST/MIRI. Not one of those instruments was built to look for megastructures, which is exactly why their nulls carry weight. The subject's dependency on instrumentation is real and is being met by facilities with entirely different primary missions.

Frontier And one dependency running outward rather than inward. Stellar Engineering depends on this brief, because a stellar engine is a special case of a structure in orbit around a star and shares its stability mathematics — McInnes's 2026 paper treats both in the same framework. This brief owns the Dyson-bubble half of that result; the Shkadov-mirror half belongs there.

Established The experiments in this subject are observations, and the most informative one has already been run twice on the same objects. The Hephaistos candidate list is the model: publish a pipeline, publish its output as candidates rather than detections, and then let independent groups and better instruments resolve them. Radio interferometry closed candidate G in 2025; JWST/MIRI closed D and E in July 2026. The whole cycle took twenty-six months.

Established The obvious remaining work is the other four candidates. A, B and G have line-of-sight galaxy explanations; D and E are resolved; C and F have a statistical explanation from Hot DOG sky densities and no direct imaging retrieved for this brief. MIRI observations of the remainder would close the list completely, and on the evidence of D and E the expected result is more background galaxies. Also missing: Hephaistos III was not located in this research pass. The series as retrieved runs I, II and IV, and this brief does not assume what III contains.

Established The methodological experiment worth transferring off this page is the wavelength test. An opaque macroscopic occulter blocks all colours equally; dust does not. Multiband photometry of the KIC 8462852 dips found differential reddening, and that single measurement settled a hypothesis that had generated worldwide coverage. It is the cheapest discriminator in technosignature work and it applies to any future transit anomaly.

Speculative The transit-shape route is the other detection channel and it is under-exploited. Wright and colleagues enumerated ten signatures that would distinguish artificial occulters from planets — anomalous silhouettes, non-Keplerian orbits, unusual transmission properties. With current transit photometry covering millions of stars, a systematic search for those signatures is a data-analysis project rather than a new facility, and no retrieved source reports one having been run at scale.

Speculative And two experiments that would bear on construction rather than detection, neither of which anyone is doing. First, an areal-density demonstration: solar sails already fly at grams per square metre, and Solar Sail Systems owns that record — a statite holding station against solar gravity would be the first physical instance of McInnes's regime. Second, and the one that would matter most, any demonstration at all of self-replicating manufacture at any scale. There is no experiment in the retrieved record that bears on it, which is why the construction side of this subject cannot be flagged above speculative.

Speculative The engineering requirements can be stated as a small set of numbers, and each one comes from a different paper because no single design exists. Collector areal density: 3.92 kg/m² for an orbiting swarm, or about 1.5 × 10⁻³ kg/m² for a levitating bubble. Total area for full interception at Mercury's orbit: 4.21 × 10²² m². Radiating area sufficient to dump the waste heat at the chosen emission temperature, which grows by three orders of magnitude per additional nine of conversion efficiency.

Speculative The station-keeping requirement is where the two architectures diverge completely. A heavy swarm is a very large number of ordinary orbits and its engineering problem is collision management and phasing across 10²² square metres of surface. A light bubble is a levitation problem, and McInnes's result says a dense enough cloud is passively self-stabilising while a single element is not — which converts an active control problem into a configuration requirement, and is the single most useful engineering statement in the subject.

Handwave The manufacturing requirement is where the specification stops being engineering. Producing 4.21 × 10²² m² of half-millimetre reflective film requires a production system with no demonstrated ancestor. The retrieved literature closes this with an assumed doubling rate and a 31-year figure, and a number produced by choosing a doubling rate is not an engineering estimate, it is a restatement of the assumption.

Established And one requirement that can be stated with certainty because it is the thing that has been ruled out. A rigid enclosing shell requires an elastic modulus of at least 10¹³ GPa. Carbyne, the stiffest known material, is nine orders of magnitude short. No engineering requirement in this brief is a target anyone could work toward; this one is a target nobody can work toward, and that distinction is worth preserving.

Frontier The tightest adjacency is Stellar Engineering, and the line between them is the star. This brief owns anything in orbit around a star — swarm architecture, collector mass budgets, orbital mechanics, the shell-instability result, the waste-heat searches and the Hephaistos candidate list with its resolution. That brief owns anything that acts on the star: moving it, lifting mass from it, mixing it, igniting a planet into one. The two share exactly one paper by design — McInnes (2026), which treats Dyson bubbles and stellar engines in a single stability framework — and each cites it for its own half.

Established Against Interstellar Archaeology the division is purpose, not dataset. That brief owns technosignature searching as a research programme — its instruments, its methodology, its nulls across radio and optical, its treatment of interstellar objects and exoplanet atmospheric pollutants. This brief owns the same instruments read as measurements about swarms: what a 2 × 10⁻⁵ upper limit says about swarm abundance, what an 85%-reprocessing limit says about galactic-scale capture, and what JWST resolving candidates D and E says about the candidate class. Both briefs cite Hephaistos; they should not cite it for the same claim.

Established Against Planetary Scale Energy Systems the division is scale and subject. That brief owns space-based solar power as an economic proposition on Earth's energy system and owns the Kardashev critique. This brief begins where the collector array stops being an energy market participant and starts being a structure. No cost figure, levelised or otherwise, appears anywhere on this page.

Frontier The genuinely useful technical adjacency is solar sailing. Solar Sail Systems owns flight-proven areal densities in the grams-per-square-metre range, which is the regime McInnes's levitation criterion lives in; a statite is a sail that does not go anywhere. Space-Based Solar Power owns the collection-and-transmission chain at the only scale anyone has costed. And Orbital Shipyards and Asteroid Mining own the assembly and feedstock capabilities that a swarm would need at a scale neither brief documents.

Established The institutional finding here is the reverse of the usual one in this category: the search is a functioning scientific enterprise and the engineering is not. The search has peer-reviewed papers in ApJ, ApJS, MNRAS and MNRAS Letters; competitively awarded time on JWST and on the European VLBI Network; a named multi-institution programme; independent groups checking each other; and published numerical upper limits. The construction literature has a review, a mass budget from a philosophy-adjacent institute, and one stability theorem from January 2026.

Established The behaviour of the search teams is the transferable institutional exhibit. Hephaistos II described seven objects as candidates and said explicitly that standard explanations could not readily account for them — not that they were detections. An independent group published a contamination analysis three weeks later. The original team then obtained JWST time and published the resolution of its own candidates. A field that publishes its candidate list, has it attacked, and then does the work of killing it itself is behaving exactly as it should, and this is worth naming because the popular coverage of the same events did none of those things.

Frontier The institutional weakness is on the construction side and it is structural rather than cultural. There is no funder for whom a Dyson-swarm engineering study is a deliverable. The nearest thing retrieved is a single university dynamicist applying solar-sail mathematics to an old question and publishing it in a mainstream astronomy journal. That is how the subject's one positive theorem arrived: not through a programme, but because a specialist in an adjacent field found the question tractable.

Established And one institutional caution about the record itself. NASA's Astrophysics Data System was unreachable throughout the research pass behind this brief, and several bibliographic records here therefore come from arXiv and publisher pages. Hephaistos III was not located at all, and this brief does not speculate about its contents. Saying what could not be reached is part of stating what is known.

Frontier The live ethical question in this subject is about announcement, and it has a worked example. A candidate list of seven objects, published responsibly with the authors' own caveats, produced global coverage describing possible alien megastructures. The resolutions — a background AGN, two background galaxies, a statistical account of all seven — received a fraction of that attention. The asymmetry is not the researchers' failure; every step of their behaviour was correct. It is a property of the reporting environment, and it is the reason a brief like this one has to state the current position rather than the memorable one.

Established Second, the ethics of null results, which run in the opposite direction to intuition. The upper limits in section 2 are the subject's most valuable products, and they exist only because instruments and observing time were spent looking for something almost certainly not there. A posture that treats megastructure searching as unserious gets no upper limits, and then the claims persist unrefuted — which is exactly what has happened on the construction side, where nobody has bothered to look and nothing is settled.

Speculative Third, and only conditionally: a detection would be the largest single fact in the history of the species, and there is no protocol in the retrieved record for a megastructure detection specifically. The searches operate under general astronomical publication norms, which worked adequately for a candidate list and would be an odd fit for something that survived follow-up.

Handwave Fourth, the far-side consideration, carried because the directive is breadth and not because it bears on anything now. Every construction scenario in this brief involves disassembling a planet. Mercury has no biosphere and no plausible claimant, so the question is not obviously a moral one — but it is the only case on this map where a proposed engineering programme consumes an entire world, and the literature treats it purely as a mass budget. That is worth noticing even though nothing turns on it for centuries.

Established The Dyson swarm's civilizational role is as a benchmark rather than a plan, and the benchmark is doing real work. It defines the Kardashev Type II energy scale, it gives technosignature searching a specific and thermodynamically motivated target, and it makes an otherwise unfalsifiable question about advanced civilisations into a photometry problem. That is a substantial contribution from a structure nobody can build.

Frontier The most interesting civilizational content is a tension between the framing and the data, and it should be stated as a tension rather than resolved. The engineering-problem framing predicts abundance: if stellar-scale capture is merely difficult, a galaxy of 10¹¹ stars over 10¹⁰ years should contain many instances. The observations set upper limits of 2 × 10⁻⁵ within 100 parsecs, 8 × 10⁻⁴ within 5 kiloparsecs, and no galaxy in 100,000 above 85% reprocessing. Those are in genuine tension, and the honest reading is that either the framing is wrong, or the structures are not built, or they are built in a form the searches are not tuned for — and Hephaistos I's own caveat about incomplete structures keeps that third option open.

Speculative The economics argument cuts against the popular image in a way that matters for what to look for. Wright's point that the optimum size is set by cost rather than by the Carnot limit implies that a real swarm would be far from complete — and partial swarms are precisely the case where the published constraints weaken sharply. If the reasoning is right, the searches are best at detecting the objects least likely to exist.

Established And the durable lesson is methodological rather than astronomical. A speculative hypothesis about KIC 8462852 was stated precisely enough to make a prediction, the prediction was tested with multiband photometry, and it failed. A candidate list was published with caveats, attacked by an independent group, and closed by its own authors using a better telescope. This is the clearest demonstration on the site that a subject can be simultaneously wildly speculative and rigorously scientific, and that the difference lies entirely in whether the claims are made testable.

These horizons track instruments and survey volumes, because that is the only part of the subject that moves:

• 10 yr: Established Expect the remaining Hephaistos candidates to be resolved, most likely as further background contamination, and expect upper limits to tighten as Gaia's final data releases and further infrared surveys arrive. Frontier Expect at least one new candidate list from a new survey, one round of coverage treating it as a detection, and one resolution paper. Speculative Expect nothing whatsoever on the construction side beyond further single-author theory papers of the McInnes kind.
• 25 yr: Frontier The interesting development at this horizon is a systematic transit-morphology search for the ten megastructure signatures across the full photometric archive, which is a data-analysis project rather than a facility. Speculative A statite demonstration — a solar sail holding station against solar gravity rather than travelling — is the one physical experiment in this brief that a real programme might fly, and it would be the first hardware bearing on any of it. Handwave Anyone forecasting swarm construction at this horizon is forecasting self-replicating industry, which has no demonstrated ancestor.
• 50 yr: Speculative If the upper limits keep tightening, the negative result becomes a genuine constraint on the abundance of advanced civilisations rather than a curiosity, and that is a real scientific outcome that requires no megastructure to exist. Speculative The construction question at this horizon is entirely downstream of autonomous manufacturing, and this brief declines to forecast it because nothing measured bears on it.
• 100 / 250+ yr: Handwave Beyond forecasting for construction. Speculative The defensible structural statement is that a Dyson swarm has no known physical obstacle and one known industrial one, and that the industrial one — self-replicating manufacture — is the same premise that carries most of the far-future literature on this map. Handwave A subject whose entire construction case rests on one unowned assumption has no meaningful base rate at this horizon.

• Depends on One edge, and it is an economic and energetic one rather than a structural one. Planetary Scale Energy Systems owns the whole space-based-solar-power case — the cost studies, the levelised-cost figures, the beaming efficiencies and the critique of the Kardashev scale — and this brief restates none of it. What it hands on is exactly what sections 2 and 4 cover: swarm architecture, the material budget, orbital mechanics and stability. The seam in one sentence: FR-I-23 treats the Dyson swarm as the asymptote of an economic argument and a detectability programme and then stops; this brief starts where the structure itself becomes the subject, and inherits from it one specific unstudied question — what an enclosing structure does to the star inside it — which is passed straight on to Stellar Engineering without being pursued here.
• Requires (not on this map) Two constraints that are not briefs on this map. The first is the premise every construction scenario rests on and nobody owns: exponentially self-replicating manufacture, which converts a planet's mass into 10²² square metres of film and supplies the only timeline anyone quotes. There is no demonstration, no partial demonstration and no experiment bearing on it, which is why the construction half of this brief cannot be flagged above speculative. The second is a scientific result nobody is producing: stars have negative heat capacity, so an enclosing structure that returns energy inward should make its star expand, cool and dim — and Wright's review states plainly that the effect has never been studied.
• Enables One edge outward: Stellar Engineering depends on this brief, because everything it describes is a special case of a structure held in position around a star and shares the same stability mathematics. McInnes's 2026 result treats Dyson bubbles and Shkadov mirrors in a single framework, and this brief owns the bubble half. No enabling edge is claimed to any energy or industrial brief, because a capability that has produced one positive theorem in sixty-six years and rests on an undemonstrated manufacturing premise enables nothing that can be recorded as a dependency.
• Adjacent Interstellar Archaeology owns technosignature searching as a research programme; this brief owns the same instruments read as measurements about swarms, and the division is purpose rather than dataset. Solar Sail Systems supplies the only flight-proven hardware in the areal-density regime McInnes's levitation criterion requires. Space-Based Solar Power owns collection and transmission at the one scale anyone has costed. Orbital Shipyards, Space-Based Manufacturing and Asteroid Mining own assembly and feedstock — and none of them owns self-replication, which is named here as an unowned premise rather than assigned.

Established “Astronomers have found seven possible Dyson spheres.” That was true in May 2024 and is not true now. Three of the seven have direct line-of-sight galaxy explanations; candidate G was resolved by e-VLBI as a background active galactic nucleus with a brightness temperature above 10⁸ K and nothing at all at the star's position; candidates D and E were resolved by JWST/MIRI in July 2026 as background galaxies at redshift 0.9 and 0.4; and an independent group's sky-density analysis accounts statistically for all seven. Established Reporting the candidate list as open in 2026 is simply out of date.

Established “Dyson proposed building a sphere around the Sun.” He proposed looking for one. The 1960 Science paper is an observing proposal whose argument is that waste heat at 200–300 K falls in a window ground telescopes can see. Established His own words are “artificial biosphere”, not solid shell, and his stated obstacle was that a whole-sky scan was “probably beyond the capability of existing techniques of detection.” The subject begins as an instrument problem and it has stayed one.

Speculative “A solid shell is unstable but could be actively station-kept.” The instability is real — by Gauss's law the interior field is zero and the shell is only neutrally stable — but it is not the killer. Established The killer is buckling: the material would need an elastic modulus of at least 10¹³ GPa, nine orders of magnitude beyond carbyne. Station-keeping cannot repair a modulus. A swarm avoids the problem entirely by not being a shell, which is why every serious treatment is about swarms.

Established “Tabby's Star was probably an alien megastructure.” It was a legitimate hypothesis, stated precisely, and it lost. Multiband photometry of the post-Kepler dips showed differential reddening favouring non-grey extinction — the dimming depends on wavelength, which an opaque macroscopic occulter cannot produce and sub-micron dust can. Frontier The star's separate long-term secular dimming remains unexplained and the two should not be conflated; the 2018 paper says so itself.

Handwave “Humanity is at Kardashev 0.7 and will reach Type I in about a thousand years.” This figure circulates widely and appears in one of the papers cited on this page. Established Planetary Scale Energy Systems established that no obtainable source supports any specific Kardashev index for humanity; the number is a convention repeated across a literature, not a measurement. The detectability arithmetic in the same paper is usable and the 0.7 is not.

Speculative “Building one is mostly an energy problem.” Unbinding Mercury — the standard feedstock in the standard scenario — costs about 1.3 hours of the Sun's total output. Handwave The energy is trivial. What is not trivial, and what every scenario assumes without argument, is exponentially self-replicating autonomous manufacture, for which there is no demonstration at any scale. The framing that calls this an engineering problem smuggles the hardest unsolved problem in as a premise.

Speculative “A Dyson swarm would be obvious, so the nulls prove nobody has built one.” The nulls are strong for the objects they were tuned for and weak elsewhere. Hephaistos I states explicitly that constraints weaken sharply for incomplete structures. Frontier And Wright's economic argument predicts that a real swarm would be substantially incomplete, because each additional nine of efficiency costs a thousandfold more radiating material. The correct statement is that nobody within the searched volumes has built one of the searched kind at the searched completeness.

Speculative “Swarm and Dyson bubble are the same idea.” They differ by a factor of about 2,600 in areal density. A collector at 3.9 kg/m² must be in a real orbit; levitating on radiation pressure requires about 1.5 grams per square metre. Frontier McInnes's 2026 stability theorem applies to the light case and Armstrong and Sandberg's mass budget to the heavy one, and the two literatures do not talk to each other.

Handwave “Nobody knows what a Dyson swarm would do to its star, so it does not matter.” The first half is exactly right and the second does not follow. Stars have negative heat capacity, so returning energy inward should make a star expand, cool and dim, and Wright names the whole question “an area of study ripe for investment.” Speculative If the effect is significant, every luminosity assumption in every construction scenario and every detection model is wrong in an unknown direction. This brief hands the question to Stellar Engineering, which owns anything acting on the star.

Established And the framing itself. “Stellar-scale energy capture is an engineering problem” fails twice. Speculative It fails on the engineering side because sixty-six years have produced one positive theorem — a stability result published in January 2026 — against a review that mainly establishes what cannot be built and a mass budget resting on an undemonstrated premise. Established And it fails on the observational side because the framing predicts abundance and the sky returns upper limits at every scale it has been probed: 2 × 10⁻⁵ of nearby stars, 8 × 10⁻⁴ across five kiloparsecs, and no galaxy in a hundred thousand above 85% reprocessing. That tension is the most interesting thing in the subject, and calling it engineering hides it.