Interstellar Archaeology

Interstellar archaeology is the search for the physical and electromagnetic traces of technology elsewhere — beacons, leakage, artefacts, waste heat, industrial markers. This brief owns the search as a programme: what has been looked for, with what instruments, to what depth, and what the nulls license as conclusions. The framing under test is that technosignatures are a searchable record.

The answer turns on one number, and it is the number this brief exists to carry. Summing every survey in the field's own census, the fraction of the eight-dimensional search space examined is 6.0 × 10⁻¹⁸. The authors' analogy is a large hot tub against the volume of Earth's oceans. NASA's own workshop report, edited by the same lead author, states the consequence without hedging: “current upper limits are quite weak.”

That does not make the nulls worthless. Several of them are real, quantified upper limits on specific things — a bound on how many nearby stars carry a transmitter of a given power aimed at us, a bound on radio emission from an interstellar visitor down to about a tenth of a watt. Those are measurements and this brief flags them as such. What the nulls do not do is bear on the population question, and Interstellar Civilization Models owns that inference.

The sharpest finding in the retrieved record is not about anyone else. It is that an observer sitting at Mars during an Earth–Mars conjunction over the past twenty years would have had a 77% chance of intercepting one of our Deep Space Network transmissions — 4 × 10⁵ times better than a random observer at a random time. Our brightest deliberate emissions are narrow in beam and narrow in time. If another civilisation resembles us, it is not leaving a searchable record; it is leaving a set of coincidences that have to be intercepted from the right place at the right moment.

One boundary is drawn explicitly and it divides on purpose rather than on dataset. Megastructure-specific results — infrared-excess candidate lists and their resolutions — belong to Dyson Swarms and Stellar Engineering. This brief lists waste heat as one channel in the agency taxonomy, notes in a clause that megastructure searches are its largest single application, and links out. It names no candidates and reports no candidate results.

Established The keystone number, with its definition attached. Wright, Kanodia and Lubar, How Much SETI Has Been Done? Finding Needles in the n-Dimensional Cosmic Haystack (AJ 156:260, 2018), sum every survey in their table and report the fraction of an eight-dimensional search space examined as 6.0 × 10⁻¹⁸. The eight axes are sensitivity to transmitted or received power; transmission central frequency; three spatial dimensions covering distance and sky position; transmission bandwidth; time and repetition rate; polarisation; and modulation.

Established The two analogies, kept separate because they belong to different authors. Wright and colleagues' own result is coverage “similar to the ratio of the volume of a large hot tub or small swimming pool to that of the Earth's oceans” — roughly eight thousand litres against the world ocean. The older and more famous image, from Tarter and colleagues in 2010 and quoted by Wright, is “akin to having searched a drinking glass's worth of seawater for evidence of fish in all of Earth's oceans.” The Tarter paper was not obtained for this brief and its analogy is reported only as quoted by the sources that carry it.

Frontier And the caveat the authors insist on, which travels with the number. “We reemphasize, however, that our numbers are very sensitive to our haystack definition and boundaries (recall that our haystack volume goes as d⁵_max).” The fraction is a function of where the box is drawn. What is robust is the order of magnitude of the deficit; the digits are not, and printing 6.0 × 10⁻¹⁸ without the fifth-power scaling misrepresents a paper that is careful about exactly this.

Frontier An interest note that cuts the useful way. Jason Wright is a SETI researcher whose institutional interest is served by “we have barely looked.” He publishes the caveat anyway, and adds the counter-caution: “We should be careful, however, not to let this result swing the pendulum of public perceptions of SETI too far the other way by suggesting that the SETI haystack is so large that we can never hope to find a needle.” A researcher arguing against the most self-serving reading of his own result raises the weight of that result.

Established The agency's own verdict on completeness, and it is a stronger statement than any sceptic's. Wright and Gelino's NASA and the Search for Technosignatures (Lunar and Planetary Institute, 28 November 2018) cites earlier coverage estimates near 10⁻²² and more recent ones near 10⁻¹⁸, and states: “These small numbers illustrate that the total searches completed to date is still quite small, and current upper limits are quite weak.” That is an agency workshop report saying the field's nulls do not mean what popular accounts take them to mean.

Established The same report supplies the institutional history that explains the coverage figure. “The field of searching for technosignatures has had no regular source of federal funding for decades”; fewer than ten PhDs have been awarded in the area; the last comprehensive review was published in 2001. The search is thin because it has been unfunded, not because it has been thorough and disappointing — and those two stories imply opposite things about what a null means.

Established The report's channel taxonomy is the organising frame for everything below. Beyond radio: optical and near-infrared, both pulsed-laser and continuous-wave spectroscopic; waste heat and megastructures; Solar System artefacts including probes, surface installations and geochemical markers; atmospheric industrial pollution; and structural signatures in transit or occultation data. Waste heat is one channel of six, megastructure searching is its largest single application, and those results belong to Dyson Swarms and Stellar Engineering. This brief does not carry them.

Frontier Methodology: how to compare searches that have nothing in common. Sheikh's Nine Axes of Merit for Technosignature Searches (International Journal of Astrobiology 19:237, 2020), developed at the 2018 workshop, states the problem it solves: “The diverse methodologies and myriad orthogonal proposals for the best technosignatures to search for in SETI can make it difficult to develop an effective and balanced search strategy, especially from a funding perspective.” The nine axes are timeliness; cost; ancillary benefits; duration and duty cycle; ambiguity; technological extrapolation; contrivance versus inevitability; detectability; and information content, with an open-source tool for plotting a search against them.

Frontier The 2026 framework adds the dimension that matters epistemically. Haqq-Misra, Framing the Possibility Space for Technosignature Searches (arXiv:2603.17741), proposes a two-parameter matrix — accessibility (is it within our observable domain?) by recognisability (could we identify it?) — yielding four strategies: Explore for the accessible and recognisable, Expand for the recognisable but out of reach, Evolve for the accessible but unrecognisable, and Exist for neither. The Evolve quadrant is the dangerous one: signals already collected and sitting in archives that we cannot recognise. A separate mission-concept framework introduces the ichnoscale — the size of a technosignature's cosmic footprint — as a way of comparing classes quantitatively.

Established Now the null record, with its actual limits, because a null without a bound is not a measurement. Enriquez and colleagues' Breakthrough Listen survey (arXiv:1709.03491, 2017) covered 692 nearby stars from 1.1 to 1.9 GHz and found no transmitters at an effective isotropic radiated power near 10¹³ W, bounding such transmitters to fewer than about 0.1% of systems within 50 parsecs. Eleven candidate events appeared and all were consistent with terrestrial interference.

Established The strongest published bound is Painter's, and it is worth quoting exactly. Painter and colleagues (AJ 169:222, 2025) reprocessed 6,630 cadences of 2,623 stars across the L, S, C and X bands from the Green Bank Telescope archive — “the largest dataset searched for technosignatures to date” — and conclude that “less than 1% of stars host transmitters brighter than 0.3 Arecibo radar equivalents broadcasting in our direction over the frequency band covered.” Read the qualifiers: in our direction, in those bands, above that power. Everything outside them is untouched.

Established The widest-field null is the Murchison Widefield Array's, and it comes with the most useful self-assessment in the literature. Tremblay and Tingay (PASA 37, 2020) surveyed 400 square degrees toward Vela at 98–128 MHz, covering more than ten million Gaia-distanced stellar sources and six known exoplanets, reaching a best EIRP limit of 6.2 × 10¹² W and a search fraction of about 2 × 10⁻¹⁶ — two orders of magnitude above the previous best and about three above the largest non-MWA survey. “No unknown signals were found with a 5 sigma detection threshold.” The authors' own conclusion: “SETI has a long way to go.”

Established Two more, filling different corners of the box. The Sardinia Radio Telescope's high-frequency survey (arXiv:2410.09288, 2024) observed 72 TESS targets and the Galactic Centre at 6 GHz and 18 GHz — the first survey ever at 18 GHz — in 36 hours, reaching an EIRP limit of 1.8 × 10¹⁹ W for roughly 500,000 stars, with “no definitive evidence of drifting signals.” And a Breakthrough Listen search around 27 eclipsing exoplanets with the Parkes ultra-wideband receiver at 704–4032 MHz (arXiv:2506.13459, 2025) generated about 1.95 million signals, of which 14,639 survived filtering, and “all events were attributed to terrestrial radio frequency interference.”

Established The pattern across all five is uniform and it is the honest summary of radio SETI: nulls with weak bounds, and interference as the dominant failure mode. The bound that means something is Painter's. Everything outside those bands, powers and pointings has not been examined at all, which is precisely what a coverage fraction of 6 × 10⁻¹⁸ encodes.

Frontier The optical channel is an instrument story rather than a results story, and its first published result is not a technosignature. PANOSETI comprises two assemblies of about 45 telescopes each, targeting roughly 4,450 square degrees with sensitivity from nanosecond to second timescales, synchronised by a White Rabbit timing network, with an instantaneous field of view near 2,500 square degrees in the full design. A baseline of more than one kilometre between telescopes viewing the same field markedly suppresses false positives from Cherenkov air showers. Its first on-sky science result was astrophysics: three gamma-ray events at about 15–50 TeV from the Crab Nebula, confirmed jointly with VERITAS.

Frontier Target selection has broadened, and the field's answer to “you only ever look at Sun-like stars” is a catalogue. Lacki and colleagues' One of Everything: The Breakthrough Listen Exotica Catalog (ApJS 257:42, 2021) contains 963 entries covering 816 distinct targets in four samples: a prototype sample — “archetype of every known major type of non-transient celestial object” — a superlative sample, an anomaly sample of “enigmatic targets that are in some way unexplained”, and a control sample. The stated rationale is that breadth serves detection chance, commensal astrophysics and systematics characterisation at once.

Speculative Artefact searching in our own system, and the finding inside the proposal is the more interesting half. Benford's Looking for Lurkers: Objects Co-orbital with Earth as SETI Observables (arXiv:1903.09582, 2019) argues that Earth's co-orbitals offer “resources an ETI might need: materials, a firm anchor, concealment” with a permanent view of Earth, naming 2016 HO3 (Kamo'oalewa) as “the smallest, closest, and most stable (known) quasi-satellite of Earth.” His framing of the search is the phrase this brief is named for: “As they would remain there after their energy supply runs out, this is extraterrestrial archeology.”

Established The checkable claim inside that advocacy is the one to carry. These objects “have been little studied by astronomy and not at all by SETI or planetary radar observations.” The parameter space closest to Earth is among the least examined. Benford is a long-standing SETI and beamed-propulsion advocate and the proposal is advocacy; the statement that nobody has looked is checkable and it stands.

Established The interstellar-object campaign of 2025 is the one place in this brief where a search is close to complete for its target, and the bounds are genuinely strong. Breakthrough Listen's observations of 3I/ATLAS: the Allen Telescope Array in July 2025 across 1–9 GHz reaching 10–110 W EIRP; MeerKAT in November at 900–1670 MHz reaching 0.17 W — described as “comparable to mobile phone transmission power at that distance”, with hydroxyl detected from ordinary cometary activity; Parkes across 704–4032 MHz reaching about 5 W at closest approach on 5 October; and the Green Bank Telescope on 18 December 2025 across 1–12 GHz reaching about 0.1 W EIRP. The Vera C. Rubin Observatory covered the optical channel. “No technosignatures have been detected in any of these searches.”

Established A tenth-of-a-watt bound on an interstellar object is worth printing precisely because so few bounds in this subject are strong. Interest runs against this null: the programme reporting it is the one that would benefit most from a detection, and it published a limit that closes the question for its own target.

Frontier And the result that turns the instrument back on us. Sheikh and colleagues, Detecting Extraterrestrial Civilizations That Employ an Earth-level Deep Space Network (arXiv:2508.15425, 2025): an observer positioned at Mars during an Earth–Mars conjunction in the past twenty years would have had a 77% chance of catching one of our Deep Space Network transmissions — “a 4 × 10⁵-fold increase over intercepting our Deep Space Network transmission versus a random observer at a random time.”

Frontier Read the other way, that is the strongest statement in this brief. A random observer at a random time has essentially no chance of detecting Earth by our brightest deliberate emissions, because they are narrow in beam and narrow in time. If another civilisation is like us, the record we are searching for does not exist in the form we are searching for it — except along sight-lines and at moments we do not control.

Frontier The keystone number is a volume fraction, and a volume fraction is not the same object as a probability. Wright and colleagues are explicit that the haystack is a box they drew and that its volume goes as d⁵_max; what follows from that and is less often said is that the coverage figure weights every cubic parsec and every hertz equally, while no searcher believes the prior is flat. Established Surveys concentrate on nearby stars, on sun-like hosts already in catalogues, and on frequencies near the water hole between the hydrogen and hydroxyl lines, because that is where the field’s priors sit. Frontier The concentration cuts both ways, and this brief should say which. Taken with the prior, the nulls are worth more than 6.0 × 10⁻¹⁸ suggests, because the examined fraction of prior mass exceeds the examined fraction of volume. Taken against the possibility that the prior is wrong, they are worth less, because a systematically misplaced prior buys a search that is deep everywhere the signal is not. Speculative No source obtained for this brief publishes a prior-weighted coverage figure beside the volume one, and that single number would do more to make the null record interpretable than another survey would.

Established The field publishes its nulls, which is rarer than it sounds, and it publishes them in a form that cannot be added up. A technosignature paper conventionally reports its own sensitivity limit and target list; it does not report the volume of the common haystack it consumed in the common units. Established That is why Wright, Kanodia and Lubar had to re-derive the coverage of every survey in their table from each paper’s own description rather than sum a published column. Frontier The consequence is structural rather than clerical: a literature whose nulls cannot be summed without a manual re-derivation will have its aggregate coverage recomputed rarely, by whoever is willing to do the work, so the field’s most-quoted number is always as old as the last person to do it. Speculative A machine-readable coverage declaration attached to every published null — frequency span, drift range, sensitivity, dwell, target list, sky solid angle — is a cheap convention nobody has adopted, and its absence is a reporting choice rather than a scientific limit.

Established Position one, and it is measured rather than argued: the search has covered a negligible fraction of the parameter space. 6.0 × 10⁻¹⁸ of an eight-dimensional haystack, with the agency's own report calling the resulting limits “quite weak.” Position two, also established: the nulls do usefully bound one specific thing. Fewer than 0.1% of systems within 50 parsecs carry a 10¹³ W transmitter in L band; fewer than 1% of 2,623 stars carry one at 0.3 Arecibo-radar equivalents aimed at us across four bands. Those are real upper limits on loud, continuous, directed transmitters, and they are the whole of what the null record establishes.

Frontier Position three: search strategy can be made comparable across methods. Sheikh's nine axes and Haqq-Misra's accessibility-recognisability matrix are published and in use; no consensus weighting exists, so two people can score the same programme differently and both be defensible. Position four: breadth of target beats quality of target. The Exotica Catalog's 816 targets spanning essentially all of astrophysics is the field's operational answer, and it has produced no detection — which is what one would expect either way at this coverage.

Frontier Position five: optical nanosecond pulses are a viable independent channel. PANOSETI is built and validated, with a kilometre-plus baseline suppressing Cherenkov false positives, and its first published science result was gamma-ray astrophysics rather than SETI. That is a genuine ancillary-benefit datapoint for Sheikh's third axis and it should be read as one.

Speculative Position six: alien artefacts could sit in Earth co-orbital space, unexamined. Benford, 2019, with an interest disclosure attached. The load-bearing part is checkable and holds: co-orbitals have been little studied by astronomy and not at all by SETI or planetary radar. Position seven: interstellar objects are artificial. A minority claim, made most loudly about the first such object; and 3I/ATLAS was searched across 1–12 GHz to about 0.1 W EIRP by four facilities with nothing found.

Frontier Position eight is the one that reorganises the subject, and it is contradicted by evidence rather than by argument: a civilisation like ours would be detectable by our methods. It is the implicit premise of radio SETI. The Deep Space Network study says otherwise — 77% at a conjunction geometry against 4 × 10⁵ times worse at a random one. The premise is not merely unproven; there is now a published result against it.

Handwave Position nine: the absence of detections is evidence of absence. The popular reading, ruled out by the coverage figure and explicitly disowned by the searchers in print. This brief's contribution is to make the disowning quotable rather than to argue it again. Position ten, out of scope on purpose: megastructure infrared excess as a detectable technosignature. It is a channel in the taxonomy and its results belong to Dyson Swarms and Stellar Engineering, which have worked it to a conclusion this brief does not restate.

Frontier And the open question that would most change the field: how much of the accessible-but-unrecognisable quadrant is already sitting in archives? Painter's reprocessing found candidates in archival data that earlier pipelines missed. That is direct evidence that the Evolve quadrant is populated by our own processing choices rather than by the sky, and it is the cheapest place in the whole subject to look.

Established Anomaly triage is the operational centre of a modern search, and the field has exactly one fully worked public example. Breakthrough Listen’s pipeline discards the overwhelming majority of hits by cadence: the telescope alternates on-source and off-source pointings, and a signal present in both is terrestrial by construction. Established A second filter requires a nonzero Doppler drift, on the reasoning that an emitter fixed to the Earth does not drift relative to the Earth. Frontier Both filters are selection effects with the sign reversed: the cadence filter throws away any signal that happens to be on during both pointings, and the drift filter throws away a transmitter that has corrected its own drift — which is what a transmitter built to be found would do. Established The candidate known as blc1, a narrowband hit near 982 MHz in Parkes data toward Proxima Centauri, survived the automated filters, was worked through a published verification framework and was attributed to local interference. Frontier Neither paper was obtained for this brief, so what is reported here is the shape of the episode rather than its numbers; the shape is the point, because it is the only case in which this field’s triage has been executed in public from hit to retraction.

Frontier There is no agreed detection threshold in this subject, and the reason is a multiple-comparison problem the field rarely states in those terms. A wideband search dissects a night into billions of frequency–time–beam cells, and a per-cell false-alarm probability means nothing without the trials factor that converts it into a search-wide one. Established Particle physics answered the same problem with a fixed, pre-registered five-sigma convention and an explicit look-elsewhere correction; technosignature work has no equivalent number, and each survey sets its own hit threshold to control its own follow-up workload. Frontier What exists instead is a communication instrument rather than a statistical one: the Rio scale, revised as Rio 2.0, scores a claimed detection for consequence and credibility so that a public statement can be calibrated after the fact. Speculative A pre-registered threshold would be worth more than a post-hoc scale, because the failure mode this field actually faces is not a fabricated claim but a defensible one made at a threshold chosen once the candidate was already in hand.

Established The first bottleneck is coverage and it scales brutally. The haystack volume goes as the fifth power of maximum range, so doubling the distance a survey can reach multiplies the space to be covered by thirty-two. Coverage is therefore not a matter of observing longer; it is a matter of collecting area, bandwidth and field of view — which is why this brief's adjudicated dependency is an instrument brief and why the dependency is arithmetic rather than thematic.

Established The second is radio-frequency interference, and it is the dominant failure mode in every survey reviewed here. Enriquez's eleven candidates were all anthropogenic; the Parkes eclipsing-exoplanet search generated 1.95 million signals, kept 14,639 after filtering, and attributed every one to terrestrial interference. The instrument that would detect a distant transmitter is optimally designed to detect our own, and separating the two consumes most of the analysis effort in the field.

Frontier The third is recognisability, and it is the least tractable. A signal can be in the archive and not be identified, because the pipeline was built around an assumed signal form — narrowband, drifting, continuous. Haqq-Misra's Evolve quadrant names the problem and Painter's archival result demonstrates it: new methods recovered candidates in data that earlier pipelines had already searched. The bottleneck is a prior about what a technosignature looks like, and it is invisible to the observers who hold it.

Established The fourth is funding, and it is documented rather than inferred. No regular federal funding for decades, fewer than ten PhDs in the area, the last comprehensive review in 2001. A field of that size cannot cover an eight-dimensional space, and the thinness of the coverage is a direct consequence of the thinness of the institution. Private programmes have changed the observing throughput substantially and have not changed the fraction by an order of magnitude that matters.

Frontier The fifth is that the search's own premise may be wrong, which is a bottleneck of a different kind. If civilisations resemble us, their brightest emissions are narrow beams used briefly for their own spacecraft, and the detection probability for a random observer is essentially zero. No increase in sensitivity fixes a geometry problem; what fixes it is either luck, or a shift to channels — artefacts, waste heat, atmospheric markers — that persist rather than transmit.

Speculative And a sixth that is a gap rather than an obstacle: the nearest unexamined parameter space is the one under our feet. Earth's co-orbital objects have not been surveyed by SETI or by planetary radar at all. An artefact search there is cheap by the standards of this field and has never been run, which says something about how the field's attention is allocated as well as about how much is left.

Frontier Data openness is a rung on this workback and it is already half climbed. Breakthrough Listen publishes raw and reduced data with documented formats, and the direct consequence is the archival reanalysis this brief already reports: candidates recovered from data that had been searched once and passed over. Established An open archive converts a one-shot observation into a standing asset whose coverage rises whenever anyone improves a pipeline, which is the only mechanism in the subject that enlarges the searched fraction without buying telescope time. Speculative A null on a closed archive can be trusted and cannot be checked, and this brief could obtain no inventory of how much technosignature-capable raw data sits in observatory archives under retention policies that will delete it before anyone reprocesses it.

Established The adjudicated dependency is Mega-Telescopes, and it is arithmetic rather than thematic. Every limit in section 2 is instrument-limited. The haystack volume scales as the fifth power of maximum range, so collecting area and bandwidth are the coverage. MeerKAT's 0.17-watt bound on an interstellar object and the Murchison array's two-order-of-magnitude jump in search fraction are both instrument stories, not analysis stories. This brief states the scaling and hands the hardware over.

Established The second relationship is the measurement-inference seam with Interstellar Civilization Models. This brief reports what was searched and what was found. That brief argues what it implies about the population. The one sentence that crosses the seam: at 6 × 10⁻¹⁸ coverage the null constrains loud, continuous, beam-at-us transmitters within tens of parsecs and essentially nothing else. The inference from there is II-13's, and it uses the same coverage figure to show why a null carries almost no information about N.

Established The third is the boundary the commission flags, and it divides on purpose rather than on dataset. Dyson Swarms and Stellar Engineering own megastructure-specific results: infrared-excess candidate lists, their contamination analyses and their resolutions. This brief lists waste heat as one channel in the agency taxonomy, notes that megastructure searching is its largest single application, and links. It names no candidate and reports no candidate outcome. In the reverse direction, those briefs cite this one for the haystack fraction, the nine axes and the completeness argument rather than re-deriving them.

Frontier Fourth, Deep Space Communications, on the transmission question. This brief owns whether a search programme should transmit; that brief owns the engineering of transmission. The Deep Space Network detectability study sits exactly on that seam and belongs in both, for different reasons: here it is evidence about what a civilisation's emissions look like from outside, there it is a statement about how our own network radiates.

Speculative And fifth, Civilization Timelines, which takes the dark forest as a periodisation claim and records that no peer-reviewed formulation was located. The argument that silence is deliberate is that brief's; this one owns only the search-programme half of the transmission decision. Link, do not re-argue.

Established Experiment one is the cheapest unexploited option in the subject: survey Earth's co-orbitals. Benford's checkable claim is that these objects have not been examined by SETI or by planetary radar at all, and the strongest named target — the quasi-satellite 2016 HO3 — is stable, close and already characterised as an astronomical body. The observation costs radar time rather than a new instrument, and its null would close a named hypothesis rather than adding a fraction of a haystack.

Frontier Experiment two: reprocess the archives with methods that do not assume the signal form. Painter's reanalysis of 6,630 cadences from the Green Bank archive is the model: it recovered previously known candidates, which validates the method, and it searched the largest dataset yet assembled without a new observation. If the Evolve quadrant is populated by our own processing choices, the archives are where the evidence for that lives.

Frontier Experiment three: complete the optical build and run it long. PANOSETI's design covers about 2,500 square degrees instantaneously at nanosecond resolution with a kilometre-plus baseline for false-positive rejection. The nanosecond-pulse channel is genuinely independent of radio and its background is dominated by an astrophysical process — Cherenkov showers — that a baseline separates cleanly. Its first result already produced usable gamma-ray astrophysics, which is the ancillary-benefit case made concrete.

Established Experiment four: keep the interstellar-object rapid-response capability. The 2025 campaign on 3I/ATLAS demonstrated that four facilities can be turned on a transient visitor and reach EIRP limits from 110 watts down to about 0.1 watt across 1–12 GHz within months. That is the one search in this brief that approaches completeness for its target, and the capability is procedural rather than instrumental — it can be preserved deliberately or lost by default.

Speculative Experiment five, self-directed and unusually informative: observe ourselves. The Deep Space Network study computed what an outside observer would see. Extending it to the full inventory of terrestrial and orbital emissions — radars, communication satellites, planetary radar, leakage — would produce the first empirical model of what a civilisation at our level actually looks like from a distance, which is the missing calibration for every survey in section 2.

Frontier Experiment six: agree, in advance, what a continued null would mean. Sheikh's axes and Haqq-Misra's matrix make searches comparable; neither converts coverage into an inference rule. A published statement of the coverage level at which a continued null becomes a constraint would convert seventy years of accumulating negatives into a measurement, and its absence is why the nulls are simultaneously real and unusable.

Established The engineering requirements in this subject are unusually concrete because the searches are instrument-limited in a measurable way. Coverage scales as the fifth power of range, so the requirements are collecting area, instantaneous bandwidth, field of view and the ability to observe many targets at once. The Murchison array's jump of two orders of magnitude in search fraction came from a wide field at low frequency, not from a longer campaign.

Established Second, interference rejection, which is where most of the actual engineering effort goes. A survey that produces 1.95 million signals and keeps 14,639 after filtering — all of them terrestrial — is an interference-classification problem wearing an astronomy hat. The requirement is provenance-tracked pipelines with published rejection criteria, because the same choices that remove interference can remove a signal.

Frontier Third, timing and baselines for the optical channel. Nanosecond-to-second sensitivity requires network-level synchronisation — White Rabbit in PANOSETI's case — and a baseline over a kilometre between telescopes viewing the same field, which converts a Cherenkov air shower from a false positive into a geometrically excluded event. The false-positive problem is solved by architecture rather than by thresholding, which is the right way to solve it.

Established Fourth, archives and reprocessability. The single most productive analysis in the recent record reused 6,630 cadences already on disk. The engineering requirement is that raw data be retained at a resolution that permits reanalysis with methods that do not yet exist, which is a storage and format decision made years before the analysis that depends on it.

Speculative And fifth, an instrument this field does not have: a dedicated wide-field, wide-band, always-on facility. Every bound in section 2 came from time borrowed on instruments built for other purposes, or from a programme funded privately. What the coverage arithmetic implies is that the search needs a facility whose figure of merit is haystack fraction per year, and no such thing has been built. Mega-Telescopes owns the hardware case.

Established The boundary that matters most here is with the megastructure briefs, and it is drawn on purpose rather than on dataset. Dyson Swarms and Stellar Engineering own megastructure-specific observational results — the infrared-excess candidate lists, the contamination analyses and the resolutions. This brief owns the search programme and its methodology: the haystack model and its coverage figure, the nine axes, the accessibility-recognisability matrix, the ichnoscale, the agency workshop's institutional history and its verdict on limits, the radio nulls and their EIRP bounds, optical instrumentation, artefact proposals, interstellar-object campaigns and the interference problem. Waste heat appears here as one channel of six and nothing more.

Established Second, Mega-Telescopes, the adjudicated dependency, because the coverage is the hardware. Third, Interstellar Civilization Models, which takes the coverage figure and argues the inference this brief declines to make.

Frontier Fourth, Deep Space Communications, which shares the transmission question and owns its engineering. Fifth, Civilization Timelines, which owns the dark forest as a periodisation claim and records that no peer-reviewed formulation of it was located — a fact worth knowing before anyone cites it as a reason for silence.

Established Sixth, Exoplanet Colonization, which shares instruments and not questions. Biosignature and technosignature spectroscopy use the same telescopes; a claim about life in an exoplanet atmosphere is that brief's, and a claim about industry — atmospheric pollution as a technosignature — is this one's. The channel is named in the agency taxonomy and no result exists in it, which is worth saying plainly.

Speculative And seventh, planetary science, which owns the objects an artefact search would examine. Earth's co-orbitals are studied as small bodies and have never been examined as candidate artefact sites. The adjacency is real and the collaboration has not happened, which is the most actionable observation in this section.

Established The institutional facts here are documented in an agency report rather than inferred, and they explain the coverage figure completely. “The field of searching for technosignatures has had no regular source of federal funding for decades”; fewer than ten PhDs awarded in the area; the last comprehensive review in 2001. A search that has covered 6 × 10⁻¹⁸ of its parameter space is what a field of that size produces, and reading the thinness as a verdict on the hypothesis inverts cause and effect.

Frontier The second institutional fact is that the recent throughput is largely privately funded, and that changes what gets searched. The largest datasets, the interstellar-object campaigns and the exotica-catalogue approach all come from a private programme. Private funding buys observing time and does not buy the standing capability — a dedicated wide-field facility, a permanent archive mandate, a graduate pipeline — that coverage at scale requires. Both facts should be stated together.

Established Third, an interest-disclosure norm that this field mostly observes and that this brief adopts. Several sources here are produced by people whose funding depends on the search continuing, and two of the most load-bearing results run against their producers' interest: a SETI researcher publishing the caution against over-reading his own coverage result, and a search programme publishing a tenth-of-a-watt null on its own highest-profile target. Marking those cases is not decoration; it is how a small field's evidence should be weighed.

Frontier Fourth, the archive mandate, which is cheap and consequential. The single most productive recent analysis reprocessed data already collected. A requirement that technosignature-capable raw data be retained and made reanalysable is the highest-return institutional decision available to this field, and it has to be taken years before the methods that will exploit it exist.

Speculative And fifth, the transmission decision, which has no institutional owner at all. Whether to deliberately signal is currently decided by whoever has a transmitter, and the arguments deployed on both sides are drawn from population models that cannot distinguish an empty galaxy from a settled one. An international framework does not exist; naming that absence is more useful than adjudicating the argument here.

Frontier The post-detection question has a document and no standing. The declaration of principles maintained by the international academy’s SETI committee sets out what a detector should do — verify, notify, publish, and refrain from replying without international consultation — and it binds nobody, having no treaty behind it. Speculative The realistic sequence for a modern candidate is that it surfaces in an open archive or a preprint before any committee sees it, which is roughly what happened with blc1 and is what the field’s own openness commitments make likely. Established That is not an argument against openness; it is an argument that the protocol was written for a world with fewer copies of the data, and no source obtained for this brief proposes a revision fitted to the one that exists.

Established The first ethical obligation in this subject is not to misreport the null, and the field discharges it better than its audience does. The searchers state in print that current limits are quite weak. The popular reading — seventy years of silence proves nobody is there — is contradicted by the people who did the searching, and repeating it is a misrepresentation of a published position rather than a defensible interpretation.

Speculative Second, the transmission question, and this brief's contribution to it is deflationary in both directions. Arguments for and against deliberate signalling rest on claims about who is out there and what they want, and those claims come from models that do not discriminate. What the search record adds is one relevant fact: our own emissions are already detectable at conjunction geometries with a 77% probability, which means the choice is not between silence and disclosure but between deliberate and accidental disclosure.

Frontier Third, the handling of anomaly claims, where the incentives are badly aligned. An artefact or interstellar-object claim generates enormous attention and a null generates none, so the reputational cost of a premature claim is borne by the field and the benefit by the claimant. The 2025 campaign is the counter-example worth holding up: four facilities, published EIRP limits down to about 0.1 watt, and a plain statement that nothing was detected.

Established Fourth, and it applies to this page as much as to any other: interested parties must be marked. The strongest artefact proposal in this brief comes from a long-standing advocate; the coverage figure comes from a SETI researcher; the interstellar-object nulls come from the programme that would most benefit from a detection. Every one of those is still usable, and two of them are strengthened by the direction their interest runs.

Speculative And fifth, a point about what an artefact would mean if one were ever found. Benford's framing — that a derelict remains after its power fails, and that finding one would be archaeology rather than contact — has consequences nobody has worked through: for ownership, for contamination, for whether an object in Earth's co-orbital space falls under any existing legal regime. Space Law and Governance owns the regime; the observation that it was not written for this case belongs here.

Frontier The civilizational finding of this brief is self-directed and it is the one worth carrying away: our own civilization is not a searchable record by our own methods. An observer at Mars during conjunction would have had a 77% chance of catching our Deep Space Network; a random observer at a random time would have had a chance 4 × 10⁵ times worse. Our brightest deliberate emissions are narrow in beam and narrow in time, and a search built to find them would miss us.

Established The second implication is about what the accumulated nulls actually mean. They are real measurements: under 0.1% of systems within 50 parsecs carry a 10¹³-watt L-band transmitter, under 1% of 2,623 stars carry one at 0.3 Arecibo-radar equivalents aimed at us, no unknown signal above five sigma in 400 square degrees at low frequency, nothing above about 0.1 watt from an interstellar visitor across 1–12 GHz. Every one of those is a bound on a specific kind of thing, and none of them is a statement about whether anyone is there.

Frontier Third, the shape of the record we are building. At 6 × 10⁻¹⁸ coverage, the honest description is not that we are reading a record slowly. It is that we have barely established that a record exists, using instruments whose strongest published bound leaves every other possibility untouched. The deficit is robust even though the digits are not.

Speculative Fourth, and it cuts against the reflexive pessimism: the search is cheap relative to what it would settle. The field has no standing federal funding line, fewer than ten doctorates and a review two decades old, and it has still produced quantified upper limits, a methodology literature, an instrument programme and a self-calibrating result about our own detectability. What it has not had is the resources that its own arithmetic says the question requires.

Established And fifth, the reframing this brief recommends. Interstellar archaeology is better understood as a survey-completeness problem than as a detection problem. The interesting quantity is not what has been found but how much of the space has been examined, and that quantity is measured, published, caveated by its own authors, and eighteen orders of magnitude short.

These horizons track search completeness and instrument capability, not the probability of a detection — nothing here supports forecasting that.

• 10 yr: Established Coverage rises with collecting area and bandwidth rather than with observing time, and on any plausible instrument schedule it remains many orders of magnitude short of informative for the population question. Frontier Archival reprocessing is the highest-yield activity in this window and requires no new hardware; whether the accessible-but-unrecognisable quadrant is populated is decidable from data already on disk. Speculative An artefact survey of Earth's co-orbitals is affordable now and has never been run; whether it happens is an attention question, not a capability one. Frontier Expect the coverage accounting to be recomputed at least once on this horizon and expect the revision to come from a reporting convention rather than from new sky, because the archives already hold more unsearched volume than the next decade of telescope time will buy.
• 25 yr: Frontier A dedicated wide-field, wide-band facility whose figure of merit is haystack fraction per year would move coverage by orders of magnitude; nothing of the kind is currently funded, and Mega-Telescopes owns the case for it. Established Interstellar-object campaigns continue as objects arrive, and the 2025 record shows the response can reach EIRP limits near 0.1 watt within months of discovery. Speculative Whether the field acquires a standing funding line is the single variable that determines everything else in this row.
• 50 yr: Speculative If coverage rises by ten or more orders of magnitude and the sky stays quiet, the null begins to constrain the loud-continuous-directed corner of the space and only that corner — and no published inference rule currently says what that would license. Frontier The recognisability problem does not improve with aperture, so the useful long-horizon investment is in methods that do not assume a signal form. Handwave Any statement about a detection at this horizon has no base rate to rest on.
• 100 / 250+ yr: Handwave Beyond forecasting, and this brief declines rather than extrapolating. Established The structural statement that survives any horizon is the self-directed one: our own brightest deliberate emissions are narrow in beam and narrow in time, and a civilisation resembling ours would not appear in a survey resembling ours except by a coincidence of geometry and timing. Speculative That fact argues for persistent channels — artefacts, waste heat, atmospheric markers — over transmissions, and the persistent channels are the least searched.

• Depends on One edge, and it is arithmetic rather than thematic. Mega-Telescopes owns collecting area, bandwidth and field of view, and in this subject those quantities are the coverage: the haystack volume scales as the fifth power of maximum range, so a factor of two in reach is a factor of thirty-two in space to be searched. Every bound in this brief is instrument-limited — MeerKAT's 0.17-watt limit on an interstellar object, the Murchison array's two-order-of-magnitude jump in search fraction from a wide low-frequency field, the Green Bank archive's 2,623-star reprocessing. The dependency is not decorative: without an instrument result the coverage figure does not move, and the coverage figure is the whole subject.
• Requires (not on this map) Two constraints that are not briefs on this map. The first is institutional and is stated in the field's own agency report: no regular federal funding for decades, fewer than ten PhDs awarded, the last comprehensive review in 2001. Coverage of an eight-dimensional space is a resourcing problem before it is a scientific one, and the thinness of the search is a direct consequence of the thinness of the institution. The second is a data-retention constraint: the most productive recent analysis reprocessed 6,630 cadences already on disk and recovered candidates earlier pipelines had missed, which means raw data must be kept at a resolution permitting methods that do not yet exist — a storage and format decision taken years before the analysis that depends on it. One further constraint arrives with the triage and threshold material and it is not a research result: an agreed detection threshold and a post-detection protocol that actually binds. Each survey currently sets its hit threshold to control its own follow-up workload, claims are scored after the fact on a communication scale rather than against a pre-registered false-alarm probability, and the declaration of principles covering what happens next has no treaty behind it. Neither gap closes by observing more sky.
• Enables No typed enabling edge is claimed. A search programme that has covered 6 × 10⁻¹⁸ of its parameter space and returned nulls with weak bounds does not enable a downstream capability. What it supplies instead is an input to an argument: Interstellar Civilization Models takes the coverage figure and shows why a null at that level carries almost no information about the population. That is a citation relationship rather than a dependency, and recording it as an edge would overstate what this brief produces.
• Adjacent Dyson Swarms and Stellar Engineering own megastructure-specific observational results; this brief lists waste heat as one channel of six in the agency taxonomy and names no candidates, which is a division by purpose rather than by dataset. Interstellar Civilization Models owns the inference from nulls to population. Deep Space Communications owns transmission engineering and shares the signalling question. Civilization Timelines owns the dark forest as a periodisation claim. Exoplanet Colonization shares the spectroscopic instruments and owns the biosignature half; industrial markers in an exoplanet atmosphere are this brief's channel and have produced no result. Space Law and Governance owns the regime that an artefact discovery would fall outside.

Established “We have been searching for seventy years and found nothing, so probably nobody is there.” The searches to date cover 6.0 × 10⁻¹⁸ of an eight-dimensional space — a hot tub against the world ocean. Established NASA's own workshop report states that “current upper limits are quite weak”, and the searchers disown the inference in print. The null is not evidence of absence at this coverage, and saying so is the field's position rather than a critic's.

Established “The sky has been examined.” What has been examined is a handful of bands, at high power thresholds, mostly toward nearby stars, mostly assuming a continuous beam pointed at us. Established The strongest published bound is that fewer than 1% of 2,623 stars host a transmitter above 0.3 Arecibo-radar equivalents aimed in our direction in the bands observed. Every qualifier in that sentence is load-bearing, and everything outside them is untouched.

Established “There have been tantalising candidate signals.” There have been many, and every one in the surveys reviewed here resolved to terrestrial interference or to ordinary astrophysics. Established One search generated 1.95 million signals, kept 14,639 after filtering, and attributed all of them to radio-frequency interference; another's eleven candidates were all anthropogenic. Interference is the dominant failure mode of the entire field and it is the reason candidate counts are not news.

Speculative “The interstellar objects might be artificial.” The claim is a minority position and it has now been tested for the most recent visitor. Established 3I/ATLAS was observed by four facilities across 1–12 GHz with limits from 110 watts down to about 0.1 watt EIRP — comparable, at that distance, to a mobile phone — plus optical coverage, and nothing was detected. Established The hydroxyl that was detected came from ordinary cometary activity. This is one of the few places in the subject where a search is close to complete for its target.

Frontier “A civilisation like ours would obviously show up in our surveys.” There is now a published result against this. Frontier An observer at Mars during an Earth–Mars conjunction would have had a 77% chance of catching one of our Deep Space Network transmissions — and a random observer at a random time would have done 4 × 10⁵ times worse. Speculative Our brightest deliberate emissions are narrow beams used briefly for our own spacecraft. If they are like us, the record does not exist in the form we are looking for.

Established “6 × 10⁻¹⁸ is a precise measurement of how much we have searched.” The authors say otherwise, explicitly. Established The haystack volume scales as the fifth power of maximum distance and the fraction is “very sensitive to our haystack definition and boundaries.” The deficit is robust across any reasonable choice of box; the digits are not, and quoting the number without the caveat misuses a paper that is careful about precisely this.

Frontier “If a signal were in the data we would have found it.” A reprocessing of the Green Bank archive — 6,630 cadences of 2,623 stars — recovered candidates that earlier pipelines had missed, using methods that did not assume the same signal form. Frontier That is direct evidence for the accessible-but-unrecognisable quadrant of the search space: the limit is a prior about what a technosignature looks like, held by the people who built the pipeline and invisible to them.

Speculative “Nobody has bothered to look close to home.” Correct, and it is checkable. Established Earth's co-orbital objects “have been little studied by astronomy and not at all by SETI or planetary radar observations” — a statement from an artefact-search advocate whose proposal is advocacy and whose factual claim stands. The least examined parameter space in the subject is the one nearest to us.

Established “The megastructure candidates are the most exciting result in technosignature science.” Whatever their status, they are not this brief's to report. Established Waste heat is one channel of six in the agency taxonomy and megastructure searching is its largest single application; the candidate lists, their contamination analyses and their resolutions belong to Dyson Swarms and Stellar Engineering, which have worked them to a conclusion. This brief divides from them on purpose rather than on dataset, and names no candidate.

Established And the framing itself. “Technosignatures are a searchable record” fails on both words. Established On searchable, because 6 × 10⁻¹⁸ of the space has been searched by a field with no standing funding, fewer than ten doctorates and a two-decade-old review. Frontier And on record, because the one civilisation we can examine directly does not leave one: our transmissions are narrow in beam and narrow in time, and a survey like ours would miss us. Speculative What follows is not that the search is hopeless but that it is aimed at the wrong noun — persistent artefacts, waste heat and atmospheric markers are records, and transmissions are coincidences.