For six decades, SETI researchers have been pointing their radio telescopes at a narrow slice of the electromagnetic spectrum and telling the world what they found — or more precisely, what they did not find. Those null results were real. But a new study presented at the Royal Astronomical Society’s National Astronomy Meeting in Birmingham last week shows that every one of them came with a hidden asterisk: the surveys that found nothing had actually searched through far more of the galaxy than anyone ever reported.
The number is stark. A major Breakthrough Listen survey that examined 1,327 telescope pointings was previously estimated, using the European Space Agency’s Gaia star catalogue, to have captured roughly 288,000 background stars in its observations. Apply a more sophisticated galactic simulation to the same data, and that estimate balloons to more than 6.1 million stars — a factor of roughly 21. The search reached not 10,000 light-years into the galaxy, but up to 25 kiloparsecs (about 81,500 light-years) — deep into the far side of the Milky Way.
Louisa Mason, a PhD researcher at the University of Manchester’s Jodrell Bank Centre for Astrophysics, presented two related papers at NAM2026 on July 24. The first described the first-ever technosignature survey conducted using the Atacama Large Millimeter/submillimeter Array (ALMA), the world’s most powerful observatory at radio frequencies above 35 GHz, located at roughly 5,050 meters (16,568 feet) above sea level on Chile’s Chajnantor Plateau. The second introduced the new stellar counting method using the Besançon Galactic Model, a Milky Way population simulator that can estimate the full distribution of stars along any line of sight — including stars too faint, too distant, or too embedded in crowded stellar fields for Gaia to catalogue reliably.
Neither paper detected a signal from an extraterrestrial civilization. Both changed what the scientific community understands about the search itself.
SETI’s Six-Decade Commitment to One Small Slice of the Radio Dial
The “water hole” is a phrase that entered SETI’s vocabulary in 1971, coined by engineer Barney Oliver to describe the narrow frequency band between 1.42 and 1.66 gigahertz (GHz). The name is a pun: hydrogen and hydroxyl — the two molecules that combine to form water — each emit natural radio signals at these wavelengths, and Oliver argued that any technologically sophisticated civilization would recognize those frequencies as a universal meeting place, the way a watering hole draws animals in an arid landscape. The logic was elegant, and it stuck.
More than 60 years of radio SETI searches have been conducted predominantly in this band, or nearby. Recent ambitious programs, including Breakthrough Listen — launched in 2016 with funding from investor Yuri Milner — have pushed the upper ceiling of searches to around 20 GHz. That is still well below where ALMA operates, which spans 35 to 950 GHz across ten frequency bands. The entire millimeter and submillimeter regime sits above what any current SETI program systematically examines.
“For decades, SETI searches have concentrated on a relatively small part of the radio spectrum,” Mason said at NAM2026. “We wanted to ask what might happen if we looked somewhere very different.” Source
There is a physical reason the water hole has been favored. Between roughly 1 and 10 GHz, the radio sky is relatively quiet: Galactic emission dims at higher frequencies, the cosmic microwave background has not yet become dominant, and the Earth’s atmosphere is still largely transparent. Above 10 GHz, atmospheric water vapor and carbon dioxide begin absorbing incoming signals, making ground-based observations increasingly difficult. ALMA overcomes this by operating at altitude — thin, dry air at 5,050 meters (16,568 feet) reduces atmospheric interference — and by using 66 antennas whose combined collecting area equals that of an approximately 84-meter (276-foot) single dish, delivering sensitivity no other facility can match at these frequencies.
But there is no physical law that requires an extraterrestrial civilization to broadcast at 1.42 GHz. The water hole assumption rests partly on a kind of cosmic optimism — that technologically sophisticated species separated by millions of light-years would independently converge on the same symbolic frequency — and partly on the practical fact that SETI has historically gone where radio astronomy already had instrumentation. Whether millimeter and submillimeter wavelengths offer a different set of communication advantages to hypothetical transmitters — higher information-carrying capacity, different propagation through the interstellar medium, or simply a different agreed-upon meeting place — remains an open question that no one had empirically tested until Mason’s work.
What Does It Actually Take to Search for Alien Signals at 90 GHz?
Mason’s approach was methodologically inventive. Rather than applying for dedicated telescope time — a scarce resource at an oversubscribed world-class facility — she mined ALMA’s public archive of observations originally collected for entirely unrelated astrophysical purposes, predominantly studies of star formation and molecular cloud chemistry. This approach, sometimes called commensal or opportunistic science, is increasingly common in SETI research: the telescope is pointing anyway; why not examine the data for narrowband spikes?
The challenge at millimeter frequencies is formidable. At 1 GHz (inside the water hole), a narrowband signal from a distant civilization would drift across the frequency spectrum at a maximum rate of roughly ±4 hertz per second (Hz/s) as the transmitter and receiver move relative to each other due to orbital and rotational motion. At 90 GHz, that drift rate scales proportionally: the maximum expected rate is ±400 Hz/s — 100 times faster. An alien signal that would sit neatly within a narrow frequency bin at 1 GHz smears across a much wider swath at 90 GHz, demanding more sophisticated signal processing to catch.
There is also the problem of spectral confusion. ALMA’s Band 3 — the frequency range Mason worked in, centered on 90.6 and 93.2 GHz — is crowded with natural molecular emission lines from interstellar molecules including carbon monoxide, hydrogen cyanide, and diazenylium. Any search for artificial narrowband signals must distinguish them from a dense forest of natural cosmic chemistry.
And then there is the hardware gap. ALMA’s existing correlator — its on-site signal processing brain — is not built for Hz-level spectral resolution of the kind that SETI searches require. Mason’s paper notes that achieving this resolution with ALMA would necessitate a specialized backend, similar to the Commensal Open-Source Multimode Interferometer Cluster installed at the Very Large Array for Breakthrough Listen, or the BLUSE system at MeerKAT in South Africa. Without such a backend, the archival data can be searched, but not at the sensitivity that purpose-built SETI instrumentation achieves.
Working within these constraints, Mason examined four archival ALMA calibrator observations in Band 3. She found no candidate technosignatures above detection thresholds. The survey covered 28 Galactic stars identified from the Gaia DR3 catalogue within ALMA’s undistorted field of view — a primary beam roughly 69 arcseconds wide at these frequencies. For the closest star in the sample, the data ruled out transmitters with an equivalent isotropic radiated power (EIRP) greater than approximately 7 × 10¹⁷ watts. That is a formidable threshold — about a hundred million times the total power output of all human radio transmitters combined — but one that an advanced civilization might plausibly operate.
The significance of the null result is not in the silence itself, but in what it establishes: a first, calibrated baseline from which future millimeter-band searches can build.
“The millimeter and submillimeter radio bands remain almost completely unexplored for SETI, so this is really about opening up a new area of parameter space to search,” Mason said.
Breakthrough Listen Was Watching 21 Times More Stars Than Anyone Reported
The second contribution from Mason’s work is equally consequential — and more immediately applicable to every SETI program that has already run.
When a radio telescope is trained on a single target star, it captures radio waves from an enormous volume of sky. Other stars happen to fall within the telescope’s field of view — “stellar bycatch” — and their signals, if any existed, would be captured in the same data. SETI researchers have long known this, and have typically estimated the bycatch population using the Gaia catalogue, ESA’s landmark survey of approximately 1.8 billion stellar objects.
Gaia is extraordinarily precise for nearby stars. But it has documented limitations that matter acutely for deep-sky bycatch counting. Its completeness drops significantly in crowded regions: in the most densely packed stellar fields, like globular clusters, Gaia’s magnitude limit for reliable detection can rise from G ≈ 20 to as bright as G ≈ 17 or brighter, meaning fainter stars are simply absent from the catalogue. Distance estimates become unreliable beyond roughly 10,000 light-years, where parallax uncertainties grow large. Only about 30 percent of objects in Gaia DR2 pass the recommended contamination filters, meaning the effective survey is substantially sparser than the headline figure of 1.8 billion implies.
Mason’s second paper proposed replacing Gaia-based bycatch estimates with the Besançon Galactic Model (BGM), a well-validated simulation of the Milky Way’s stellar population developed at the Observatoire de Besançon in France. The BGM models four stellar populations — thin disc, thick disc, bulge, and spheroid — applying star formation histories, initial mass functions, and evolutionary tracks to generate a complete predicted distribution of stars for any given line of sight, including populations that Gaia cannot observe.
Applied to the 1,327 pointings of the Breakthrough Listen Enriquez/Price survey at the 100-meter (328-foot) Green Bank Telescope in West Virginia, the difference was striking. The Gaia-based estimate: approximately 288,000 background stars captured in the survey data. The BGM-based estimate: more than 6.1 million stars, reaching out to distances of up to 25 kiloparsecs — roughly 81,500 light-years — across the galaxy.
This is not a minor methodological footnote. It means that every major SETI null result — every “no signal found” published in the peer-reviewed literature — was reported against a coverage baseline that understated the actual number of stars examined by a factor of up to 21. The cosmic haystack that SETI has searched is meaningfully larger than its own published maps of it.
“One of the most exciting things about this work is realising that we’ve surveyed many more stars than initially thought,” Mason said. “Even a very small observation can contain a huge number and diversity of stars that we might never have intended to study. By combining high-frequency observations with galactic simulations, we can better understand exactly what we’ve searched and where we should look next.” Source
Why Absence of Evidence Matters More Than It Sounds
Mason is careful not to overstate what these two findings mean for the existence of extraterrestrial intelligence. The ALMA survey covered only four archival observations, at two spectral windows, in one of ALMA’s ten frequency bands. The BGM improvement, however dramatic numerically, does not tell us that the 6.1 million stars examined by the Enriquez/Price survey were searched at all frequencies, at all sensitivities, or for all types of signals — only that the stars were present in the data.
What has changed is the conceptual accounting. The framework for understanding how much of the galaxy SETI has examined — the “cosmic haystack” in the language of a 2018 paper by Penn State astronomer Jason Wright and colleagues — depends on accurate coverage fractions in all relevant dimensions, including how many stars were actually in the data. Prior coverage fractions, calculated using Gaia bycatch estimates, were systematically understated. Future surveys applying the BGM approach will have a far more accurate picture of what they have examined.
The work was conducted with Professor Michael Garrett, also of the University of Manchester and Leiden Observatory; Dr. Andrew Siemion of the SETI Institute; and Dr. Kelvin Wandia of the University of Manchester. It was presented as a poster, “Strategies Utilising High-Frequency Interferometric Data to Explore SETI Parameter Space,” as part of NAM2026’s Statistical Challenges for Next-Generation Astronomical Surveys session.
The two underlying papers — “Conducting high-frequency radio SETI searches using ALMA” (MNRAS 536, 2127–2134, 2025) and “Simulating the stellar bycatch: constraining the prevalence of extraterrestrial transmitters within radio SETI surveys” (MNRAS 545, 2026) — are published in Monthly Notices of the Royal Astronomical Society. Paper 1 Paper 2
How Do We Know We Have Searched Enough?
SETI has been quietly grappling with this question since at least 2010, when Jill Tarter and colleagues estimated the cumulative coverage of 50 years of radio SETI as equivalent to searching “1.6 cups of water from Earth’s oceans” for evidence of fish. Wright et al. (2018) updated the metaphor: even after major Breakthrough Listen investments, the fraction of the searchable parameter space examined is comparable to a hot tub’s worth of the world’s oceans.
ALMA’s unique frequency coverage makes it not a supplement to existing SETI surveys but a genuinely new window. No other facility currently conducting SETI operates at the frequencies where ALMA excels. Whether an extraterrestrial civilization — if any exists — would choose to transmit at 90 GHz, at 1.42 GHz, or at frequencies no one has yet searched remains one of the most profound open questions in science. Mason’s work cannot resolve it. But it opens the aperture a little wider, and corrects the record on how wide the aperture already was.
Frequently Asked QuestionsWhy has SETI focused on the water hole for so long when there are so many other frequencies?
The water hole — the frequency band between 1.42 and 1.66 GHz bounded by the natural emission lines of hydrogen and hydroxyl — was proposed in 1971 by engineer Barney Oliver as a logical interstellar meeting frequency, on the reasoning that any civilization would recognize the symbolic significance of the molecules that make water. The frequency range also coincides with a region of relatively low cosmic background noise, making signals easier to detect. Decades of radio astronomy infrastructure was built at these frequencies, and SETI research followed that infrastructure. The assumption that any transmitting civilization would converge on the same symbolic frequency has never been independently verified — it is a cultural projection, not a physical law.
What did ALMA actually find, and does the null result mean there are no aliens?
ALMA found no candidate technosignatures — no narrowband signals consistent with artificial transmission — in the four archival observations Mason examined, at two frequency windows in Band 3 (around 90.6 and 93.2 GHz). This rules out transmitters above a certain power level at those specific frequencies, toward those specific directions, during the observation period. It does not rule out weaker signals, signals at other frequencies, signals in other directions, or transmitters that were not broadcasting at the time. SETI researchers emphasize that the total fraction of the multidimensional search space examined to date remains tiny — comparable, in one widely cited analogy, to a hot tub’s worth of the Earth’s oceans.
How does the new Besançon Galactic Model finding change what we know about past SETI results?
Every major SETI survey publishes an estimate of how many stars it covered, which is then used to calculate upper limits on the prevalence of alien transmitters. If that coverage estimate was based on Gaia and Gaia was systematically undercounting background stars — by a factor of roughly 21 in the Enriquez/Price survey — then the published upper limits were also off. The correct implication is not that SETI “failed to find aliens” across a small sample, but that it conducted an unacknowledged, far deeper search than its own records stated. Future surveys applying the BGM approach will be able to state precisely which stars were in their data, rather than which ones Gaia happened to catalogue.
Should ALMA be used for dedicated SETI searches going forward?
Mason’s paper identifies both the promise and the practical obstacles. ALMA’s sensitivity at millimeter frequencies is unmatched, and its archive already contains a substantial body of data that could be searched. But its existing correlator is not optimized for the Hz-level spectral resolution that SETI searches require — a specialized backend, similar to those built for the VLA or MeerKAT, would need to be installed or dedicated time purchased with compatible processing. A commensal approach, piggy-backing on observations taken for other purposes, appears the most practical near-term path — extending the archival method Mason demonstrated to a larger fraction of ALMA’s ten frequency bands.