Finding alien life will not be one discovery. It will be a claim that survives every independent check. This is a plan to put the public inside that checking.
The real sky: 1,637 naked-eye stars, and 536 radio bursts from other galaxies, replayed in the order the CHIME telescope caught them. One year in 45 seconds.
Press play. This is real: 7 minutes 39 seconds of Jupiter's radio bursts, recorded by the University of Florida and hosted by NASA. Slow swells like surf on a beach, rapid pops like a crackling fire. It is not a message. It is the first sound every listener has to learn, because a search for anything else starts by ruling it out.
Recording: University of Florida, from NASA's Radio JOVE library. Four more recordings, and live telescopes you can tune from your browser.
110 real places on a 3-D map of nearby stars, each with its own page: how far it is, how long a probe would take, and what is known about it.
Why one dish is not enough, and what a worldwide network would be for. No alien signal has been detected. The sound under the narration is real: radio waves recorded by NASA's Van Allen Probes.
A real recording to play here, then six live receivers, free, in your browser. None of them plays a signal from another civilization, because none has ever been confirmed. This is what listening really sounds like.
These sites are run by their own teams, not by One Sky Alien Array. Checked on October 3, 2026.
The dome is the sky over one place on Earth, worked out from the clock. Circles are the stars humanity has sent messages to. Diamonds are the worlds and directions astronomers rank highest for a search. Radio passes through daylight and cloud, so a target counts as up whether or not you could see it.
Showing Greenwich, England. Enter a latitude and longitude to see your own sky. West and south are negative numbers.
The big nearby-star surveys watched each star for about fifteen minutes and only counted a signal that appeared in all three of their scans. A powerful beacon that reaches Earth for five minutes a day, or an hour a week, could never have been recorded by them, at any power. That gap is the whole reason this project exists, and the arithmetic fits on a page.
The network does not exist yet. Every figure outside the Worlds tab is a real measurement or a calculation from one, with its source named. The Worlds tab is a simulation and says so. Solar readings are a snapshot from 2026-10-03 01:01 UTC.
Real recordings you can play here, and real radio telescopes you can open in a browser right now, free.
Each was made by a real receiver and is hosted by NASA. None is a message. These are the natural radio sounds every listener has to learn first, because a search for anything else starts by ruling them out.
A sequence of sample bursts from Jupiter, picked up by antennas on Earth. Jupiter's bursts come in two kinds: slow swells that sound like surf on a beach, and rapid pops like a crackling fire. A home station with a wire antenna can catch them.
A burst recorded at 20.1 and 20.8 MHz during a strong flare. This kind, called Type II, comes from a shock wave moving out through the Sun's atmosphere.
A Type III burst at 20.1 MHz. These come from beams of electrons thrown out by the Sun and last only seconds.
Falling tones recorded in space by NASA's Van Allen Probes. These radio waves are slow enough to play directly as sound. In a whistler the high tones travel faster and arrive first, which makes the pitch fall.
Rising chirps, like birds at dawn, from the same spacecraft. They are radio waves in the belts of charged particles that surround Earth.
These are real antennas you can tune from a browser, free. Nothing here plays a signal from another civilization, because none has ever been confirmed. What you get is the real thing a home station would hear: the Sun, Jupiter, meteors, lightning, and a great deal of human interference. Each link was checked on October 3, 2026.
A 1956 research dish now run by volunteers, with a receiver you tune yourself. When the dish is not live it plays recordings of signals bounced off the Moon. A second antenna listens for meteor echoes from a radar in France.
Students and amateurs with home-built radio telescopes record radio storms from Jupiter and bursts from the Sun. A team member streams a receiver live on YouTube, linked from the project site, and years of recordings are in its archive.
An antenna aimed at a distant TV transmitter in Canada that is normally out of range. Each ping you hear is a meteor burning up and its trail reflecting the signal for a moment.
Radio made by Earth itself, at frequencies low enough to play straight through a speaker. The steady crackle is lightning, often from thousands of kilometers away.
Small solar radio spectrometers around the world, with a map that updates about every 15 minutes and a burst of the day. Solar bursts are the event our own network has to catch first, in step 4 of the build plan.
One receiver that many people tune at once. Almost everything on it is human-made: broadcasters, aircraft, amateur radio. It is good practice for the main skill in this work, telling interference from sky.
SALSA, at Onsala Space Observatory in Sweden, lends 2.3 m dishes over the internet at no charge. You create an account, book a time, and map the hydrogen in the Milky Way's spiral arms, the same 1420 MHz line our station kit would measure. PICTOR, a 1.5 m dish in Athens, takes observation requests through a web form with no booking, but it was under maintenance when checked on October 3, 2026.
These sites are run by their own teams, not by One Sky Alien Array. Some use plain http addresses, so a browser may show a "not secure" note.
What humanity has found so far, how sure anyone is, and the messages already on their way.
In 2021 NASA scientists proposed a seven-step ladder for judging any claim of life beyond Earth, the Confidence of Life Detection scale. A claim starts at the bottom and climbs only as independent checks succeed. Here is the ladder, and where the real cases sit.
Steps paraphrased from Green et al. 2021, Nature 598, 575. The scale was written for biosignatures; radio signals are judged on the separate Rio scale, but the logic of independent confirmation is the same.
A rock nobody can bring home. A spectrum only one telescope can take. A radio signal nobody else was pointed at. In each case the first observation was real, and the second look never came or did not agree.
Second looks are something a worldwide public network can supply. That is the purpose of everything below.
The Perseverance rover found organic carbon and mineral spots that microbes can produce. NASA called it a potential biosignature. The cored sample is still on Mars, and the US Congress confirmed cancellation of the return mission in January 2026.
A team reported a gas that on Earth comes mainly from ocean life. Independent reanalyses of the same James Webb data found insufficient evidence; ordinary hydrocarbons fit as well or better.
A radio-telescope detection of a gas with no known source there. Other teams reprocessed the data and found a much weaker signal or none. Still unresolved.
A 72-second narrow-band signal at Ohio's Big Ear telescope, logged as 6EQUJ5 on the printout. Nobody else was watching that patch of sky and it never repeated. It can be neither confirmed nor dismissed.
A candidate from the Parkes telescope in the direction of the nearest star. The search team's own follow-up traced it to human interference. The checking worked.
No claim has passed step 2. Saying so plainly is part of doing this honestly, and it is why the count at the top of this page reads zero.
Since 1974, radio telescopes have beamed deliberate messages at nearby stars. Each one travels at the speed of light, so its position today is simple arithmetic: the years since it was sent. The bars below update from your clock.
Each transmission lasted minutes to hours and was aimed once. To catch one, a listener would need a large dish pointed at the Sun at the right moment. Nobody should expect an answer. If one ever came, these are the earliest dates it could arrive.
Seen a new claim of alien life, a paper that changes one of the cases on this page, or a message to the stars that is missing from the list? Send it with a link to the source. Every submission is checked before anything on this site changes.
submissions received so far.
Your submission goes into a queue. It is read against its source. If it holds up, the record on this page is updated and you are credited if you gave a name.
Anything without a checkable source is not used. Sightings and personal experiences are outside what this record covers.
The idea in one page, then five checks, on real data, of whether a worldwide network of small stations could do useful work.
Radio telescopes have pointed at nearby stars since 1960 and heard nothing, and that silence is usually read as a verdict: if anyone were broadcasting, we would know. That reading holds for one kind of broadcast, a steady tone, because that is what the searches were built to catch. It does not hold for a powerful signal that reaches Earth briefly and repeats, five minutes a day or an hour a week. For that kind of signal, the silence is not evidence of absence. It is the sound of not having listened.
The largest nearby-star surveys of the past decade observed each target for three five-minute scans inside about half an hour, and most stars were observed once. A beacon that lights up Earth for five minutes once a day falls inside those scans during about 1% of its cycle. For a beacon that repeats weekly, about 0.15%. The telescopes were not deaf. They were elsewhere.
To count as a candidate, a signal had to appear in all three of those scans. A burst shorter than the gaps between them fails that rule by construction. No short, isolated burst, at any power and any repeat period, could have reached the published candidate lists. The silence those surveys reported says nothing about such a signal.
Suppose a search had accepted a signal seen in any one scan, and recovered half of what the timing allowed. In a thousand-star survey the limit on steady transmitters would be about 0.6% of stars. For a once-a-day five-minute beacon the same arithmetic gives 57%. Beyond a repeat period of about 1.7 days there is no limit at all: a beacon at every star is compatible with total silence.
A hypothetical survey of 1,000 stars, three five-minute scans per star starting at 0, 10 and 20 minutes, any single-scan detection accepted, 50% recovery. Where a curve leaves the top of the chart, the silence proves nothing. The dashed line is the proposed campaign: 144 paired five-minute visits per star on 300 stars, for the far more powerful beacons a small-dish network can hear. Timing geometry under ideal detection, reproduced from Section 2 of the research proposal.
Move the sliders. The survey is the hypothetical one in the chart; the campaign is the proposed network's schedule. Both assume the signal is strong enough to see when the telescope is looking.
A small dish cannot hear leakage, radar or ordinary communication traffic from another star. A three-meter dish needs a transmitter delivering sixteen times the power of the Arecibo planetary radar toward Earth to notice a one-hertz tone at ten parsecs. The only thing it can hear is a beacon built to be heard, and the published analyses of what such a beacon would cost conclude that it would not be a steady tone: a transmitter minimizing cost per detection sends powerful, directional pulses swept across many targets. The one signal class that inexpensive instruments can reach is the one class no survey has tested.
First, an audit that costs nothing but time. Using only the timestamps in public Breakthrough Listen data, measure which beacon rhythms every observed star has and has not been covered for, and publish the table and the code. If coverage turns out to be high, the project stops there and says so.
Second, a benchmark with locked rules. Inject synthetic bursts into real archive noise and test a short-window search against the standard pipelines, with the rules registered in public before any result is opened.
Third, and only if the first two justify it, a global ring of small stations in pairs, spread around the world because a single site can never reach more than about a third of the phases of a once-a-day beacon. Every stage has a stop condition, and a stopped stage is a result.
We are not saying anyone is transmitting. We are not saying small dishes can out-hear Green Bank; they cannot, by a factor of several thousand. The published limits on steady transmitters stand. We are saying something narrower: for the one kind of signal a cost-conscious civilization would plausibly send, and the one kind a worldwide network of modest dishes could plausibly hear, the record of the search so far is close to empty, and the arithmetic that shows it fits on a page.
On October 24, 2026 we ran the first version of the audit above on the public Breakthrough Listen archive: the timestamps of 7,174 L-band scans of 3,734 stars by the Green Bank Telescope, of which 1,121 stars had the full three-scan cadence, with the three scans about 11 minutes apart. On the generous any-scan terms used in the chart, that record limits steady transmitters to about 0.5% of those stars, but limits a once-a-day five-minute beacon only to 43% of them, and says nothing at all about a beacon repeating less often than every 2.3 days. Under the published rule, which needs the signal in all three scans, the record has no sensitivity to any burst shorter than the gap between scans.
These numbers are preliminary: our own computation from archive timestamps, reproduced with the method in Section 2 of the proposal, and not yet reviewed by anyone else. The code and the full per-star table will be published as Step 3 of the build plan.
The argument in two pages: the companion brief. The derivations, tables, references and reproduction program: the research proposal, Sections 2, 6 and 7 (codename Lost Voices). The author is responsible for every number and claim in both.
The US Air Force runs four solar radio observatories spread around the globe. Each one measures the Sun's radio output near its own local noon. One of their frequencies, 1415 MHz, sits 5 MHz from the hydrogen line that SETI searches favor.
Flux in solar flux units (1 sfu = 10,000 Jy), both axes logarithmic. Hollow markers are readings the observatory itself flagged as uncertain.
Snapshot read from NOAA on October 3, 2026, 01:01 UTC.
Australia read 59, Italy 62, Hawaii 56 sfu. Three stations a third of the planet apart, hours apart, agree within about 5 percent.
The check also catches bad data. Hawaii's 610 MHz reading was 16 sfu against Italy's 52, and the observatory flagged it as uncertain. One station alone could not have known.
Separately, Canada's Penticton observatory measured the standard 10.7 cm index at 92 sfu on October 2 (90-day mean 115). Source: NOAA Space Weather Prediction Center.
Pick a real target. The lit half of the globe has it above the horizon at this moment, by your device's clock. A flash from that direction reaches each station at a slightly different time, and those gaps are what triangulation measures.
Or click the globe to drop a station. It stays in your browser only.
The flash is a calculation, not a signal: it shows the wavefront crossing Earth and the moment it reaches each station, slowed about 235 times. The darker half of the globe is where it is night right now.
Fast radio bursts are millisecond flashes from other galaxies. Canada's CHIME telescope logged these 536 in its first catalog, between July 2018 and July 2019. They are natural, as far as anyone can tell, and their cause is still not fully understood. Click any point.
Nothing appears below about −11° declination. The bursts are there; CHIME sits at 49° north and that part of the sky never rises for it. One telescope, however good, is blind to the sky under its feet.
Rate matters too. From this catalog, CHIME's team estimated that about 525 bursts bright enough for it reach Earth every day, across the whole sky. The catalog itself averages fewer than two a day.
This is the standard radiometer equation, the same one professionals use to size a telescope. Set the dish, the receiver noise and the bandwidth, and see what clears a 10-sigma detection.
Defaults describe a 1 m WiFi grid dish, a low-noise amplifier and an RTL-SDR dongle: about US$200 of parts. Assumes 1420 MHz, 50% aperture efficiency, one polarization.
A lone receiver is fooled constantly by phones, radar, satellites and its own electronics. A real flash from space must reach every station that faces it within 85 milliseconds, the time light takes to cross the Earth and back. Random glitches almost never line up that way.
Two different jobs. Catching a one-off burst from an unknown direction, the case worked through here, takes hundreds of stations watching the whole sky. The experiment in our research proposal, listening for a repeating beacon from a known star, takes tens: see the station guide.
Assumes each station's false alarms are random and unrelated to the others, which holds better the farther apart the stations are. Uses the dish settings from Evidence 4. Sky coverage is a best case: stations spread evenly and aimed so that three share each patch.
More than 100 real places, from Mars to other galaxies: how far they are, how long a probe would take, and what the planets among them might look like.
Pick a kind of place, a destination and a speed, then launch. The map is drawn from real positions: the 4,059 cataloged stars within 100 light-years, in their real colors, and 3,617 stars known to have planets.
This tab is a simulation. Distances, positions and speeds are real. Whether anything lives at any of these places is unknown. No sign of life has been detected at any of them. Only the two groups of candidate worlds, and the ocean worlds of our own solar system, are places astronomers rank highly in the search for life. The other groups are here because of what has been measured or seen there.
Travel time is distance divided by speed, with no time spent speeding up or slowing down. Voyager 1 and the Parker probe speeds are measured. The 20% of light speed figure is a design goal for a gram-scale craft that has not been built.
No telescope can yet show the surface of any planet beyond the solar system. What is known is each planet's mass, size, orbit and the light of its star. The picture is built from those, then filled in with one of the surfaces the data allows. Switch between them: every one is still possible.
measured from telescope data in the catalog. calculated from measured values. estimated from a rule of thumb, because the value has not been measured. Oceans, ice, clouds and the shapes of the land are imagined, and no vegetation or cities are drawn because none are known.
This is the Drake equation. Astronomers have measured the first three terms reasonably well: about 2 new stars a year in our galaxy, planets around nearly all of them, and a rocky planet in the temperate zone of roughly 1 star in 5. Nobody knows the other four. Set them yourself and see what follows.
The starting positions are placeholders, not findings. Published opinions on these four terms differ by many orders of magnitude.
How to take part now, and what has to be built before the network is real.
Add yourself and you become part of the record: a dot on the member map, and a name on the list we write to when alerts, station kits and the forum open. It takes a minute and costs nothing.
founding members so far.
It puts you on the list and, if you choose, on the globe in the Evidence tab. Nothing else happens yet. We write to you only when something you signed up for opens.
You can remove yourself at any time, in the privacy section on this page.
Two of these you can start today through projects other scientists already run. The other two are what One Sky Alien Array has to build.
Search real telescope data from a laptop. No equipment and no training.
Give your location. When a real telescope reports a flash, you learn within a minute whether it is above your horizon.
The feeds already exist and are public. CHIME sends burst alerts about 13 seconds after detection, and NASA's General Coordinates Network relays alerts from space observatories. Nothing yet passes them to the public by location. This needs no hardware, so it comes first.
Run a station. A US$200 starter kit teaches the sky and detects the galaxy's hydrogen; a reference station, about US$900 to 2,000 with a GPS-locked clock and a 2.4 to 3 m dish, can join the network. The network needs tens of those in clusters around the world, not thousands, and the first pair is ours to build. Later, keep the reply schedule: listen toward each star we have messaged when an answer could first arrive.
The station logger exists (onesky-station 0.1, simulation-tested, not yet run on a radio). The scheduler and the server that corroborates candidates between stations are not written yet.
Until then you can listen through receivers other people already run: see Listen live. For the two kinds of station and the parts lists, see the station guide.
If a signal is ever confirmed, who answers for Earth? The international SETI protocols, under revision since 2025, say no reply should be sent until there has been international consultation.
No public forum for that consultation exists. Members would have one here, on the record, before anything is ever detected.
The plain answer to the first question people ask: the experiment in our research proposal needs tens of good stations in three or four clusters around the world, not thousands. Adding dishes does not make the network hear fainter signals; it buys observing time and longitude. The first pair is ours to build. The rest will come from people who already own a dish, if the software is simple enough to plug in and forget.
Using only the timestamps in public Breakthrough Listen data, measure which beacon rhythms every observed star has and has not been covered for, and publish the result. Anyone with a laptop can help. Zero dollars. If the coverage turns out to be high, the project stops here and says so.
Two stations within an hour or two of longitude, built to the specification below, documented, and copied. They prove the two things a network needs: that both stations hear the same sky at the same time, and that the small, predictable frequency difference between them separates a signal from space from anything fixed to the ground.
Regional pairs spread around the world, so that between them they can reach every phase of a once-a-day beacon that no single site can. Recruited from amateur radio astronomers, clubs and schools that already own a dish, with the station software as the pitch. The proposal's arithmetic: half coverage of a daily cycle for 300 nearby stars is about 7,200 station-hours, roughly 90 nights for ten stations.
The published build from the RTL-SDR Blog's hydrogen-line guide, their design, not ours: a 2.4 GHz WiFi grid dish (about $50), a NooElec SAWbird+ H1 amplifier with a built-in filter ($44.95), an RTL-SDR Blog V3 software radio ($21.95), a tripod mount ($39.99), a Type N to SMA adapter and a USB extension cable (about $17). Fit the feed with its long side along the grid bars, put the amplifier straight on the antenna socket, tune to 1420 MHz in SDRSharp with the IF Average plugin, record empty sky for 6 to 7 minutes, then point at the Milky Way and look for the broad bump near 1420.40 MHz.
It detects the galaxy's hydrogen and shows the galaxy rotating. It cannot be a network station: its clock drifts about 1.4 kHz at this frequency and it has no input for an outside reference. It can run our logger, and it is the right way to learn pointing and calibration before spending more.
Prices as listed in the guide, which totals $179.40 and warns they have risen. Read the full guide at rtl-sdr.com.
Parts list dated October 8, 2026. Prices are estimates and list prices; real quotes replace them before any money is asked for.
What it can hear, in the proposal's own numbers: a transmitter aimed at Earth delivering about 3 × 10¹⁴ W of equivalent isotropic power within a 1 Hz line, sixteen times the Arecibo planetary radar, at 10 parsecs. Nothing fainter. That is the whole point: the only signal such a station can hear is a beacon built to be heard, and that is the signal no survey has checked for.
Exists today, free, used by professionals. turboSETI, blimpy and setigen from Berkeley SETI search, read and inject signals into the standard file format; Virgo runs small-dish hydrogen-line observations; GNU Radio and SoapySDR capture from the radio; SDR++ and SDRSharp show the live spectrum; Stellarium plans the sky; the sgp4 and skyfield libraries with CelesTrak's public orbital elements screen satellites.
Ours, written: onesky-station 0.1. The station logger. It captures from a software radio or a built-in simulator, writes standard filterbank files that the Berkeley tools search directly, records where its time came from (GPS, synchronized clock, or unverified), keeps a rolling buffer of raw signal so a candidate's recording can be frozen, and uploads a small summary of each file. It has been tested in simulation, including recovery of injected drifting and intermittent tones by turboSETI. It has not yet been run on a radio. Download the source (Python, about 80 KB).
Ours, not written. The scheduler that decides what each station listens to, the server that corroborates candidates between stations, and the public evidence registry.
Why this design: the two-page brief and the full research proposal (codename Lost Voices).
Not ready to buy? You can steer a real dish for free first: see Listen live.
Every step ends in something that either works or does not. No step claims more than it has shown.
This page.
Test: every figure traceable to a named source. Done.
A domain, public hosting, sign-up with a city-level location, a privacy policy, a transparent way to give, and the code in an open repository.
Test: 100 members in 10 countries.
Connect to the public CHIME and NASA alert feeds. Tell each member when a real event is in their sky.
Test: a real alert reaches members in view in under 60 seconds.
No hardware. From the timestamps in one public Breakthrough Listen data release, compute how much of each possible beacon rhythm every observed star has been covered for, and publish the table and the code. This is the first result of the research proposal.
Test: a published, reproducible coverage table for one named release. If day-scale coverage is already high, the network is not justified by it, and this page will say so. Status: first run done October 24, 2026; preliminary result under The idea; code and table not yet published.
Two stations to the reference specification, within an hour or two of longitude, running the station logger with GPS-locked clocks. Built once, documented, and copied.
Test: both stations detect the Sun and the galaxy's hydrogen on the same day, and the predicted frequency offset between them for a known source matches the measured one to within a few hertz.
A server that keeps only candidates that pass the local interference monitor, the satellite screen, and a second station at the predicted frequency, and publishes everything it keeps. Then grow to six to twenty stations in three or four clusters around the world, recruited from people who already own a dish. Share data with the Society of Amateur Radio Astronomers, the SETI Institute and university groups. Open the member forum on whether and how Earth should reply.
Test: a test signal injected by cable at both stations is recovered and corroborated, a known satellite pass is correctly vetoed, and a professional observatory follows up one of the network's published candidates.
If you join: your email, the city you type, the role you pick, your country as our host reports it, and, only if you tick the map box, a location rounded to about 10 km. If you send a submission: what you wrote, the optional name, and your country.
Your email is used only to tell you when something you signed up for opens. It is never shown on this site, sold or shared. The map shows dots with no names attached.
It is stored with Cloudflare, the company that hosts this site, until you remove it. We run no advertising or analytics trackers. Joining is for people aged 16 or older. Questions, or to have something removed by hand: [email protected].
Most of this project costs time. Here is where money comes in, and what happens to it.
A Ko-fi page run by the project's founder. Money given there goes to the person building One Sky Alien Array and is his to use for it: time, tools, hosting and new ideas.
It is personal support. It is not a donation to a charity and it is not tax-deductible.
Support on Ko-fiThe link opens ko-fi.com/oneskyalienarray, where payment is handled by Ko-fi. Time helps too: the ways to join that are open today cost nothing.
A separate pot that can only be spent on stations: first one reference pair, built and documented, then the server that corroborates their signals, then help for people who already own a dish to join the ring. It opens when the pair is ready, as its own campaign or under a nonprofit sponsor.
Estimates, about US$3,750 in all. Real quotes replace them before any money is asked for.
| Part | Listed price |
|---|---|
| 1 m WiFi grid dish | about $50 |
| Hydrogen-line amplifier with filter | $44.95 |
| RTL-SDR software radio | about $21.95 |
| Tripod | $39.99 |
| Cable, adapter, terminator | about $22.50 |
| Starter kit, guide total | $179.40 |
| Reference station, our specification | $900 to 2,000 |
Starter-kit prices as listed in the rtl-sdr.com hydrogen-line guide, which notes that prices have risen since it was written. The reference station is our specification and its price is an estimate; the itemized parts list, dated October 8, 2026, is in the station guide.
A planning sum only. Reference stations are priced at US$1,500 each; running costs at US$750 a year are taken out first. The network needs about 20 such stations.