There was no public vote on whether Earth should advertise itself to the stars. No planetary authority weighed the risks, chose a frequency and authorised a greeting. Yet for decades, some of humanity\u2019s strongest persistent radio transmissions have been leaving the Solar System as a by-product of something much less theatrical: keeping in touch with spacecraft at Mars.<\/p>\n
A peer-reviewed analysis in The Astrophysical Journal Letters<\/a> reconstructed where those transmissions went from 1 January 2005 through 31 December 2024. Pinchen Fan of Penn State led the study with Penn State astronomer Jason Wright and T. Joseph W. Lazio of NASA\u2019s Jet Propulsion Laboratory.<\/p>\n The 77 percent in the headline needs careful handling. It is not the probability that extraterrestrials heard Earth, and it says nothing about whether extraterrestrials exist. It is a duty cycle for a very particular piece of sky: a distant observer positioned to see Earth and Mars align would have been inside one of NASA\u2019s deep-space uplink beams 77 percent of the time during the analysed period.<\/p>\n This is one study of one communications network, not settled consensus about Earth\u2019s detectability. Its value lies in replacing a vague idea of radio leakage with a measured pattern, then asking whether the same pattern could guide searches for technology around other stars.<\/p>\n A radio beam does not stop at Mars<\/p>\n The NASA Deep Space Network<\/a>, or DSN, is an array of large radio antennas at Goldstone in California, near Madrid in Spain and near Canberra in Australia. Their placement roughly 120 degrees apart in longitude lets one complex take over as a spacecraft drops below another\u2019s horizon.<\/p>\n Those dishes carry commands out and bring tracking and scientific data back. The network supports missions spread across the Solar System, from spacecraft at Mars to New Horizons beyond Pluto and observatories around the Sun\u2013Earth Lagrange points.<\/p>\n For an uplink, a DSN dish concentrates radio energy into a narrow cone aimed at a receiver on a spacecraft. Narrow, however, is not the same as perfectly pencil-thin. The 70-metre dishes have a half-power beamwidth of about 0.128 degrees at S band and 0.038 degrees at X band, according to the paper. That is several arcminutes across.<\/p>\n A spacecraft antenna is tiny within that cone. Even Mars usually covers only a small central patch. The craft receives enough energy to recover the command, while most of the wavefront passes around it and continues in the same direction. Nothing in the radio signal knows that Mars was its intended destination.<\/p>\n From sufficiently far away, the geometry compresses. Earth, the spacecraft and Mars appear close together on the sky. An observer almost directly beyond the spacecraft would sit along the continuation of the uplink, like someone standing behind a small target illuminated by a distant spotlight.<\/p>\n Rebuilding 20 years of invisible traffic<\/p>\n The researchers began with public DSN uplink schedules. They matched the target code in each entry to a mission, retrieved spacecraft trajectories from JPL\u2019s Horizons system and reconstructed the beam\u2019s outward direction. They also corrected for light-travel time, because the apparent location used to receive a downlink is not automatically the location toward which an earlier uplink had to be sent.<\/p>\n The resulting map contains 92.5 antenna-years of cumulative transmission time. That total can exceed the 20-year calendar interval because several antennas may transmit at once. The team represented the sky in one-arcminute cells and calculated how long each direction was illuminated.<\/p>\n Ka-band uplinks were excluded because their transmitting power was below one kilowatt and because only a few antennas supported them. Low-Earth-orbit communications were also outside the study. The map is specifically an account of stronger deep-space uplinks in S and X bands, not every artificial radio emission Earth produced.<\/p>\n The logs did not record the actual power of every uplink. That makes this principally a study of timing and direction, not a claim that every beam could be recovered at the same interstellar distance.<\/p>\n Where the 77 percent comes from<\/p>\n Mars emerged as the busiest corridor. For a line of sight within two arcminutes of the planet, the mean duty cycle was 77 percent, equivalent to about 9.4 months of transmission per year. Between two and three arcminutes, it was still about 58 percent. Beyond three arcminutes, it fell quickly.<\/p>\n In the paper\u2019s comparison, the region within two arcminutes of Mars was illuminated about 400,000 times more often than the average direction across the whole sky. The chance associated with an alignment involving a planet other than Mars was about 12 percent, more than 60,000 times the random all-sky case.<\/p>\n So the finding is conditional twice over. First, an observer has to occupy the narrow extension of the Earth\u2013Mars line. Second, the observer has to watch during the alignment. The result does not mean DSN beams have swept 77 percent of the galaxy, that 77 percent of nearby stars received them, or that a recipient noticed them.<\/p>\n It does mean that the choice of where and when to listen can transform the odds. A randomly placed observer at a randomly selected time would be very unlikely to intercept a deep-space uplink. Move that observer onto the Mars line during conjunction, and routine mission operations make the signal present more often than not.<\/p>\n Why Mars became Earth\u2019s accidental transmitter mast<\/p>\n Mars dominates because it has dominated robotic planetary exploration. Over the study period, NASA communicated with orbiters, landers and rovers there through overlapping missions. A command intended for one machine could illuminate a distant background direction even though neither the command nor its content had anything to do with interstellar communication.<\/p>\n The larger map bears the imprint of the Solar System\u2019s flat architecture. Most planets and interplanetary spacecraft remain close to the ecliptic, the plane of Earth\u2019s orbit. The team found that 79 percent of total DSN transmission time stayed within five degrees of that plane. Eighty-four percent of one-arcminute sky cells in the band were illuminated at least once, compared with fewer than 3 percent outside it.<\/p>\n The Earth Transit Zone is narrower still. This is the strip from which a distant observer could see Earth cross the Sun. Its average DSN duty cycle was about 13 minutes per year, 20 times the 0.65-minute average across all ecliptic latitudes. A system able to notice Earth by transit therefore also has a better-than-random position from which to encounter its deep-space radio traffic.<\/p>\n Two other concentrations appeared toward and directly away from the Sun. They reflect communications with spacecraft around the Sun\u2013Earth L1 and L2 points. The anti-solar peak grew fourfold in the 2022 to 2024 subset, which the authors attribute largely to transmissions directed toward the James Webb Space Telescope around L2.<\/p>\n Being in a beam is not the same as hearing it<\/p>\n The authors estimated that a typical DSN transmission could be detected by present-day human radio technology from roughly seven parsecs, or 23 light-years. There are 128 known star or brown-dwarf systems within that distance. The estimate is a useful scale, not a census of systems that actually received a particular uplink.<\/p>\n Detection would require a receiver with sufficient sensitivity, observing at the right frequency and time. The signal weakens with distance. Interstellar plasma can affect it. A civilisation would also have to distinguish an engineered transmission from local and astronomical radio noise. Recognising a carrier as artificial is not the same as decoding a command designed for a NASA spacecraft.<\/p>\n The absent power values add another uncertainty. The study identifies when a location was illuminated, but not whether an instrument at that location could recover every event. Fan and colleagues explicitly separate those questions.<\/p>\n