Posted by Start with the awkward part. There is no agreed scientific definition of extraterrestrial intelligence, and most of the people who have spent careers looking for it will admit as much before you finish asking the question. What the field uses instead is a working substitute: a short list of things that can actually be measured, chosen in large part because they can be measured. That substitution is not a scandal. It is how the search functions, and once you see it, almost everything about where the telescopes get pointed makes sense.
Ask a working astronomer what would count as a detection and you will hear about narrowband radio carriers, nanosecond laser pulses, excess infrared, industrial gases in a transit spectrum. Notice what is absent from that list. Minds. Nobody is proposing to detect thought, or curiosity, or language. The search is for evidence of engineering, and engineering is being quietly asked to stand in for whoever built it.
Nobody Can Define It on Earth Either
Before blaming astronomers for fuzziness, try defining intelligence on the one planet where we have samples. An octopus opens jars, navigates mazes and escapes tanks it was never meant to leave. New Caledonian crows manufacture hooked tools out of twigs and keep the good ones. Dolphins carry signature whistles that work something like names. None of them will ever build a radio transmitter, and none of them ever could, and it would be a strange definition of intelligence that rules them out entirely.
So the honest version of the phrase is narrower than it sounds.
When researchers say extraterrestrial intelligence, they generally mean a technological civilization: something that manipulates energy and matter at a scale large enough to change what a distant instrument sees. A species could be ancient, social, philosophically sophisticated and completely invisible to every telescope humanity has ever built. Nothing in the science rules that out. It simply puts them outside the reach of the method, which is a limitation of the method and not a claim about them.
Drake’s Quiet Substitution
In 1961, ahead of a small meeting at the Green Bank observatory in West Virginia, Frank Drake wrote out a chain of factors as an agenda rather than a formula. Star formation rate, planets, habitable ones, life, intelligence — and then the two terms that give the game away. One is the fraction of intelligent species that develop technology releasing detectable signs into space. The other is L, how long they keep releasing them.
Drake was not defining intelligence. He was defining detectability, and letting intelligence ride along as a passenger. Every serious program since has inherited that structure, from Project Phoenix, which worked through roughly 800 nearby Sun-like stars between 1995 and 2004, to Breakthrough Listen’s sweeps with Green Bank and Parkes. The unit of the search has never been a mind. It has been an output.
A Definition Made of Hardware
Freeman Dyson made this explicit in a two-page paper in 1960. A civilization with a growing appetite for energy might capture a large fraction of its star’s output, and thermodynamics being what it is, that energy has to come back out as low-grade heat in the infrared. So look for the heat. Dyson was not describing aliens; he was describing a thermal signature and arguing that it was the only part of the problem an observatory could get hold of. Nikolai Kardashev did something similar in 1964 by sorting hypothetical civilizations according to how much power they command, planetary through stellar through galactic. Crude, and openly so. But again the axis is energy, not thought.
The modern word for all of this is technosignature. Jill Tarter has pushed for it for years, partly because the older label leads audiences to expect a message, and messages are only one line item on a long list. A technosignature is any observable consequence of technology, deliberate or accidental — the interstellar equivalent of finding a wheel rut rather than meeting the cart driver.
Which answers the definitional question by dodging it. Operationally, extraterrestrial intelligence is whatever leaves an industrial fingerprint big enough to see from a very long way off.
The Narrowband Bet
Radio searches lean on a specific piece of reasoning about what nature cannot do. Astrophysical sources are messy: a hydrogen cloud, a pulsar, a quasar all emit across broad ranges of frequency. A tone a few hertz wide, holding steady, is on this planet always something somebody built. Add a second filter — genuine motion between a distant transmitter and a rotating, orbiting Earth should push the frequency along at a predictable drift rate, while a phone mast on the next ridge sits stubbornly still. Those two criteria carry most of the weight in most surveys.
On 15 August 1977, the Big Ear telescope at Ohio State recorded exactly that kind of thing near the hydrogen line: a narrowband burst that rose and fell over 72 seconds as the beam swept past. Jerry Ehman circled it on the printout and wrote “Wow!” in the margin. It never came back. Decades of re-observation have found nothing at that spot. The Wow! signal stayed famous precisely because it passed the shape test and failed the repeat test, which is the whole point of having two tests.
Breakthrough Listen had its own version. Observations of Proxima Centauri from Parkes in 2019 turned up a narrowband tone near 982 MHz with a plausible drift, and by late 2020 it had a name, BLC1, and a great deal of press. Within a year the team had traced it back to human equipment. They published the postmortem in detail, which mattered more than the candidate ever did.
What a Civilization Leaks
There is a useful inversion here: point the criteria at Earth and ask what we would look like from outside. The popular assumption is that our television broadcasts have been announcing us for eighty years, but the analyses are less flattering. Ordinary leakage is weak, spread across frequencies and beamed in every direction at once. Planetary radar, like the transmitter Arecibo carried before the dish collapsed in 2020, was enormously more powerful, but it pointed at a handful of targets for short stretches, so the odds of anyone catching it are slim.
The chemical route may be better. Chlorofluorocarbons have no plausible natural source. Nitrogen dioxide accumulates where things are burned at industrial scale. Both have been studied as targets for transit spectroscopy with instruments of the JWST class, and both come with the same caveat: only for nearby systems, only for transiting planets, and probably only after an expensive pile of observing hours. City lights, orbital debris and waste heat sit further down the same list, technically detectable and practically out of reach for now.
Anomaly Is Not the Same as Artifact
The gap between “we cannot explain this” and “somebody built it” is where the field earns its credibility, mostly by refusing to close it. KIC 8462852, better known as Boyajian’s Star, showed irregular dips in the Kepler data that swallowed up to a fifth of the starlight on no discernible schedule. Megastructures got mentioned. Then follow-up showed the dimming was wavelength-dependent, which starlight blocked by a solid object would not be, and dust took over the explanation.
The same discipline shows up elsewhere. ‘Oumuamua, the interstellar object that passed through in 2017, generated an artificial-origin hypothesis that most of the community treats as unsupported. The claimed detection of dimethyl sulphide in JWST spectra of K2-18b, floated in 2023 and pressed harder in 2025, has been met with independent reanalyses arguing the feature is far weaker than advertised. Phosphine on Venus, announced in 2020, is still being argued about six years on.
None of these were embarrassments. They were the null hypothesis doing the job it exists to do.
The Bar for Yes
Confirmation would need more than a strange plot. It would need the thing to come back, on a different instrument, at a different site, run by people using unrelated software, with the source tracking the sky rather than the building. The Rio scale exists to keep the announcement proportional to the evidence, and it exists because the temptation to skip that step is real and human.
Underneath the verification protocols sits the unresolved question. We would not have found intelligence. We would have found a machine, or the exhaust from one, and inferred the rest. That inference is probably sound. It is still an inference, and it is worth saying out loud rather than burying in a footnote.
If that tension interests you more than a tidy answer would, SETIworld is a decent place to sit with it — the surveys, the retracted candidates, the arguments about what a detection would even mean. Come read, argue, and follow the next candidate through its inevitable debunking with the rest of us.
