Only 6.6% of astrobiologists surveyed after the K2-18b announcement agreed that scientists had probably found extraterrestrial life. Nearly two-thirds disagreed outright.
That gap between the headlines and the people who actually do this work is the finding here, and it’s the part almost nobody reported.
Because 2025 looked, from the outside, like the year we got close. A press release claimed the “strongest hints yet” of extraterrestrial life on the exoplanet K2-18b. Months later, discussing a sample collected from a rock named “Cheyava Falls” on Mars, NASA Administrator Sean Duffy said this was the “closest we have ever come” to discovering life on the red planet.
Nobody actually counts
Here’s the odd thing about how science reaches you. When a controversy or a breakthrough takes over the news cycle, press officers and journalists quote a handful of experts. Those views can be sharp and useful. But they rarely tell you what the wider research community thinks.
And yet public argument leans constantly on “the science says” or “scientists believe”, as if someone had run the numbers.
Usually nobody has. Systematic evidence about scientific opinion is mostly missing. My colleagues and I tried to fix that for astrobiology, surveying hundreds of astrobiologists from across the global research community within days of each of the two big 2025 announcements. One question: did scientists think extraterrestrial life had probably been found?
Two claims, two very different reactions
The first case was K2-18b. In April 2025, researchers reported possible traces of molecules called dimethyl sulfide and/or dimethyl disulfide. On Earth, those are associated with biological activity. Coverage was heavy, and plenty of it framed the result as an extraordinary step forward in the search for alien life.
The second came in September, when NASA announced that Cheyava Falls appeared to preserve a potential biosignature. The so-called “leopard spots” are mineral rings that on Earth are often formed by microbial activity. Headlines ran hot again, and so did NASA officials.
The survey numbers didn’t match the volume. For K2-18b: 6.6% agreed, nearly two-thirds disagreed, 28.0% stayed neutral. For Mars, confidence climbed but stayed guarded, with 15.1% agreeing, disagreement down to 44.6% and neutrality up to 40.3%.
The number that tells the real story
Read only the agree and disagree columns and you miss what moved.
Strong disagreement collapsed. It went from 35.1% on K2-18b to 11.1% on Mars. Most of that shift wasn’t rejection turning into endorsement. It was hard rejection loosening into something more tentative.
Expert opinion moved, in other words, but it moved in a structured way. The community got more open to the possibility without signing up for it.
Part of that is probably the evidence itself. K2-18b rested on possible atmospheric signatures picked up across interstellar distances. The Martian case involved a rock that can be studied directly, in far more detail. Astrobiologists have also known for a long time that lifelike features can arise through non-biological processes. The hard part usually isn’t imagining how life might produce a signal. It’s accounting for every way nature could fake one.
Five answers, not two
Public discussion tends to treat a scientific community as a switch: agrees, or doesn’t. The distribution matters just as much. Strong agreement, agreement, neutrality, disagreement and strong disagreement each say something different about how a field is absorbing a claim.
A big neutral block can mean several things at once. Scientists may find the evidence genuinely inconclusive. They may sit at some middle level of confidence. Or they may think a claim is too speculative to either endorse or reject. Movement from strong disagreement to ordinary disagreement can signal softening even while overall disagreement stays high. Flatten all of that into “for” or “against” and you lose the information.
This isn’t really about aliens
The same problem shows up in climate science, pandemics, artificial intelligence and medical research, where public conversation invokes scientific consensus constantly.
Sometimes that consensus is real. Sometimes it isn’t. Either way we rarely have a systematic way to measure what scientists actually think, particularly where evidence is still emerging or uncertainty runs deep. What fills the gap is selective quotation, the loudest individuals and assumptions about what everyone else believes.
Some of that is starting to change. At Durham University, our research group C-Scope, the Centre for Scientific Community Opinion Polling and Evaluation, studies how expert opinion is distributed and how it shifts over time. We’re not trying to replace evidence with polling, and we don’t treat majority opinion as truth. The point is to understand how scientific communities handle uncertainty.
Science advances through uncertainty, disagreement and slow revision. So the next time a press release tells you we’ve never been closer, ask a narrower question: closer according to whom, and how many of them? On K2-18b, the answer was 6.6%.
