It looked good. Printed in Nature, no less. But Till Sawala read that headline two weeks ago and felt his stomach drop.
“I thought, ‘Okay, either this is the most important thing in a decade, or it is just wrong,’” the University of Helsinki cosmologist says.
His bet? Wrong.
Always is.
The paper claimed to shatter decades of cosmic consensus. It used data from the Dark Energy Spectroscopic instrument, or DESI. Forty-seven million galaxies. Quasars stretching back eleven billion years. The usual stuff, really—filaments and voids forming that familiar cosmic web. Except this team said the web wasn’t random. They claimed the filaments aligned. Preferentially. In specific directions across billions of light years.
That breaks the cosmological principle. The rule that the universe looks roughly the same in every direction at large scales. If the data proved otherwise, everything changes.
“If we missed that,” Sawala says, “the community has some explaining to do.”
He didn’t think they had missed it. He thought they made a calculation error.
An elemental one.
The Nature authors used “luminosity distance” to map the galaxies. Sawala says you can’t do that. Not without adjusting for the universe expanding while the light traveled. You need “comoving distance” to fix that expansion. The original authors forgot to scale the data.
It sounds technical. It isn’t really.
It is a math mistake.
Once Sawala applied the correct metric, the mystery vanished. No alignment. No violation. The data settled back into the consensus. The universe remained boringly uniform.
Francesco Sylos Labini of the Enrico Fermi Center, a co-author, pushed back. He argued that orientation mattered more than the patchiness Sawala focused on. Sawala disagreed. The mistake stood, regardless of perspective.
So Nature published it. Why?
Because extraordinary claims get space in top journals. It is their job to feature groundbreakers. “To be in Nature you have to be groundbreaking,” Sawala notes. This was definitely that. But breakthrough status is not a synonym for correctness.
David Spergel of the Simons Foundation is less gentle. “Disappointing,” he calls the oversight. He thinks Nature’s editors should tighten up.
Sawala isn’t sure he would have caught it even if he had reviewed the paper. He admits peer review is broken by design. Reviewers know their niche, not the whole code base.
Daniel Eisenstein at Harvard agrees. These bugs hide. They sit in code for years, waiting.
It is easy to see how this sat unnoticed for so long.
The irony stings. A sensational claim makes the rounds, grabs headlines, and then—once debunked—disappears from view. The public remembers the shock, not the correction. Science moves forward, but the narrative sticks around like gum on a shoe.
This is why physicists prefer preprint servers. ArXiv.org allows the whole room to read the draft at once. One or two anonymous referees? That is a lottery. An open community? Someone, eventually, finds the bug.
This specific paper? It wasn’t posted to ArXiv before Nature published.
They kept it under embargo. A secret shared only with journalists days before the drop. It makes for better press. It makes for a cleaner narrative arc.
It does not make for better science.
“I think these embargos serve the publication,” Sawala says. He pauses. “Not the science.”
