Publication Bias
How science learned to hide its own failures.
1 min read · from UNINTENDED by Mayank Mehta
In science, there's an unspoken hierarchy of results. Discover something new, confirm a hypothesis, produce a striking effect, and journals are eager to publish your work. Find nothing, no effect, no difference, no breakthrough, and the work tends to disappear. Not because it's wrong. Because it's boring.
This preference, which sounds reasonable in isolation, has over time reshaped the entire scientific record.
Researchers learned early in their careers what the system rewarded. Positive results attracted publications, citations, grants, and promotions. Negative results attracted nothing. Slowly, and without anyone designing it this way, the published literature became a curated highlight reel. Studies that worked were visible. Studies that didn't were filed away in desk drawers, forgotten.
The consequences were most dangerous in medicine. Clinical trials that showed a drug working were far more likely to be published than trials that showed the drug did nothing. A treatment could appear effective simply because the failures were invisible. Doctors made prescribing decisions based on partial evidence. Patients trusted conclusions drawn from an incomplete picture.
The distortion compounded through meta-analyses, studies designed to aggregate all available evidence on a question. If the available evidence was already biased toward positive results, the meta-analysis inherited the bias. Science didn't lie. It just spoke selectively.
Nobody designed this system to produce distorted knowledge. Editors chased impact. Researchers chased publication. Funders chased results. Each decision, individually, was rational. Collectively, they produced a world in which failure wasn't disproven. It was simply unseen.
Publication bias is a quiet reminder that truth doesn't disappear when it's inconvenient. It just stops being visible. And when only success is allowed to speak, knowledge itself becomes quietly unreliable.