When you read about a scientific finding, it can feel solid -- after all, it made it through peer review. But the studies that reach you have already passed through a filter that strongly favors exciting, positive results. Research that found nothing interesting often never gets published at all, which means the picture you see is systematically brighter than reality.

There is a strange kind of silence in science: the silence of studies that found nothing. Researchers who run experiments and get null results often just file them away. Those studies sit in desk drawers and hard drives, invisible to anyone trying to understand what the evidence actually says. The published record tells you what worked; the file drawer holds what did not -- and you never get to see it.

When someone funds a study, the results have a quiet tendency to come out in the funder's favor -- not because researchers are lying, but because money shapes questions, designs, comparisons, and emphasis in ways that are hard to see from the outside. The study looks methodologically sound. The bias lives in the choices that were made before any data was collected.

In science, being cited is a kind of currency -- and like any currency, it does not get distributed fairly. Studies that find positive, exciting results get cited far more than equally valid studies that find nothing. Over time, this creates a world where certain findings look foundational simply because everyone keeps referencing them, while contradictory evidence sits in plain sight, technically published but effectively invisible.

Imagine running a study that measures twenty things, finding that nineteen showed nothing and one looked promising. Now imagine writing the paper as though that one promising result was the whole point. That is outcome reporting bias -- the quiet practice of spotlighting the results that worked and dimming the lights on everything that did not, so the study reads like a targeted success rather than a broad search that got lucky once.

There is something deeply satisfying about a theory that perfectly explains the data -- and that satisfaction is exactly the problem. HARKing happens when a researcher looks at their results first, builds a theory to explain the pattern they found, and then writes the paper as though the theory came first and the data confirmed it. The narrative reads like prediction. It is actually hindsight wearing a lab coat.