Philosophy of Science
Why 'Is It Science?' Is a Harder Question Than It Sounds
Karl Popper thought he had solved philosophy's most stubborn boundary dispute in a single afternoon — he hadn't, and the fallout is still reshaping what we trust.
The Idea
The demarcation problem is the question of where science ends and non-science begins. It sounds administrative, like a border dispute between bureaucratic departments, but the stakes are surprisingly high: the answer determines which claims get taught in schools, which theories get funded, and which expert testimony is admissible in court. Popper's famous answer was falsifiability — a claim is scientific if, in principle, an experiment or observation could prove it wrong. Einstein's general relativity, for instance, made a bold prediction about starlight bending around the sun during a solar eclipse; it could have failed and didn't. Freudian psychoanalysis, by contrast, could absorb any counter-evidence into its framework. Unfalsifiable, said Popper. Therefore not science. Clean, elegant — and incomplete. The philosopher W.V.O. Quine pointed out that no single hypothesis ever faces evidence alone; it always arrives bundled with auxiliary assumptions. When an experiment fails, you can always blame the instruments, not the theory. Thomas Kuhn then showed that scientists in practice don't abandon theories just because anomalies pile up; they hold on until a better framework comes along. What this leaves us with is genuinely unsettling: there may be no single criterion that separates science from non-science across every case. Instead, science looks more like a cluster of overlapping features — testability, predictive power, openness to revision, institutional peer review — none of which are individually sufficient, and all of which admit of degrees.
In the World
The stakes of the demarcation problem became vivid in a courtroom in Little Rock, Arkansas, in 1981. The state of Arkansas had passed a law requiring public schools to give 'balanced treatment' to creation science alongside evolutionary biology. A federal judge, William Overton, had to decide: is creation science actually science? He turned to philosopher Michael Ruse, who laid out a set of demarcation criteria — science is guided by natural law, it is explanatory, it is testable, its conclusions are tentative and revisable, and it is falsifiable. Creation science, Ruse argued, failed on all counts: it invoked supernatural intervention, which cannot be tested or revised in response to evidence. Judge Overton agreed and struck down the law. But Ruse's criteria drew immediate fire from his fellow philosophers. Larry Laudan, one of the most respected philosophers of science at the time, wrote a scathing response arguing that Ruse had caricatured both science and creationism to reach a predetermined legal conclusion. Many creationist claims are falsifiable, Laudan noted — they're just false. And plenty of accepted science involves untestable auxiliaries. Laudan's point wasn't that creationism belongs in classrooms; it was that using a flawed philosophical criterion to win a legal argument was intellectually dishonest and would eventually backfire. It's a warning the scientific community has never quite fully absorbed: the boundary between science and non-science cannot be drawn by whoever needs to win the argument right now.
Why It Matters
This isn't just an academic puzzle for people who like arguing about definitions. We live in a moment when the authority of science is simultaneously more important and more contested than at any point in living memory. When someone dismisses climate projections, or promotes an alternative medical treatment, or questions the validity of psychological research, they're implicitly making a claim about what science is and what it can do. If you think science is defined by falsifiability alone, you'll run into trouble with string theory, which currently makes no testable predictions. If you think peer review is the gold standard, you have to reckon with replication crises in fields like social psychology. If you trust institutional consensus, you need a story about how institutions sometimes get captured by orthodoxy. None of this should push you toward the comfortable cynicism of 'it's all just opinion.' The better move is a more sophisticated trust — one that weighs predictive track records, the quality of opposing evidence, mechanisms of revision, and the specific history of a field. The demarcation problem doesn't dissolve science's authority; it asks you to understand where that authority actually comes from.
A Question to Ponder
If there's no single criterion that makes something scientific, what is it that actually earns your trust in a particular scientific claim — and is that the same thing that should earn it?
Get a new one of these every morning.
Start learning with Thinkable