We trust science. We fund it, teach it to our children, and invoke it when making decisions about health, climate, and technology. But what exactly is science? Where does it end and pseudoscience begin?

This question—the demarcation problem—has occupied philosophers for over a century. You might think the answer is obvious: science is what scientists do, or what follows the scientific method. But these intuitions crumble under scrutiny. Astrology once had university chairs. Bloodletting was medical consensus. Today's fringe theory might become tomorrow's paradigm.

The stakes aren't merely academic. Courts decide which expert testimony counts as scientific. Schools debate what belongs in biology class. Funding agencies allocate billions based on implicit judgments about scientific legitimacy. Understanding why this boundary is harder to draw than we assume reveals something important about how human knowledge actually works.

The most famous attempt to solve demarcation came from Karl Popper: science is what's falsifiable. Real scientific claims risk refutation by evidence. Astrology, by contrast, can explain any outcome after the fact, which makes it unfalsifiable—and therefore unscientific.

Elegant as this sounds, it fails in practice. Historical astronomy contained unfalsifiable auxiliary hypotheses. Darwin's theory of evolution wasn't immediately testable in Popper's strict sense. Meanwhile, clever astrologers can make falsifiable predictions—they're just usually wrong. Being wrong doesn't make something unscientific; being untestable supposedly does.

What about requiring empirical methods? Scientists observe and experiment, while pseudoscientists speculate. But historians use evidence too, and nobody calls history a science. Mathematicians proceed without experiments, yet mathematics underlies all physical science. The boundary keeps shifting.

Predictive success seems promising—real science makes accurate predictions. Yet string theory has produced no testable predictions in decades, while traditional Chinese medicine can predict certain treatment outcomes. We'd hesitate to call string theory pseudoscience or acupuncture genuine science. Every criterion we propose meets a counterexample that forces us back to the drawing board.

Takeaway

The failure of every simple demarcation criterion isn't a failure of philosophy—it reveals that science is a more complex, historically contingent practice than any single principle can capture.

Wittgenstein introduced the concept of family resemblance for categories that resist definition. Games, for instance, share no single common feature. Board games involve competition, but so does business. Children's games involve play, but so does jazz. Yet we recognize games when we see them because they share overlapping similarities—a web of features, none strictly necessary.

Perhaps science works the same way. Physics uses mathematics; biology often doesn't. Chemistry involves controlled experiments; astronomy mostly observes. Geology interprets historical traces; psychology studies present behavior. These disciplines share institutional features, methodological commitments, and standards of evidence—but different combinations in each case.

On this view, asking is this scientific? is like asking is this a game? There's no binary answer, only degrees of similarity to paradigmatic cases. Physics and chemistry sit comfortably at the center. Archaeology and economics occupy hazier territory. Astrology sits far from the core, resembling science in superficial ways while lacking its institutional accountability and track record.

This approach frustrates those seeking clear rules. It means demarcation involves judgment, context, and sometimes disagreement. But it better describes how working scientists actually recognize their practice—not through checklist verification, but through recognizing kinship with exemplary cases.

Takeaway

When a concept resists definition by necessary and sufficient conditions, it may be a family resemblance category—understood through paradigm cases and overlapping similarities rather than sharp boundaries.

You might conclude that demarcation is a philosopher's puzzle with no real-world significance. This would be a mistake. Consider courtrooms, where judges must decide whether expert testimony meets scientific standards. The Daubert standard in American law requires judges to evaluate methodology—but methodology varies across legitimate sciences.

Educational policy involves constant demarcation decisions. Should intelligent design appear alongside evolution in biology class? The question isn't whether intelligent design is true—it's whether it counts as science at all. Similar debates surround teaching climate change, vaccination, and human evolution.

Research funding operates through implicit demarcation. Grant committees distinguish promising heterodoxy from crankery, groundbreaking interdisciplinary work from dilettante dabbling. These judgments allocate billions of dollars and shape which questions get pursued. They rely on recognizing scientific legitimacy despite lacking precise criteria.

The absence of a perfect philosophical criterion doesn't mean anything goes. We can still recognize that homeopathy lacks the institutional accountability, evidential track record, and theoretical coherence that characterize good science. Demarcation functions less like a geometric boundary and more like a gravitational field—clear at the center, fuzzy at the edges, but still meaningfully structured.

Takeaway

Practical decisions about science don't require perfect philosophical criteria—they require informed judgment about how closely a practice resembles our best examples of knowledge production.

The demarcation problem resists solution because science isn't a natural kind waiting to be discovered. It's a human practice that evolved historically, varies across disciplines, and continues to change. No criterion will perfectly capture something that was never designed according to one.

This shouldn't breed cynicism. The failure to define science precisely doesn't mean science and pseudoscience are equivalent. It means our concept is richer than any simple formula can express—like art, or justice, or democracy.

Understanding demarcation as judgment rather than algorithm puts appropriate responsibility on the institutions that make these calls. Courts, schools, and funding agencies should cultivate the expertise to make good judgments rather than seeking mechanical rules. The question isn't what makes something scientific in the abstract—it's whether this practice, in this context, deserves the epistemic authority we grant to science.