In July 1965, Karl Popper and Thomas Kuhn faced each other at an International Colloquium in London organized specifically to debate their competing views of science. Popper had spent decades arguing that the hallmark of science is the willingness to falsify — to make bold conjectures and ruthlessly eliminate those that fail empirical testing. Kuhn had just argued that most scientists do precisely the opposite — that they cling to their paradigms in the face of anomalies and only abandon them when forced to by overwhelming crisis. The two men agreed on almost nothing, and their disagreement remains the most productive fault line in twentieth-century philosophy of science.
The Kuhn-Popper debate is not only historical. It bears on pressing questions in contemporary science: What should we make of the replication crisis in psychology? Is AI research a paradigm in Kuhn's sense, and if so, is it in crisis? What does the continued failure to detect dark matter or develop a quantum theory of gravity tell us about fundamental physics? To answer these, you need both Kuhn's and Popper's tools and a clear sense of where they diverge.
Popper's falsificationism
Karl Popper's philosophy of science is organized around a single guiding principle: a theory is scientific if and only if it is falsifiable — if it makes predictions precise enough to be definitively ruled out by possible observations. Science advances through what Popper calls "conjecture and refutation": scientists propose bold, risky hypotheses; experiments attempt to falsify them; falsified hypotheses are eliminated; survivors are provisionally retained — not as true, but as the best available conjectures. On this picture, the proper attitude of a scientist is permanent critical openness — the willingness to abandon any theory if the evidence demands it. A scientist who protects their theory from testing by adding ad hoc modifications to explain away disconfirming results is not doing science; they are doing what Popper called "degenerative research" — clinging to a theory past its useful life.
Popper's model is normative: this is how science should work, not necessarily how it always does. But Popper believed that the best science in history — Newton's mechanics, Einstein's relativity, Darwin's evolution — had in fact proceeded by this method, and that this was the explanation of science's success.
Kuhn's counter
Kuhn attacks Popper on both the historical and the normative levels. Historically: "No theory ever solves all the puzzles with which it is confronted at a given time; nor are the solutions already achieved often perfect." If scientists abandoned theories whenever they faced anomalies, no theory would survive long enough to be useful. Newton's mechanics faced anomalies throughout its history (the precession of Mercury's perihelion, for example, was unresolved for decades); the rational response was not to abandon it but to work harder within it. The anomaly was not evidence of fundamental failure — it was a puzzle to be solved. Kuhn's account of normal science describes a practice that is, in Popper's terms, "unscientific" (it protects the core theory from falsification) but that is historically the norm in mature sciences. Methodologically: Kuhn argues that Popper's falsification principle cannot actually guide scientific practice because there is no theory-neutral observation language in which falsifying observations can be described. Observations are always already interpreted through a theoretical framework; what counts as a "falsifying instance" depends on what paradigm you are working within. The deeper disagreement
The sharpest point of disagreement concerns the value of normal science. For Popper, periods where scientists do not critically test their core assumptions are periods of stagnation — failures of scientific rationality. For Kuhn, those periods are conditions of productivity — the paradigm's stability enables the focused, cooperative work that makes genuine discovery possible. As the Stanford Encyclopedia summarizes: "Popper contested that periods of 'normal science' are not normal but undesirable aberrations and that science can and should aim to maintain a perpetual state of Kuhnian revolution." Kuhn found this picture not only historically inaccurate but practically absurd: a community in permanent revolutionary crisis would never accomplish anything. The two views are not, ultimately, incompatible at every level — both reject naive positivism, both emphasize the role of bold theoretical speculation, and both insist that science is a rational enterprise rather than mere data-collection. But their differences about the role of paradigms, the value of normal science, and the mechanism of progress are deep enough to generate importantly different views about what good science looks like and what contemporary scientific controversies tell us.
Starting around 2011 and accelerating through the mid-2010s, the psychological sciences began confronting what has come to be known as the replication crisis: a large proportion of prominent findings — including famous studies on priming effects, ego depletion, power poses, and social facilitation — failed to replicate under independent testing. The Open Science Collaboration's 2015 study attempted to replicate 100 published psychological experiments and found that only about 36-39% reproduced the original results with similar effect sizes. How do Kuhn and Popper frame this differently?
Popper's framing: The replication crisis reveals that much of social psychology was insufficiently falsificationist — researchers used small samples, flexible analysis strategies (p-hacking), and publication bias to produce results that appeared to confirm their theories without genuinely testing them. The crisis is a correction — scientific rationality asserting itself against degenerative research practices. The solution is more rigorous testing, pre-registration, and willingness to abandon confirmed "findings" when they fail replication.
Kuhn's framing: The replication crisis is better understood as a paradigm crisis. The dominant paradigm in social psychology — centered on laboratory priming experiments, small samples, and theoretically rich but methodologically light experimental designs — accumulated anomalies (failed replications, inflated effect sizes) until the community's confidence in its core assumptions began to erode. What we are now witnessing is not simply methodological correction but the early stages of a paradigm shift: toward larger samples, open data, pre-registered designs, computational modeling, and a different set of theoretical commitments about what psychology should explain and how. The Kuhnian framing captures something the Popperian framing misses: the resistance that established researchers have shown to methodological reform is not simply stubbornness or dishonesty — it is the normal response of practitioners deeply invested in a paradigm who see the challenges as puzzles to be solved within the existing framework (better statistical methods, more sensitive designs) rather than as evidence of fundamental failure. The generational dimension is visible: younger researchers have been the primary drivers of the open science movement; older researchers with established reputations in the old paradigm have often been its most vocal critics.
Applying Kuhn to contemporary science raises a difficulty: When does an anomaly become a crisis? Kuhn's account of this transition is vague. He describes crisis in phenomenological and sociological terms (confidence erodes, rules blur, alternatives proliferate) but gives no threshold for when anomalies trigger crisis. That vagueness means his framework works better as a retrospective explanatory tool than a predictive one. We can see, in retrospect, that geology was in crisis in the 1950s; it is harder to say, in the present moment, whether a given field has reached a crisis state or is merely experiencing a productive period of puzzle-solving.
String theory presents a particularly thorny case. For decades, string theory has been the dominant approach to quantum gravity and has attracted enormous intellectual resources. It faces the deep anomaly of making no testable predictions in the energy ranges accessible to current experiments. Some critics argue that string theory is a paradigm that has entered crisis without producing a successor — a state of protracted anomaly that Kuhn's model does not clearly account for. Others argue that string theory is still in a normal science phase, with the testability problem a puzzle to be solved rather than an existential threat.
The Kuhn-Popper debate has been productive because both positions capture something real about science. Popper is right that critical testing is essential to scientific rationality and that theories which cannot be tested are scientifically problematic. Kuhn is right that real scientific practice involves extended periods of normal science within stable frameworks, that paradigms are not abandoned at the first anomaly, and that the social and psychological dimensions of science are genuinely epistemically relevant.
The most productive contemporary approaches to philosophy of science — including Imre Lakatos's research programmes, Philip Kitcher's division of cognitive labor, and Helen Longino's social epistemology — draw on both figures, trying to capture the genuine insights of each while avoiding the excesses of pure falsificationism and pure paradigm-relativism. For the learner, the key takeaway is this: Kuhn does not undermine science's rationality or authority. He enriches our picture of how science actually advances — not as a smooth machine of conjecture and refutation, but as a complex social practice that is profoundly human in its organization, its psychology, and its history, while remaining genuinely progressive and epistemically powerful.
The interactive steps that follow ask you to apply Kuhn's framework to specific scientific cases, test his arguments against Popper's, and reflect on what the philosophy of science reveals about rationality and progress in knowledge.