A primer on what science actually does when it works, why falsifiability matters more than verification, and where the framework cuts cleanly. Written in honor of the tradition, not against it.
Two British expeditions, one to Principe and one to Sobral in Brazil, set up telescopes to photograph a total solar eclipse. They were not there to see the eclipse. They were there to put a theory on the line.
Arthur Eddington led the team at Principe. The plan was simple in concept and excruciating in execution. During totality, the moon would block the sun for a few minutes. Stars near the sun's edge would become visible. Einstein's general theory of relativity, published in its final form three years earlier, predicted that the sun's mass would bend the light from those stars by a specific amount, roughly twice what Newton's theory predicted. Photograph the stars during the eclipse. Photograph the same stars again later, when the sun was nowhere near them. Compare the positions. The deflection would either match Einstein, match Newton, or match neither.
It matched Einstein. Within months, the news made Einstein a global celebrity, the first scientist to be a household name in the modern media age. But the more important point is the one most people miss when they retell the story. The experiment could have killed the theory. If the photographs had shown stars in their Newtonian positions, Einstein's framework would have taken a wound that even its author would have struggled to staunch. Einstein himself was clear about this. Asked what he would have said if Eddington's results had gone the other way, he replied that he would have felt sorry for the dear Lord. The theory was correct.
That posture, the willingness to specify in advance what would prove you wrong and then to expose your idea to exactly that test, is what a young Karl Popper, then a teenager in Vienna watching all this unfold, would later identify as the soul of science. Not the brilliance of the conjecture. Not the elegance of the math. The willingness to be killed by the world.
To understand Popper, you have to understand the project he grew up next to and ultimately refused to join.
In the 1920s, a group of philosophers, mathematicians, physicists and economists began meeting in Vienna, mostly in a room at the University. The group included Moritz Schlick, Rudolf Carnap, Otto Neurath, Hans Hahn, Friedrich Waismann, and others. Their ambition was breathtaking. They wanted to settle, once and for all, the question of what counts as a meaningful statement about the world. They called their program logical positivism, and their core tool was the verification principle.
The verification principle, in its cleanest form, said this. A statement is cognitively meaningful only if it can in principle be verified by sense experience, or if it is true by definition (as in pure mathematics and logic). Everything else, claims about God, claims about absolute moral truth, claims about Hegelian Spirit unfolding through history, was not false. It was something stranger. It was meaningless. Not even wrong, as the physicist Wolfgang Pauli would later put it. A noise with grammar.
You can feel the appeal. After the First World War, after the collapse of the old metaphysical certainties, after watching how grand systems like German Idealism had been turned into nationalist rhetoric, the desire to clean house was visceral. The positivists wanted to take the rigor of physics and the precision of mathematical logic and use them as a broom. Sweep out the nonsense. Keep only what could be checked.
The program ran into trouble almost immediately, and the trouble was the kind that does not go away with patching. Apply the verification principle to itself. The statement "a sentence is meaningful only if it is empirically verifiable or analytically true" is itself neither empirically verifiable nor analytically true. By its own standard, it is meaningless. The principle eats itself.
There were other problems. Universal generalizations of the kind science depends on, statements like "all metals expand when heated," cannot be verified by any finite number of observations. You can check a million metals; the million-and-first might behave differently. The positivists tried weaker formulations (confirmability instead of verifiability, probabilistic confirmation, observation languages versus theoretical languages) and each move bought a little time at the cost of a little coherence. By the 1950s, even sympathetic philosophers were treating logical positivism as a noble failure.
Popper grew up around all this. He attended the Circle's meetings occasionally, knew the players, took the project seriously. But he was never quite of it. He thought the positivists had the right enemies and the wrong weapon. The dream of cleanly separating sense from nonsense by an empirical test was a good dream. The verification principle was just not the way to do it.
Popper's Logik der Forschung appeared in Vienna in 1934. It was translated into English in 1959 as The Logic of Scientific Discovery. The argument is simple enough to state in a paragraph and deep enough to chew on for decades.
Start with a familiar example. The statement "all swans are white" was, for centuries of European observation, a perfectly respectable empirical generalization. People saw white swans. They kept seeing white swans. Each new white swan was added to the confirming pile. By the eighteenth century, "all swans are white" had the kind of weight that comes from millions of confirming instances.
Then, in 1697, Dutch explorers in Western Australia found black swans. Cygnus atratus. One observation, one bird, falsified the universal claim that had stood for centuries.
This is the asymmetry. No finite number of confirming instances can establish a universal claim. A single counter-instance can refute it. Confirmation is weak, almost decorative. Refutation is decisive. If you take the asymmetry seriously, it inverts the entire picture of what scientists are doing when they do good work. They are not, primarily, building piles of confirmations. They are stating bold claims that forbid certain observations, and then looking for those forbidden observations as hard as they can.
This is a stronger claim than it sounds. The positivists thought science worked by accumulating verified observations until a generalization emerged. Popper said no. The generalization comes first, as a conjecture, often a wild one. The work of science is then to try to kill it. The conjectures that survive serious attempts at refutation, especially attempts where they very nearly died, are the ones we keep, provisionally, while we keep trying to kill them.
Note the word provisionally. Popper was clear, in a way that vulgar Popperians often forget, that no scientific theory is ever proved. Theories are corroborated by surviving tests. Corroboration is not proof. Tomorrow's experiment might be the black swan. Even the most magnificently confirmed theory in physics, general relativity, is held with a kind of confidence that is structurally provisional. We hold it because nothing has killed it yet, and many things could have.
Popper's most influential single contribution was a criterion for telling one kind of inquiry from another. The criterion is not about subject matter and not about method. It is about a structural property of claims.
The demarcation problem, in its classical form, asks: what makes physics different from astrology, biology different from numerology, economics (sometimes) different from tea-leaf reading? The positivists had answered: physics is meaningful because its claims are verifiable; the others are meaningless because theirs are not. Popper recast the question. The right question, he said, is not whether a claim can be verified but whether it can be refuted.
A scientific claim is one that forbids something. Einstein's general relativity forbade the stars near the eclipsed sun from being where Newton said they would be. The theory put itself on the line. It risked being killed by an observation, and the observation could have killed it. That risk-taking is the mark of a scientific theory. It is a theory with skin in the game.
Now consider astrology. A horoscope tells you that as a Capricorn you are reserved and ambitious. If you behave reservedly, the horoscope is confirmed. If you behave wildly, the horoscope is confirmed in the form of "you are working against your nature" or "Mercury is in retrograde, masking your true temperament." Whatever happens, the horoscope is confirmed. The claim forbids nothing. Therefore it tells you nothing about the world. It can feel illuminating without conveying information, the way a flattering Rorschach reading can feel illuminating.
This is a much sharper instrument than the verification principle. Astrology is not meaningless. The words have meanings, the sentences parse. What astrology lacks is the kind of meaning that distinguishes a description of the world from a description that would be equally true of any other world. A claim that is compatible with every possible observation tells you which world you are in by telling you nothing.
The criterion has a beautiful side effect. It locates the virtue of science not in any one experiment, but in the attitude scientists take toward their own theories. A scientist who proposes a bold theory and then tries to kill it is doing the work. A scientist who proposes a vague theory and then explains away every awkward result is not, regardless of credentials. The criterion is methodological and ethical at the same time. It is a description and a discipline.
Popper's most famous applications of the falsifiability criterion were not to astrology, which everyone already suspected, but to two of the most influential intellectual systems of the twentieth century: psychoanalysis and Marxism. The critique landed, and it still lands, but only against certain forms of those systems.
In his autobiographical essay "Conjectures and Refutations," Popper recounts the moment in 1919 when his disillusionment with Freud and Marx crystallized. He had been an enthusiast for both. He worked briefly with Alfred Adler in Vienna, doing volunteer social work with troubled children. What struck him, increasingly uncomfortably, was that the psychoanalytic explanations seemed to fit every case. A child who hit other children was acting out an Oedipal aggression. A child who refused to hit other children was repressing an Oedipal aggression. Whatever the child did, the theory had a slot for it.
Popper compared this with what he was hearing about Einstein's relativity in the same period, which was about to be tested by Eddington and which made specific, narrow, dangerous predictions that could have killed it. The contrast was not in the brilliance of the thinkers. Freud and Marx were obviously brilliant. The contrast was in the relationship of the theory to the world. Einstein's theory said: if X happens, I am wrong. Freud's theory, as it was actually practiced, did not seem to say that.
Be careful with this critique, because Popper himself was. He was not saying Freud was stupid, or that there were no important psychological insights in psychoanalysis, or that the unconscious does not exist. He was saying that the specific claims, in the form they were made, did not put themselves at risk in the way scientific claims do. They functioned more like interpretive frameworks than predictive theories. The frameworks might be illuminating, even profoundly so, but calling them science was a category error.
The critique cuts hardest against the most ambitious versions of early psychoanalysis: claims about specific dream-mechanisms (a falling dream means X, a flying dream means Y), claims about the universal structure of childhood development, claims about the specific etiology of specific neuroses. Contemporary research in clinical psychology, including the parts that owe much to the depth-psychology tradition, has by and large taken the lesson on board. Therapeutic frameworks are tested for outcomes; specific mechanistic claims are constrained by what brain imaging and behavioral data can actually show.
The lesson is not that the depth-psychology tradition is worthless. Anyone who has worked with their own dreams seriously, or done IFS work with a part of themselves, or watched a Jungian analyst gently reframe a recurring image, knows that something real is happening in that tradition. The lesson is that the kind of knowledge being generated there is not the same kind of knowledge being generated in a particle accelerator, and pretending otherwise serves neither tradition.
The critique of Marxism follows the same logic. Marx made specific historical predictions. Revolution would arrive in the most industrialized capitalist countries first. The proletariat would become progressively immiserated. The state would wither. When the predictions failed, Marxist theorists did not, by and large, treat the failures as falsifications. They added auxiliary hypotheses. The revolution had not come because of "false consciousness," because of "imperialism exporting the contradictions," because of "the labor aristocracy." Each rescue saved the theory locally and weakened it as a piece of science. The system became unfalsifiable not by being meaningless but by being infinitely flexible.
Popper's critique of Marxism, expanded in The Open Society and Its Enemies (1945), was politically explosive in a way the Freud critique was not. Popper had escaped Vienna for New Zealand in 1937, just ahead of the Anschluss. He wrote The Open Society during the war as what he called his "war effort." The book attacks Plato, Hegel and Marx as ancestors of totalitarianism, philosophers whose claims to know the unfolding logic of history were used to justify dispensing with the open, fallible, piecemeal reform that Popper saw as the only humane politics. The argument is more than an application of falsifiability. It is a moral argument about what political claims can responsibly do. We will come back to it.
Falsifiability is the famous part of Popper. The deeper part is what falsifiability is in service of. Popper's positive vision is of science as a particular kind of evolutionary process: bold conjectures, ruthless attempts at refutation, the slow accumulation not of truths but of useful errors discarded.
The image is biological. In nature, organisms produce variation; the environment ruthlessly selects against the unfit; what survives is what was, in that environment, viable. The unfit do not get to reason their way back into existence. They are simply gone, leaving the field to their less unlucky cousins. Popper saw scientific theories as occupying an analogous position. A scientist proposes a bold conjecture (the variation). Other scientists, and the world itself in the form of experiments, attempt to refute it (the selection). What survives, for now, is what has not yet been killed.
This image cuts against several flattering pictures of science that scientists themselves sometimes tell. It cuts against the picture of science as the patient accumulation of facts; in Popper's view, facts are interesting mostly when they wound a theory. It cuts against the picture of the scientist as a neutral, presupposition-free observer; in Popper's view, all observation is theory-laden, because you cannot decide what to observe without some prior conjecture about what would matter. It cuts against the picture of induction as the engine of science; in Popper's view, induction does not exist, or rather, the move from "I have seen many white swans" to "all swans are white" is not a logical move but a guess, and guesses earn their keep by surviving tests.
Late in his career, Popper extended the picture in a direction that startled some of his admirers. In Objective Knowledge (1972), he proposed an ontology of three worlds. World 1 is the physical world. World 2 is the world of subjective mental states. World 3 is the world of objective contents of thought: theorems, theories, problems, arguments, the contents of libraries. World 3 objects, Popper argued, are real. They have causal powers. A mathematical theorem, once stated, has consequences that no one has yet thought of, and those consequences are there to be discovered rather than invented. Science is, on this view, the exploration of World 3 by means of World 1 experiments conducted by World 2 minds.
You do not have to buy the three-worlds ontology to see what Popper was after. He was insisting, against a certain kind of psychologism, that knowledge is more than what is going on in some individual head. Knowledge has a public, criticizable, improvable structure. A theory is something you can hand to a stranger and they can attack it. That handing-over and that attacking are how knowledge grows. Scientific knowledge, on Popper's view, is the part of human knowledge that has organized itself to make this handing-over and attacking as efficient as possible.
Falsifiability, taken as a clean prescription for how science actually proceeds, ran into trouble of its own. The trouble came from sympathetic critics who took Popper seriously enough to argue with him for decades. The mature view of scientific method that emerges is more complicated than Popper's, but it is still recognizably Popper's.
Thomas Kuhn published The Structure of Scientific Revolutions in 1962, and the book changed how educated people talk about science. Kuhn's point, drawn from close reading of the actual history of physics and chemistry, was that scientists do not, in fact, spend most of their time trying to refute the dominant theory. Most of the time they do what Kuhn called "normal science": puzzle-solving within an accepted paradigm. The paradigm sets what counts as a real problem, what counts as a good solution, what instruments are reliable. Anomalies, observations that do not fit, accumulate quietly for a long time before they trigger a crisis, and a crisis is resolved not by a clean falsification but by a messy revolution in which a new paradigm displaces the old. Newton was not falsified into Einstein. Newton was, over a long generation, replaced by Einstein.
Kuhn was widely misread as a relativist. He was not one; he insisted that science makes genuine progress, even if not in the cumulative way the textbooks suggest. The point that stuck was the sociological one. A scientific community is a community, with norms, prestige structures, training routines and shared problems. The reigning theory has institutional weight, and that weight is not entirely a corruption. Some of it is what allows productive work to happen at all. Popper's image of the scientist as a perpetual revolutionary, always trying to overthrow the current consensus, is at best a caricature of any human community that has ever existed.
Imre Lakatos, a student of Popper's who became his most rigorous friendly critic, proposed a middle way. Science, Lakatos said, proceeds not by single theories tested against single observations but by research programmes. A research programme has a "hard core" of central assumptions that the community treats as off-limits to refutation, and a "protective belt" of auxiliary hypotheses that can be adjusted to handle apparent counter-evidence. The hard core of Newtonian mechanics, for instance, included the inverse-square law. When the orbit of Uranus did not match predictions, the protective belt added the hypothesis of an undiscovered planet, and Neptune was duly found. The protective-belt move had saved the programme productively. By contrast, when the orbit of Mercury also misbehaved, the same trick (the postulation of a planet "Vulcan") did not pay off, and eventually Einstein's reworking of the hard core itself was needed.
Lakatos's criterion is not falsification at a single test but the long-term character of the programme. A progressive programme is one whose protective-belt adjustments predict novel facts that turn out to be true. A degenerating programme is one whose adjustments only ever explain away the awkward results after the fact, predicting nothing new. Marxism, in Lakatos's view, had become a degenerating programme. Einsteinian relativity, by contrast, had been progressive for decades. The judgement is not made on a single experiment but on the trajectory of the research.
Paul Feyerabend, also for a time in Popper's circle, took the critique further than Popper could stomach. In Against Method (1975), Feyerabend argued, with deliberately provocative flair, that there is no single scientific method, that every methodological rule has been violated by successful science at some point, and that the only methodological principle that survives the historical record is "anything goes." Feyerabend's point was not that science is bad. His point was that the actual history of science is so heterogeneous, so dependent on lucky guesses, rhetorical maneuvers, ad hoc rescues and aesthetic preferences, that the philosophers' attempts to legislate a single Method are themselves unscientific. They do not match the data.
You do not have to swallow Feyerabend whole to learn from him. The lesson is that a clean rational reconstruction of scientific method, of the kind Popper attempted, is always a stylization. Real science is messier. Pretending otherwise is a small failure of the very honesty Popper was trying to enforce.
And then there is the replication crisis, which has been the largest applied lesson in philosophy of science in our generation. Starting around 2011 with Joseph Simmons and colleagues' "False-Positive Psychology" paper, and consolidating with the Open Science Collaboration's 2015 report that successfully replicated only about 36 percent of 100 prominent psychology studies, it has become widely accepted that large stretches of published "scientific" findings, especially in social psychology, biomedical research and some areas of economics, do not survive serious attempts at re-testing. The crisis has spread, in varying degrees, to nutrition science, cancer biology, machine learning evaluation, and beyond.
The replication crisis is, in a sense, Popper's revenge. It shows that when communities of researchers stop genuinely trying to refute their own results (because of publication incentives, career pressure, p-hacking, the file-drawer effect) the findings they produce do not have the resilience that scientific findings are supposed to have. The cure, the open-science movement's emphasis on pre-registration, replication, public data, and adversarial collaboration, is essentially a reinstallation of Popperian discipline in places where it had quietly lapsed.
Three generations of argument later, no serious philosopher of science holds a pure falsificationism, and no serious one holds a pure relativism either. The working synthesis is messier and, I think, more interesting than either pole.
Science is a social-epistemic enterprise. It involves bold conjectures (Popper); rigorous testing where rigorous testing is possible (Popper, again); peer review and replication (institutional norms that mostly date from the twentieth century and that are still being refined); the ability of a community over time to detect and discard error (Kuhn); the long-term assessment of research programmes as progressive or degenerating (Lakatos); and an acknowledgement that the practice is plural, historical, and humanly fallible (Feyerabend, taken in moderation).
The reason this synthesis is hopeful rather than deflationary is that it explains why science actually works. Newtonian mechanics ran the Industrial Revolution and got us to the moon. General relativity gives us GPS. Quantum mechanics gives us semiconductors. Antibiotics, vaccines, the structure of DNA, the age of the universe: these are not "socially constructed" in the sense some careless writers have suggested. They are real findings, produced by a real method that, when it functions, is the most powerful epistemic engine humans have ever built. The point of the synthesis is not to weaken any of that. It is to describe it accurately enough to keep it working.
The phrase I find most useful is one Susan Haack uses: science is not a method, it is a continuum of inquiry, the part of inquiry that has organized itself most successfully to expose its claims to evidence and to correct them in light of that evidence. The continuum runs from rigorous physics through messier biology through messier-still psychology through messier-still economics, and it does not stop there. Detective work, careful history, investigative journalism all sit on the same continuum, distinguished from physics not by kind but by how much of the Popperian discipline is available given the subject matter.
The synthesis does not soften the blade where the blade still cuts. Several contemporary disputes are clarified, sometimes painfully, by asking the Popperian question: what would this claim look like if it were wrong, and what is the community doing to check?
Parapsychology has been studied seriously, with reasonable methods, for over a century. Telepathy, precognition, psychokinesis. The Society for Psychical Research was founded in 1882. The field has been generously funded at moments, has had prestigious advocates, and has produced thousands of experiments. The honest summary is that the field does not progress. Effects appear in some studies, fail to replicate in others, shrink as methodology tightens, and never get to the point of letting anyone do anything they could not do before. By the Lakatosian criterion, this is a degenerating programme. By the Popperian criterion, the theories on offer have shown themselves capable of accommodating any result. The field is not "wrong" in the way phlogiston theory was wrong; it is wrong in the way astrology is wrong, by failing to expose itself to the kind of test that could matter.
Young-earth creationism is a clearer case. The claim that the earth is roughly six thousand years old is in direct conflict with every dating method we have, with the structure of the geological record, with genetic evidence, with the speed of light. The response of young-earth creationists has, with notable consistency, been to add auxiliary hypotheses ad infinitum. The light from distant galaxies was created in transit. The geological record was laid down by the Flood. Radioactive decay rates were different in the past. Each rescue saves the theory at the cost of any predictive content. Nothing in geology, biology or cosmology could, in principle, refute the position, because every observation can be reinterpreted as part of the divine design. This is not a difference of opinion to be settled by more evidence. It is a structural feature of the theory.
Adaptationist just-so stories in some forms of evolutionary psychology. The critique here, made forcefully by Stephen Jay Gould and Richard Lewontin in their 1979 "Spandrels" paper, is that not every human behavior needs an adaptive explanation, and that the practice of inventing one for any behavior you happen to notice can produce a literature that is more like folklore than like science. A claim that "men prefer waists with a 0.7 ratio because in the Pleistocene this signaled fertility" is doing real evolutionary biology only if it forbids something, makes a novel prediction, can be tested against cross-cultural data, can be checked against the actual archaeological and genetic record. Many such claims are not. They are decorative stories told over the same data. Some evolutionary psychology is rigorous; some is not, and the Popperian question is the most efficient way to tell the difference.
Certain alternative medicines. Homeopathy makes a specific physical claim (that water "remembers" substances diluted out of existence) that is in conflict with what we know about chemistry and that has been tested, repeatedly, with consistent negative results. Defenders respond not by abandoning the claim but by reformulating it in ways that drift toward unfalsifiability ("the effect is too subtle to measure," "the wrong outcomes were measured," "the patient's mental state matters"). This is the structural pattern Popper identified. The pattern does not prove homeopathy false; it identifies the kind of intellectual move being made to keep it alive. Discrimination here is not gatekeeping. It is honest.
The point of these examples is not to score points against anyone in particular. The point is that Popper's criterion is a real instrument that does real work, and that work is genuinely valuable in distinguishing claims that are accountable to evidence from claims that have, intentionally or otherwise, slipped that accountability. Where the blade applies, it should be used.
The next two pieces in this series take up at length the question of what kinds of knowledge falsifiability cannot reach. I will only flag the point here, because it matters for an honest reading of Popper himself.
Falsifiability is a criterion for empirical science. It is a criterion for the kind of claim that is supposed to describe how the natural world is, in a way that the natural world could push back against. It is not a criterion for ethics, for aesthetics, for meaning-claims about a life, for the description of subjective experience, for poetry, for the question of how one should treat one's father, for the question of whether a given symphony is great, for the question of what it is like to wake up at 3 a.m. with a fear you cannot name.
Some popularizers, and some scientists in their less careful moments, have extended the criterion beyond its proper domain and concluded that any claim that cannot be empirically falsified is meaningless or unworthy of serious thought. This is a vulgarization, and it is one Popper himself would have rejected. The Open Society and Its Enemies is a work of political philosophy and ethics, full of normative claims that Popper did not pretend were "scientific" in his own sense. He was clear that the scientific method is one form of rational inquiry among others, suited to certain kinds of question and not to others. The mistake of confusing the criterion for science with the criterion for meaning is not Popper's. It is a mistake of his less careful readers, and it has done real damage in the wider culture.
I will say more about all this in the second piece. For now: the strength of Popper's criterion within its domain is a reason to apply it carefully, not a reason to apply it everywhere. A scalpel is better at being a scalpel because it is not also a hammer.
Even within the domain that science indisputably owns, there is at least one question where the most rigorous physicalists openly admit their method does not yet reach. The question is consciousness.
In 1974, the philosopher Thomas Nagel published "What Is It Like to Be a Bat?" in The Philosophical Review. The argument is short and famous. A complete physical description of a bat's brain, sonar system, neural firing patterns, evolutionary history, would tell you everything about the bat from the outside. It would not tell you what it is like to be the bat from the inside. The fact that there is something it is like to be the bat (the subjective quality of its experience, the felt-from-the-inside) is a fact that physical description, however thorough, leaves out. Nagel did not conclude that physicalism is false. He concluded that we do not yet have the conceptual tools to understand how subjective experience relates to the objective physical world.
In 1995, the philosopher David Chalmers sharpened the point in a paper that introduced the now-standard distinction. The "easy problems" of consciousness are the problems of explaining how the brain performs various cognitive functions: discriminating stimuli, reporting on internal states, integrating information, controlling behavior. These problems are hard in the colloquial sense, but they are easy in the sense that we know what kind of explanation would solve them. We need a story about neural mechanisms. The "hard problem" is to explain why any of this is accompanied by experience at all. Why is there something it is like to see red, rather than mere information processing of red wavelengths? No story about neural mechanisms, by itself, seems to answer this question. It seems to be a question of a different kind.
The hard problem is contested. Some philosophers (Daniel Dennett most famously) think it dissolves on closer examination. Some think it will yield to future neuroscience. Some think it requires a fundamental revision of our physics. The point for our purposes is not to take a side. The point is that even within the most rigorous traditions of analytic philosophy and cognitive neuroscience, there is a live recognition that the subjective, the experiential, the felt-from-the-inside, is not yet captured by the objective, third-person description that science excels at producing. This is hard science conceding, openly, the limits of its current reach.
I flag this here only as a preview. Piece 3 takes it up in earnest. For now: the bat is the point. There is something it is like to be the bat. Whatever that something is, it is real, and any honest account of knowledge has to find a place for it.
Science, when it works, is the best machine we have ever built for finding out which of our beliefs about the natural world are wrong. Popper gave us the cleanest articulation of what that machine is doing and why it has the power it does. The machine is not magic. It is a disciplined, social, fallible practice of putting bold guesses at risk and being willing to lose. The successors and critics of Popper added crucial detail: the practice happens within paradigms and programmes and communities, not as a series of isolated decisive tests; the machine can corrode when communities stop genuinely trying to refute themselves; the method is plural and historical and not reducible to a single rule.
None of these qualifications undermine the achievement. They locate it. Science is a particular tool, the best tool we have for a particular kind of question. Honoring science means understanding what it is for, what it is uniquely good at, and where it is honestly admitting its limits.
The next piece in this series turns to a different family of questions. Not how the world is, but how to live in it. Not mechanism, but meaning. The tools for that work are not Popper's, and pretending they should be is a kind of disrespect to both traditions. But the rigor and honesty that Popper modeled, the willingness to expose your beliefs to what could refute them, is portable. It transfers. The next piece is not a retreat from rigor. It is rigor applied to a different terrain.
Real books that exist. Pick from any cluster according to taste. The Popper himself cluster is the place to start if you want to read one thing.