Science advances by revolutions, not steady fact-gathering

1

How fields become sciences

A field is not mature science until practitioners share one framework of assumptions, tools, and exemplary problems; before that there are only competing schools.

Kuhn rejects the image of science as a continuous march that simply gathers facts. In a field's early life, rival camps disagree on basics. Optics before Newton had schools that did not even share a view of what light is, and no contemporary standard cleanly ranked them.

What ends the scatter is not one decisive experiment. It is the rise of a paradigm: shared assumptions, methods, model solutions, and standards for a good answer. The paradigm is broader than a single theory. It sets which questions count, which instruments matter, and what a satisfying solution looks like.

Newton's Principia worked this way for physics. It offered not only claims but a pattern others copied when they faced new problems. Once that pattern sticks, workers stop arguing foundations and start filling open details. Efficient accumulation becomes possible because the community agrees on what winning looks like.

The sting for working scientists is clear. Rigor, on this account, rests as much on a socially agreed starting point as on evidence alone. Without that agreement there is no shared scoreboard for progress.

In practice: Competing pre-Newton optics schools could not settle what light was, so there was no single science of light to advance until a shared template took hold.

2

Normal science as puzzles

Most scientific work is not bold theory-testing but solving small puzzles the reigning framework already guarantees have solutions.

Once a paradigm is set, Kuhn says daily science looks more like a crossword or jigsaw than dramatic discovery. The worker trusts a solution exists and that accepted theory and instruments can reach it. Ingenuity goes into fitting pieces, not into asking whether the picture on the box is wrong.

Typical tasks include measuring a constant more tightly, extending a theory's math to a new case, or pulling observations closer to predictions. None of that challenges the framework; it completes it. Specialization and speed follow because first principles are not reopened with every run.

There is a built-in cost. Puzzles the rules cannot touch get set aside, not because they are trivial, but because the paradigm offers no approach. Training teaches which questions are legitimate and which are not worth asking. That narrowing is a feature of paradigms, not a personal failing.

So what looks like stubbornness toward an odd result can be rational inside the game. At first an anomaly is only an unsolved puzzle, not proof the theory is false.

In practice: A researcher spends months refining a physical constant rather than reopening whether the underlying theory of measurement is sound.

3

When anomalies become crisis

Revolutions begin not from one odd result but when repeated puzzle failures accumulate until the community doubts its own foundations.

Normal science produces its own undoing. As workers push a paradigm into new territory, some results refuse the expected solution. At first these are treated as challenges to technique, not as refutations.

Strange findings can sit for a long time. Kuhn notes X-rays and anomalous planetary motions appeared well before anyone read them as paradigm threats. What raises the temperature is persistence plus centrality. If serious repeated tries fail, and if the trouble touches a problem the paradigm treats as fundamental, unease spreads.

His central case is the run-up to quantum mechanics. Classical physics could not absorb blackbody radiation or the photoelectric effect no matter how the puzzle was reframed inside old assumptions. Research character then changes. People question once-taken-for-granted assumptions, float ad hoc fixes, and sometimes turn philosophical about foundations—work normal science usually discourages.

A crisis still does not mean instant abandonment. Ready alternatives are rare. The field instead tolerates proliferating variations and half-formed ideas. The threshold is social and psychological as much as logical; there is no fixed anomaly count that forces a break.

In practice: Classical physics kept reframing blackbody radiation and the photoelectric effect as solvable puzzles until communal confidence in the old foundations cracked.

4

Replacement, not mere rejection

A paradigm is never dropped just because it fails; it is replaced only when a viable alternative already exists.

Kuhn is blunt: anomalies alone do not kill a paradigm. A flawed framework that still solves most problems beats having no framework, because normal science cannot run without one.

What moves people is a rival that handles the stubborn cases while keeping most prior successes. In the unsettled interval, scientists test several provisional pictures at once and accept logical looseness they would reject in ordinary times.

Before oxygen theory won, multiple patched versions of phlogiston chemistry circulated, each aimed at a different anomaly. In astronomy, Copernicus did not at first out-predict Ptolemy on planetary positions; appeal came from conceptual simplicity and the sense of addressing the crisis directly rather than patching around it.

This move severs paradigm change from simple empirical scorekeeping. A new framework can gain adherents while still weaker on some measures, because elegance, future problem-solving promise, and grip on the core anomaly also count. Data cannot settle the choice alone when data are partly read through the paradigm under review.

In practice: Copernicus won early support through simplicity and promise even while planetary predictions were not yet better than the geocentric model's.

5

Different worlds after a shift

A new paradigm changes not only beliefs but perception, so workers on either side of a revolution operate in importantly different worlds.

Kuhn's sharpest claim is that a shift is not a theory swap with fixed facts underneath. It alters what counts as a fact, which instruments are trusted, and which questions matter.

He borrows Gestalt language: the same marks can switch aspect, like a duck-rabbit figure. Laboratory results get seen anew. Priestley and Lavoisier examined what we now call oxygen. Priestley saw dephlogisticated air inside the old chemistry; Lavoisier saw a new element inside a new one. Neither made a simple factual blunder; each saw through a different frame.

Terms drift across the divide. Mass in Newtonian mechanics and mass in relativity connect formally yet assume different things about space, time, and motion. Kuhn later labeled the resulting mutual unintelligibility incommensurability. Pre- and post-revolution scientists can talk past each other.

He does not say anything goes. A new paradigm must still solve the old successful puzzles and face the central anomalies. But comparing paradigms is not like grading two answers to one well-defined question. If he is right, there is no fully neutral observation language that can referee the dispute by raw paradigm-free data.

In practice: Priestley and Lavoisier stood over the same gas; one reported dephlogisticated air, the other a newly identified element.

6

Conversion, not proof alone

Scientists switch paradigms the way people convert, moved by evidence mixed with aesthetic judgment and social pressure rather than by decisive proof.

If no single neutral scale ranks paradigms, how does a community settle? Kuhn's answer is that the switch resembles conversion more than deduction. Individuals yield to different mixes: superior handling of one anomaly, mathematical elegance, an early adopter's prestige, or none of these.

He notes, following Planck's observation, that a new framework sometimes prevails less by persuading every opponent than by the older generation retiring while the next is trained inside the new rules from the start. Hold-outs can keep doing productive work in the old frame for decades after the majority has moved.

Kuhn denies this is sheer irrationality. Predictive accuracy, simplicity, consistency with neighboring theories, and fruitfulness for future work all weigh. They simply do not combine into one formula that always names a winner, and reasonable people weight them differently.

After settlement, textbooks rewrite the story so the new paradigm looks like the obvious cumulative outcome. The real uncertainty and rivalry of the transition disappear from view.

In practice: A generation trained under a new framework accepts it as natural while senior hold-outs continue publishing inside the old one until retirement.

7

Textbooks hide the breaks

Science teaching systematically erases past revolutions, presenting each current framework as the smooth correct endpoint of steady progress.

Kuhn turns to pedagogy. A physics text does not stage Newtonian mechanics, relativity, and quantum theory as successive contested paradigms. It stages the present framework as the logical culmination, with earlier work cast as approximation or stepping stone.

He likens this to political victors rewriting the rise of the new regime as inevitable and the old as an obvious error corrected. Two effects follow. Scientists absorb a self-image of their field's past as smooth truth-accumulation, which hardens resistance to the thought that their own paradigm might later be overturned rather than merely polished. Philosophers who trusted that textbook story then built confirmation and falsification theories on a false continuity.

Kuhn's method was to return to notebooks, letters, and early publications instead of retrospective textbook narratives. That is how the discontinuities become visible again.

The chapter answers a puzzle raised by the rest of the book. If shifts are as disruptive as claimed, why do most practitioners not see them in their own field's past? The discontinuity is real; education trains it out of sight.

In practice: A standard physics text lines up earlier theories as stepping stones to the present instead of as once-dominant frameworks toppled in fight.

8

Progress without a final truth

Scientific progress is movement away from earlier, less adequate problem-solving, not movement toward one fixed true picture of nature.

Kuhn ends on the threat his argument seems to pose: if paradigm choice is not neutral comparison to independent truth, is progress real? His answer is a qualified yes, with the target redefined.

He compares scientific change to biological evolution. Species do not climb toward a predetermined ideal organism; they adapt away from earlier forms under pressure. Likewise a later paradigm is not necessarily closer to a final nature-as-it-is. It is reliably better at the puzzles the community now cares about, including the anomalies that broke the last framework.

Later paradigms usually retain and explain earlier successful predictions, so there is cumulative content. There is not cumulative certainty about ultimate reality. Kuhn rejects full relativism: puzzle-solving power is a genuine, checkable achievement, so not every framework is equal. He also rejects the portrait model in which each stage simply draws nature more accurately.

Readers are asked to drop the equation of better science with more true science, and to accept better as more capable, in its historical moment, of solving the problems that moment has generated.

In practice: A successor framework keeps earlier accurate predictions while solving the anomalies that stalled its predecessor, yet still is not claimed as the final portrait of nature.