Kuhn's book was disruptive because it described science historically rather than logically. Philosophers had been asking what method scientists ought to follow. Kuhn asked what they actually did, and found a pattern that fitted neither the steady-accumulation story nor Popper's picture of constant attempted refutation.
Most of the time, he argued, scientists do normal science: detailed work inside an accepted paradigm — a framework comprising theory, methods, standards, and above all exemplary solved problems that show newcomers what good work looks like. Normal science is not a criticism. Kuhn's point is that a shared framework is what allows genuine precision, because researchers can stop re-arguing foundations and get on with hard specific problems.
Normal science generates anomalies — results the paradigm cannot readily accommodate. Crucially, and contra Popper, a single anomaly does not overturn anything. It is filed as an unsolved puzzle, and sensibly so, since young theories are always surrounded by unexplained data. Only when anomalies accumulate, resist repeated effort, and strike at central commitments does crisis set in. Then alternatives proliferate, foundational questions reopen, and eventually a new framework reorganises the field.
The claim that caused most argument is incommensurability. Rival paradigms, Kuhn suggested, may lack a fully neutral standard for comparison, because standards of what counts as a good explanation or a relevant problem are themselves part of the paradigm. Terms can shift meaning across the divide — mass in Newtonian and relativistic physics is not quite the same concept — so the theories are not straightforwardly rivals answering an identical question.
Critics read this as making theory change irrational, a matter of persuasion or sociology rather than evidence, and some enthusiasts adopted precisely that reading. Kuhn spent much of his later career rejecting it. His position was that scientists have good reasons — accuracy, scope, simplicity, fruitfulness — which genuinely bear on the choice, while denying that these reasons form an algorithm that compels a unique answer. That is a claim about the nature of judgement, not a denial of rationality, and it describes how most difficult expert decisions actually work.
Two cautions are worth stating, because the term has escaped into general use. First, "paradigm shift" now gets applied to any moderately novel idea, which drains it of content — Kuhn meant a change in the entire framework of a mature science, and those are rare. Second, none of this implies that scientific knowledge is arbitrary or that any framework is as good as another. Kuhn held that science makes real progress in problem-solving power; what he denied was that it progresses by converging on a single fixed picture of reality.