Kuhn, Thomas S. The Structure of Scientific Revolutions. Third edition. Chicago: The University of Chicago Press, 1962.
 
In his groundbreaking Structure of Scientific Revolutions, T. S. Kuhn argues that scientific knowledge is not accumulated incrementally; rather, there are defined paradigms of scientific knowledge – "constellations of facts, theories and methods" – and scientists make a revolutionary move of faith when they transfer their allegiance from one paradigm to another. (1) Often, but not always, a new paradigm resolves the anomalous observations that the old paradigm could not explain, and usually the new paradigm, while initially unsophisticated, suggests new directions and new boundaries for exploration. The Structure of Scientific Revolutions also argues that scientific thinking is analogous to puzzle-solving: to be scientific, a paradigm must predict that all of the data, all of the "pieces," have a place in the puzzle, that there is one right relation between the pieces, and that there is a solution to the whole puzzle. A scientist doing "normal" work engages with the puzzles of his or her paradigm.
 
Kuhn structures his book (he calls it an essay) in 13 chapters. Chapters 1-5 describe normal science; Chapter 6-8 discuss the "element of arbitrariness" present in scientific endeavor; and Chapters 9 and 10 the nature of scientific revolutions. Chapters 11-13 are grouped together to ask three questions: why scientific revolutions are so difficult to see; what the process of evaluating scientific theory ought to be; and how development through revolution can be scientific.
 
Normal science, Kuhn says, is "research firmly based upon one or more past scientific achievements, achievements that some particular scientific community acknowledges for a time as supplying the foundation for its further practice." (10) Normal science is governed by a paradigm, an organizing principle sufficiently engaging to attract an "enduring group of adherents" and sufficiently broad to offer a range of problems for the adherents to solve. Researchers pursue three kinds of facts: those that support the framework of the paradigm (boiling points, acidity, specific gravity); those that confirm the paradigm itself; and those that resolve potential anomalies in the previous and present paradigm. The effort of most professional scientific work is not to challenge the paradigm but rather to confirm, via "puzzle-solving," the relations of the pieces (data) that the paradigm predicts should fit together. By grouping "shared rules and assumptions" (49) paradigms permit scientific community subgroups to share some but not all of the same ideas, thus opening the interstices through which revolutionary anomalies can affect one or more of these subgroups.
What happens when scientific paradigms are confronted by anomalous data? Scientists may simply add to accretion of adjustments needed to make the paradigm "work," such as those required by the Ptolomaic universe. Often, anomalous data are not recognized as such at first; it is more likely for the researcher to believe that he or she has made a mistake in performing the experiment. In this regard Kuhn cites an interesting experiment done with "impossible" cards (red spades, black hearts) that were persistently misidentified as "normal" cards by test subjects. Research done under the aegis of the paradigm generally indicates only what the paradigm predicts, but newly developed equipment (as Kuhn describes the Leyden jar) allows for new explanations to be considered. Looking back, it may be difficult to pinpoint the date of a new discovery, since the name for the new "thing," such as oxygen rather than phlogiston, may depend upon a new paradigm being created. Paradigms are sometimes ruptured when they seem too complex to be "real" – persistent errors and clumsy "patches" are required. (Simplicity being a hallmark of reality) It may also be that a previously undiscovered effect, such as the X-ray, is discovered. "As in manufacture so in science – retooling is an extravagance to be reserved for the occasion that demands it." (76)
 
Scientific revolutions, then "are those non-cumulative developmental episodes in which an older paradigm is replaced in whole or in part by an incompatible one." (92) Frequently the new paradigm raises as many problems as it solves, and it may not do a better job of accounting for questions that have already been "solved" in the old paradigm. Kuhn is not a positivist: he argues that the acceptance of a new paradigm does not necessarily mean that a more "realistic" version of the world has been arrived at. Rather, he argues that influential people in the field accept the new explanations, as well as the attendant puzzles and problems, sometimes because the solution seems better, and sometimes for other reasons. For whatever reason, accepting a new paradigm means that the same object will be seen very differently. Kuhn returns often to the example of a weight on a string – after Newtonian physics, it is no longer a weight being constrained, but a pendulum. Again, Kuhn cites an experiment in which the subjects were given goggles with inverting lenses that made the world appear upside down. After a few days of confusion and disorientation, the subjects internally "reversed" what they saw, so the world re-appeared right side up.
 
Why are scientific revolutions silent? Kuhn suggests that scientific revolutions are obscured by those textbooks that give students a seamless and second-hand version of inventions and events. A-historical revisioning is a necessary concomitant to creating paradigms, but such revision denies the catastrophic nature of discovery. Neither falsification nor confirmation is the method by which truth is decided upon, because members of two paradigms who are in conversation with each other cannot even discuss questions using the same verification methods. Competition is the method by which truth is determined, and scientific communities must be nurtured in order to ensure the continuing growth of precise and detailed data. (170)
 
In his 1969 Postscript Kuhn addresses questions and problems raised by his work. He defends and extends the definition of paradigm, while acknowledging the need to have defined scientific communities as prior to paradigms, and calling for further research on professional communities of all sorts. His commentary on problems as exemplars is suggestive of linguistic theory of language learning. In fact, Kuhn posits that those working from different paradigms must translate to be understood by one another. Kuhn's work functions like one of his own paradigms
-- after reading it, it is difficult to see scientific progress in any other way. However, one question does seem worth asking: Kuhn's system seems not to permit us to look at the inventions and discoveries that make one paradigm qualitatively better than another. While he might argue that paradigm is not necessarily better at solving puzzles than another, would he argue that one might produce better results than another? Kuhn says unequivocally that he is a convinced believer in scientific progress in that "later scientific theories are better than earlier ones for solving puzzles in the often quite different environments to which they are applied" (206). How would one evaluate paradigms if one posits that some problems obtain in all the environments in which humans find themselves?