The premise of the book is that science doesn't progress by the cumulative addition of knowledge, but instead advances by major shifts in paradigms that replace, rather than increment, large parts of previous paradigms. To begin with, scientific research in a specific subject is carried out within the bounds of a generally accepted framework that defines what scientists already know about the field, as well as the questions that remain unanswered. This is what Kuhn calls a paradigm. A paradigm is useful because it defines puzzles that need to be solved and gives a set rules for them to be solved in. Over time, the paradigm is more fully explored and is broken down into smaller and more specific problems. To solve them, scientists develop specialized equipment and detailed experiments are carried out. Scientists experiment not to generate an unknown result, but with a hypothesis that has an expected result. The paradigm they're working under has helped them predict results and expect an answer.Eventually however, problems are discovered that can't be effectively solved within the rules of the paradigm. At first, these types of problems can be worked around by making adjustments in rules of the paradigm. Ultimately, as it is explored more deeply and the rules become more complex, a problem or problems arise that simply cannot be answered elegantly with the paradigm. As these difficult problems gain notice, they become recognized as the problems in most urgent need of a solution. When there is a big problem like this it can either be ignored until better equipment is available, made to fit by adjusting the current paradigm or, most interestingly, it can lead to the development of a new theory, or group of theories that attempt to solve the problem. As alternate competing theories are proposed to address the crises, eventually one gains enough traction among scientists to become the new paradigm. One of the examples used in the book is how Einstein's relativity became the paradigm that replaced Newtonian physics. What happened was not that Newtonian physics was found to be outdated and immediately replaced with the theory of relativity, in fact that theory is still useful within a large number of applications today. Instead, it was recognized that there is a very limited set of parameters in which Newtonian physics is accurate--specifically for calculating interactions between objects moving at relatively low velocities, but that outside those parameters, Newtonian physics will lead to incorrect assumptions. The theory of relativity solves the same problems that Newtonian physics does, but it also works with objects moving at high velocities. Rather than just building upon Newtonian physics incrementally, relativity supplanted large parts of it, even as scientists recognized that parts of Newtonian physics remain useful in certain contexts.To me, a non-scientist, rather than being controversial, this is a really useful way to think about science, and beyond science to how change and progress occur in almost any field. To a scientist, I can see how Kuhn's ideas are controversial. They mean that what scientists see and look for in observational and experimental data is not analyzed and recorded completely objectively but that scientists are heavily biased by what they believe and expect they're going to find. It 'accuses' scientists of viewing data and the experiments they choose to perform relatively, rather than objectively or positively (there is a long and hairy philosophical argument on relative knowledge that I will avoid getting into). Kuhn, rather than criticizing scientists for their subjective view on data, believes that viewing science this way is unavoidable, and in fact beneficial because it trains scientists to recognize patterns in data and to become adept with the data that they deal with. When a scientist is proficient at viewing data within the bounds of a paradigm, they are, in turn, well-prepared to view anomalies in the way their paradigm interprets data. This leads to the tough problems that are escalated to criseses in the paradigm and eventually, to the development of new paradigms.