The Structure of Scientific Revolutions

What follows is a (truncated) summary from my reading of The Structure of Scientific Revolutions by Thomas Kuhn, as well as some personal reflections. You can read the full original text, complete with postscript, here.


The key question

How does science progress over time?

Science is not (merely) cumulative

A common portrayal of scientific progress is that of a cumulative process, where new ideas build upon old ones in a clear trajectory towards the truth.

Kuhn argues that this is an inaccurate view of the trajectory of science. Instead, science consists of two interleaving phases: normal science and revolutionary science.

Normal science

Normal science resembles the then-conventional view of science as a cumulative process.

Kuhn describes the activities of normal science as "puzzle-solving". Puzzles are well-defined problems with reasonably straightforward solutions. Normal science results in an accumulation of puzzle-solutions.

The aim of normal science is not to discover novel things, but to:

  1. Gather facts: constants, laws
  2. Match facts with theories
  3. Articulate theories

Normal science is guided by past achievements which have become foundational to a discipline (paradigms).


The word "paradigm", now rooted in common parlance (especially in corporate-speak, yuck), was notoriously amorphous in the main body of Structure. In the postscript, Kuhn later clarifies two distinct usages of "paradigm":

  1. Sociological: beliefs, values, techniques
  2. Exemplars, or model problems

This second definition of the paradigm as a set of model problems and solutions is easier to pin down. Some examples Kuhn provides include seminal texts containing foundational theories and their applications to important problems, such as:

Paradigms are necessary for a mature science. They specify the boundaries of a scientific discipline by:

  1. Suggesting new puzzles
  2. Providing potential approaches to solve those puzzles
  3. Providing a standard to evaluate proposed solutions

Note that Kuhn also uses the term "disciplinary matrix" interchangeably with "paradigm".


Anomalies are puzzles that resist solving by a paradigm.

According to Karl Popper, any given anomaly is enough reason for a scientific revolution. Kuhn argues that it usually takes more than one anomaly for this to happen, and there is no scientific basis for which particular anomalies lead to revolutions.

When enough anomalies accumulate, more and more scientists lose confidence in the current paradigm and a crisis occurs. There are three potential responses to a crisis:

  1. Normal science is able to resolve it
  2. The anomalies are left as problems for future generations to solve with better tools
  3. Scientific revolution (paradigm shift)


When a scientific revolution occurs, the old paradigm is replaced partly or entirely by the new one. Kuhn likens this process to a political uprising, where members of a community lose confidence in the existing institutions. Any number of new candidate paradigms come into play to be chosen as the new paradigm.

In addition to purely scientific factors, sociopolitical factors such as the nationality and personality of scientists may influence which paradigm prevails in a revolution. Kuhn gave the example of Kepler's "sun worship" (I'm not sure how literal his worship was) as being pivotal to his adoption of Copernicus' heliocentrism.


Lots of syllables that mean "lacking a common standard of measurement"—or more colloquailly, "apples and oranges". Two differing paradigms cannot be easily compared because paradigms are incommensurable with one another. There are three reasons for incommensurability:

1. Methodological differences

Each paradigm has its own accepted methods and criteria for evaluating a given theory.

2. Perceptual differences

The prevailing positivist view was that observation is a way to independently evaluate competing theories, but Kuhn argues that observation is affected by the paradigm a scientist subscribes to. Scientists practicing in different paradigms perceive the world differently.

3. Semantic differences

The same term may have different meanings. See, for example, Newtonian mass and Einsteinian mass.

So what's the big deal?

Paradigms as exemplars help explain how scientists come up with new hypotheses and solutions. Kuhn suggests that these exemplars help scientists frame new problems in terms of established model puzzles, which makes it easier to come up with puzzle-solutions. The acceptability of new solutions is based not on rational rules, but on their similarity to exemplar solutions in the reigning paradigm.

The idea of paradigm shifts may not seem all that, er, revolutionary, but Kuhn was the first to formulate a theory of scientific change that deviated from the conventional heroic, cumulative view. Kuhnian revolutions are also discontinuous: for example, a new paradigm is often unable to solve all the problems addressed by its predecessor (this phenomenon is known as "Kuhn loss").

Another far more controversial idea is that through revolutions, scientific communities develop theories with better puzzle-solving abilities (simplicity, precision, etc.), but these theories do not necessarily draw closer to the truth. The reason for this is the aforementioned incommensurability: if two paradigms cannot be meaningfully compared, the reason scientists pick one over the other is based not on approximation to the truth, but on changes in world view. Despite likening scientific change to revolutions throughout much of the book, Kuhn drops an analogy to evolution in the final chapter. While science may evolve from historical progenitors into diverse, specialised forms, it may not converge towards a single idealised goal.

Many of the ideas in this book were interepreted (not always fairly) as attacks on the rational core of science. Kuhn was accused of being a relativist, which is the position that objective reality does not exist. Kuhn also cast doubt on the picture of scientists as neutral, rational agents. In a period of crisis, proponents of competing paradigms argue their cases using rhetoric rather than relying only on cold, hard evidence. Revolutions are by and large invisible to scientists because the writers of textbooks provide a conventional account of progress at the cost of diminishing the role of past, outdated achievements.

Personal reflections

Structure kicked off a slew of debates that remain far from resolved. Kuhn's arguments and the immediate responses to them should therefore not be taken at face value. Indeed, Kuhn's own views on the matter changed significantly as he revisited these topics through the course of his life. Make sure you read the postscript as well as the responses by Kuhn's contemporaries to avoid missing important context. Even then, they form the mere tip of the iceberg. See the SEP entry on Kuhn for a nice overview of a pretty darn deep rabbit hole.

As someone who is keen on a career in biomedical research, I'm glad I finally read this book in its full, raw glory after having come across its ideas in many editorials and articles. Hot takes about truth and rationality aside, the sheer breadth and scope of ambition of Structure is infectious (there's more than a brief nod to Wittgenstein, and some appeals to gestalt psychology). However, you should know that this is firmly big picture stuff, dealing with philosophical questions about the nature of science itself. I honestly don't see how this affects the working scientist in any meaningful way, other than perhaps forcing you to think hard about whether your chosen field is a legitimate science. Don't get me wrong, this is thought-provoking and mind-expanding—but don't expect it to change the way you think or work as a scientist.

Day 33 of #100DaysToOffload

← Home