Vansh Kumar | The Structure Of Scientific Revolutions

Exquisite book. A very different view of how the process of science occurs than the textbook-style edited narrative you learn in school. Almost guaranteed to make you rethink what you know and see science for the sociological process it is.

Believing is seeing is believing

Before Kuhn wrote this book, many people (including scientists) saw science as an objective falsifiable process where “seeing is believing” rang true. A scientist has a hypothesis, does an experiment, and objectively compares results against their hypothesis & prevailing theories. If it doesn’t match, the theory is updated based on the results of objective measurement. Kuhn points out that this is not how things work at all.

The hypothesis a scientist chooses to test in the first place, how they set up their experiment, whether they interpret their results as theory-falsifying or an experimental error. All of these things are heavily influenced by what they already believe – what Kuhn calls the current scientific paradigm.

His notion of normal science is revealing. There is an existing paradigm, or way of looking at & organizing the world. Most science at any time is normal science and is a kind of puzzle-solving. A scientist doing normal science may set up an experiment with a hypothesis, but crucially they do not expect to discover anything surprising. Rather, the experiment is typically done with outcome already in mind – extending an existing theory to an incrementally new area, measuring some constant to a higher level of precision, etc. If something anomalous is discovered, it’s often explained away as experimental error or via duct-taped additions to the existing theory.

One of the classic examples is the precession of Mercury. Mercury’s precession was observed to deviate from what Newtonian physics would predict in 1859. Instead of this anomaly immediately leading to a major upheaval of Newtonian physics, various ad-hoc solutions were proposed, including the suggestion of a hypothetical planet named Vulcan. It took 56 years for this to be resolved by Einstein’s theory of general relativity.

Doing normal science in this way may seem to go against some rigorous notion of “the scientific method”, but it is in fact very good sense. How else can you reasonably deal with a world where a nearly infinite number of theories can explain any data, on top of any data being subject to question based on the measurement technologies of the time? Persisting with a particular paradigm despite anomalies, until the anomalies have both piled up and there’s a replacement theory, is logical. It’s more likely than not that an observed anomaly is actually error. There were indeed challenges to Newtonian physics that ended up just being math errors.

Revolution: changing what you believe changes what you see

As the name normal science implies, the above is what science typically looks like. Incremental improvements to existing theory in various ways. So how do changes in paradigm happen?

A paradigm can find itself in a state of “crisis” when people know that anomalies have piled up and a theory is full of duct-taped fixes or things it simply can’t explain:

the astronomical tradition he [Coperniucs] inherited had finally created only a monster. By the early sixteenth century an increasing number of Europe’s best astronomers were recognizing that the astronomical paradigm was failing in application to its own traditional problems.

The core element of a crisis is the breakdown of the technical puzzle-solving of normal science. There are many other relevant factors (eg social pressure for calendar reform in the case of Copernicus), but that is the primary one.

Oddly, a replacement theory doesn’t seem to have to be more accurate, explain things better, or even be simpler. At the time Copernicus’ heliocentric theory was not more accurate or much simpler than the Ptolemaic model. He did not do away with the epicycles that littered the Ptolemaic model. All that’s needed is recognition that the existing paradigm isn’t working and it may be time to build something new.

Even Copernicus’ more elaborate proposal was neither simpler nor more accurate than Ptolemy’s system. Available observational tests, as we shall see more clearly below, provided no basis for a choice between them. Under those circumstances, one of the factors that led astronomers to Copernicus (and one that could not have led them to Aristarchus) was the recognized crisis that had been responsible for innovation in the first place. Ptolemaic astronomy had failed to solve its problems; the time had come to give a competitor a chance.

One needs there to be a crisis to motivate looking for a replacement. It’s not worth it to overhaul the functional, if creaky, machinery of normal science unless there’s a good reason. And it is a complete overhaul. A new theory reorganizes and gives new meaning to many of the experiments, data, and results that precede it. Changing what you believe changes what you see.

And this may be part of why it takes time for a new paradigm to be accepted after it’s been proposed. A complete reorganization of your worldview is difficult, especially when you can always hold out hope that some new experiment will show the new theory is bunk. “Science advances one funeral at a time” and all that.

Takeaways

For me the biggest rethink due to this book was that science is a sociological process. Who decides what is a worthwhile question to ask & which methods are valid to answer it? I had not even thought about this before. The cliché that science is really about asking the right questions speaks to this point. I noticed this myself when trying to do open-ended research – it took me around a month to get to a point where I could even ask a useful question.

While I am not sure Kuhn’s process of normal science punctuated with revolutions applies to all sciences (eg I don’t think biology conforms well), I do think his description of normal science is spot on. It highlights the tacit knowledge that shapes which questions are asked, which experiments are run, and how the results are interpreted.

Aside: wtf is a paradigm?

The word “paradigm” is used throughout the book and Kuhn doesn’t really give a definition. I certainly assumed it was nearly interchangeable with “theory,” or perhaps more broadly “set of shared beliefs about how the world works.” And I still kind of use it that way.

The postscript, written ~10 years later, has Kuhn note that his loose use of this word was one of the biggest objections to the ideas in the book. He then provides a clarification: paradigm is something akin to “shared exemplar.” It’s a solution to an important recognized problem that acts as a beacon for. He illustrates this with the example of $f = ma$:

That expression proves on examination to be a law-sketch or a law-schema. As the student or the practicing scientist moves from one problem situation to the next, the symbolic generalization to which such manipulations apply changes. For the case of free fall, $f = ma$ becomes $mg = m \frac{d^2s}{dt^2}$; for the simple pendulum it is transformed to $mg sin(\theta) = – ml \frac{d^2 \theta}{dt^2}$

The student discovers, with or without the assistance of his instructor, a way to see his problem as like a problem he has already encountered. Having seen the resemblance, grasped the analogy between two or more distinct problems, he can interrelate symbols and attach them to nature in the ways that have proved effective before. The law-sketch, say $f = ma$, has functioned as a tool, informing the student what similarities to look for, signaling the gestalt in which the situation is to be seen.

A paradigm is a way of seeing problems. Kuhn suggests it is transmitted primarily through the example problems students work out in school, often at the end of textbook chapters. This certainly matches my experience.

Aside 2: progress

Why does science seem to progress? Kuhn’s potential answer: science may be defined as that which progresses, mostly from normal science.

Social science seems like “less of a science” precisely because it doesn’t seem to progress. People are often still arguing about what constitutes a valid solution to a problem in the field. Shared beliefs & exemplar solutions accepted by the whole field are much rarer. This may be because the problems social science works on are based on what matters to society, whereas fields like physics are free to work on whatever problems seem tractable with current tools. It’s a matter of problem selection.

This view also helps explain why a social science like economics appears to have made a lot of progress in its decades-long turn towards mathematical modeling, but it’s unclear if it actually has gotten better at solving real-world problems.

Provenance