Kuhn’s Paradigms, and Foucault’s Tower

This is a textbook chapter, and its subject is a field. Sismondo is introducing science and technology studies, and chapter 2 is the origin story of that field: the moment philosophy stopped asking what makes a statement scientific and started asking what a community of scientists does. So the chapter spends its first half on Kuhn and its second half on what people built afterwards, and the second half is the part a reader expecting a summary of Kuhn will miss.

His one-line version of Kuhn's move is that Kuhn rejected the formalist view with its rules about what science ought to be and looked at the activities instead, so that science is merely what scientists do. Everything else in the chapter follows from taking that seriously.

Whig history, a phrase Sismondo credits to the historian Herbert Butterfield in 1931, is history written as a series of steps toward the present. In the history of science the temptation is to treat every move toward what we now believe as more rational, more natural, and less in need of explanation than the opposition to it. Copernicus gets no explanation because he turned out to be right; his opponents get explained away as stupid or frightened.

Sismondo's objection is about causes, and it is short. Events must follow their causes, so a sequence of events cannot be explained by where it ended up. Explaining 1543 by what happened in 1687 is explaining a thing by something that had not happened yet.

Why it matters for the rest: the story has been told as though history were aiming at its own ending all along, which is what teleology means. Take that out, and there is room for changes that are not improvements. That room is where the whole account of revolutions goes.

Normal science is the science done when a field shares four things: recognition of key past achievements, beliefs about which theories are right, an understanding of which problems matter, and methods for solving them. A group that shares those four shares a paradigm.

The word came from grammar, where a paradigm is the table of endings you learn and then copy, and Sismondo keeps that sense deliberately: it points at an achievement that serves as an example for others to follow. Newton's mechanics, Lavoisier's chemistry and Mendel's genetics each did that, structuring the research that came after them.

A paradigm has two sides, and the second one is the one that gets forgotten. The theoretical side is a worldview: categories and frameworks to slot phenomena into. The practical side is a form of life, the phrase borrowed from Wittgenstein: patterns of behaviour, ways of working, what you do with your hands all day.

Normal science is puzzle solving, and the word is doing work. A problem is to be solved within the terms of the paradigm, so failing to solve one is a mark against the researcher and never against the theory or the method. An unsolved problem is not evidence against anything. It is an anomaly, left for somebody later.

Sismondo's explanation for why the paradigm is not questioned is training, and he is blunt about it. The natural sciences are exceptionally good at socialising their practitioners: textbooks that present a standardised view of the field and its history, long apprenticeships, and years of being set problems that are well understood and already have known answers. A student is not taught to doubt the framework, because every exercise they are given presupposes it.

Anomalies accumulate, because a paradigm can only ever be a partial way of dealing with its subject. Some of them stop being puzzles and start being real problems, which produces discomfort with the terms of the framework itself. That is crisis.

Then the generational part. If an alternative appears that solves some of the central unsolved problems, the scientists who take it up first are the younger ones, the ones not yet fully indoctrinated into the way of life of the older paradigm. As the older and more conservative scientists become marginalised, the alternative becomes a paradigm itself and a new period of normal science begins.

Progress exists inside normal science and not across revolutions. Within a paradigm everyone can recognise everyone else's achievements, so progress is easy to talk about. A revolution both builds and destroys: some of the research done under the old paradigm will no longer make sense under the new one.

Read that carefully, because it is a report and not an endorsement. Sismondo is saying what follows if Kuhn is right, and the next four pages are him arguing that this part of Kuhn is not right.

The claim is that there is no shared ruler. Two paradigms cannot be laid side by side and measured against each other, because there is no scale both of them accept, so asking which is better has no way of being settled. The word for that is incommensurable, which literally means having no common measure. Sismondo gives Kuhn three separate reasons for the claim, and keeping them apart is what lets him knock them down one at a time.

1. Meaning. From the positivist picture, where the meaning of a theoretical term comes from the observations it implies. Kuhn extends it: a term's meaning depends on the whole constellation of claims it sits in. Change the paradigm and every key term shifts, so no term in one maps neatly onto a term in the other.
2. The theory-dependence of observation. Borrowed from N. R. Hanson in 1958. There is no raw observation in ordinary circumstances: we do not see dots and lines, we see recognisable objects, and what we recognise is guided by the concepts we already have. So during a revolution people stop seeing one way and start seeing another.
3. The historian's experience. Kuhn's own, from doing the work: past science is genuinely opaque, and parts of it look bizarre. Sismondo adds the sharp version of this. If somebody finds very old research easy to understand in present-day terms, they are probably doing violence to it. Newton's physics looks strikingly modern in a modern textbook, much less so in its original printing, and less so again once his religious and alchemical work is put beside it.

This is the pivot of the chapter, and a reader skimming for a summary of Kuhn will read straight past it. Having set out the three roots, Sismondo argues that the strong claim fails on all of them.

Meaning. The case for semantic incommensurability has had a great deal of attention and almost all of it negative. Meanings do change. They do not change so much, or so systematically, that claims using them cannot be compared. Most philosophers and linguists who have worked on it have concluded the claim cannot be sustained, and Donald Davidson argued in 1974 that a change of meaning that radical cannot even be made sense of.

History. That problems, concepts and methods change is uncontroversial. The difficulty of interpreting past science can be overcome, and the proof is that historical research keeps overturning present-centred readings of the past. If the past were really sealed off, that could not happen.

Kuhn moved first. By 1965 he had already retreated to saying that by incommensurability he meant only incomplete communication or difficulty of translation, sometimes ending in communication breakdown. On that reading it is still a real phenomenon, and a useful one: people working on the same subject can and do fail to understand each other.

And if there is no radical incommensurability there is no radical break between paradigms either. Sismondo turns this into a methodological principle. Complete ruptures are inexplicable. When a historian explains an innovation they explain it as a reworking of resources that were already available, because every new idea, practice and object has sources, and to assume otherwise is to invoke magic.

That is the chapter's own best line and it is worth holding. Science hangs together not through one continuous thing running through it but through overlap. It gets what Sismondo calls an ad hoc unity from the fact that its parts rarely change together.

These are the two named concepts most likely to be asked for, and neither of them is Kuhn's. They are what the field built once it accepted that different specialties really do work differently.

Start with the ordinary case. A molecular biologist's work is not obviously interesting or even comprehensible to an evolutionary ecologist, and with some translation it sometimes becomes so. Disciplines are what Karin Knorr Cetina calls epistemic cultures, with different orientations to their objects, different social units, different ways of interacting. Yet they interact anyway, and science gets some unity out of it. The question is how.

A trading zone, the term is Galison's, is a place where practices from different fields interact through a simplified shared language, a pidgin, without either side having to assimilate. Traders exchange goods without adopting each other's culture. Collaborations in a trading zone can work even when the two sides do not agree about what the problems are or what the words mean.

Boundary objects, from Susan Leigh Star and James Griesemer in 1989, put the weight on things. An object works as a bridge if it can hold the attention of different social worlds at once and stay recognisably itself in each of them.

Sismondo is sceptical of both, and says so. The boundary object concept has been applied to an enormous range of things, including, within the original article alone, the museum itself, the animal species in it, the state of California and the specimen records. Engineers' sketches, the parameters of climate models, and breeds of rabbits and poultry connecting fanciers to geneticists have all been called boundary objects since.

His conclusion is the deflationary one. The sheer number and variety of these bridges suggests that coordination and communication across boundaries are common, and that a good deal of it happens by plain translation: researchers learn what colleagues in other fields know and put things in terms those colleagues can follow. Trading zones and boundary objects may therefore overstate the difficulty. Incommensurability, as it turns up in practice, is often not a serious barrier at all.

The conclusion lines Kuhn up against four positions, three of which are in the week 2 packet.

• Against functionalism. Scientific communities are organised around ideas and practices, from the bottom up. Functionalism had them organised around ideals of behaviour, serving an overarching goal.
• Against positivism. Changes in theory are driven by changes of vision. Data follows: if a worldview is a theory, then data sits under theory, which is the reverse of what positivism said.
• Against falsificationism. Anomalies are typically set aside. Only during a revolution are they used to justify rejecting a theory.
• Against all of them. On the largest scale the history of science is a story of change. Uninterrupted progress is the story it is usually told as.

On the irrationalist reading, Sismondo is dismissive. The book was read as saying science is fundamentally irrational, or is mob rule, because it violated almost everybody's ideas about rationality and progress. Looking back, he says, it is difficult to find much irrationalism in it, and possible to read the book as fairly conservative, Steve Fuller having argued it is politically conservative as well.

And the verdict, which is oddly generous and oddly brutal at once. Few of Kuhn's specific ideas have survived fully intact. The book profoundly affected everything that came after it anyway, because what it did was create a space for studying the practices of science locally, in their own terms, with no need to score them against progress or against an ideal.

Sismondo's boxes are set apart from the text and carry named material. All three are inside the assigned pages.

Thomas Kuhn was an American physicist who moved into the history of science, and in 1962 he published The Structure of Scientific Revolutions. It is short, it is the most cited book of its century in the humanities and social sciences, and it put the word paradigm into ordinary English, where it is now used by people selling software.

Where it came out is the joke. It was published as a monograph inside the International Encyclopedia of Unified Science, which was the Vienna Circle's own publishing project. Chapter 9 of last week's packet is the Vienna Circle manifesto, arguing that science is one unified structure built up from verified observation statements. Kuhn's book, which says science does not build up and observation statements are not neutral, came out as a volume of their encyclopedia. The people who commissioned it published the thing that undid it.

What the book was read as saying was that science is not rational. That a paradigm wins by fashion, politics or the deaths of the people who disagree, that one is as good as another, and that there is no such thing as getting closer to the truth. That reading is why "paradigm shift" now means any change of mind at all, and it is the reading Kuhn spent the rest of his life denying.

What he wrote is narrower. There are real reasons for preferring one theory to another, and he named five: accuracy, consistency, breadth, simplicity and fruitfulness. His claim is that those five do not settle it, because two people can weigh them differently and neither is being stupid. One prefers the theory that fits the measurements better, the other the theory that covers more ground, and there is no further rule that says which of them is right. So the choice is reasoned and underdetermined at the same time, which is a long way from irrational.

Progress he never denied. Later science solves more puzzles, more precisely, than earlier science. What he denied was the shape the textbooks give it.

The textbook shape is a wall. Each generation lays a course of bricks on the one below, nothing already laid is ever taken out, and the wall only gets higher. On that picture Newton is still standing inside Einstein, as the special case you get at low speeds.

Kuhn says bricks come out. Newtonian mass is a fixed quantity that a body simply has; relativistic mass changes with speed and converts into energy. They are not the same quantity with a correction term, so the sentence "mass is conserved" is not a true claim that Einstein kept and refined. It was thrown away and something else was put in its place that happens to give nearly the same numbers when things move slowly. Phlogiston went out of the wall entirely. Caloric, the weightless fluid heat was made of, went out. The crystal spheres carrying the planets went out. A wall that loses courses is not a wall being built up, and that is the whole of what "science does not accumulate" means.

Kuhn's first move is to describe what scientists actually do all day, which is not what the textbook picture says. The textbook picture is that a scientist proposes a bold conjecture and tries to refute it. Kuhn looked at the historical record and found that almost nobody does this almost ever.

What they do instead he calls normal science: solving puzzles inside a framework that is not itself in question. Measure this constant more precisely. Extend the theory to one more case. Build an instrument that gets at the quantity the theory says exists. The framework is the tool they work with.

Puzzle is his word and it is deliberate. A puzzle is something you already believe has a solution, and the interest is in whether you are clever enough to find it. A crossword has an answer. If you fail, the crossword is not discredited, you are. Normal science works the same way, and that is why a failed experiment is normally read as a failed experimenter.

Kuhn used paradigm in more senses than he meant to. A critic named Margaret Masterman went through the book and counted twenty-one distinct uses. Kuhn accepted the complaint and in a postscript split it in two, and those two are the ones to hold.

• The disciplinary matrix. Everything a scientific community shares: the equations it writes down, the models it pictures things with, the values it uses to judge a good theory, the list of problems it thinks are worth solving. Not a theory, a whole working culture.
• The exemplar. A concrete problem already solved, which students work through in training and then imitate for the rest of their careers. The inclined plane. The hydrogen atom. Kuhn said this was the sense he cared about most, and it is the one usually forgotten.

The exemplar sense is the interesting one, because it says a scientist is trained less by learning rules than by copying worked examples. Nobody can state the rules of their own field completely. What they have is a stock of cases and a feel for which new problem resembles which old one. That is a craft, and it is learned the way crafts are learned, which puts Kuhn unexpectedly close to the Greek technē in last week's Schadewaldt chapter.

Anomaly. A result the framework cannot digest. Most anomalies get absorbed, adjusted for, or put on a shelf marked "not yet", and there is a good reason for it: a framework that collapsed at the first awkward result would never get any work done.

Crisis. Anomalies accumulate at a point the community cares about, the fixes start looking like patches, and confidence goes. Kuhn's claim is that this is a state of the community. The same anomalies could have been known for decades and the crisis arrives when confidence does.

Revolution. A new framework takes over. Not by being added to the old one, but by replacing it, and Kuhn's strongest claim is that the new one does not contain the old as a special case in the way textbooks pretend.

Incommensurable means there is no common measure. Kuhn borrowed it from geometry, where the side and the diagonal of a square cannot both be counted in whole units of the same ruler. Applied to paradigms it means you cannot settle the contest by totalling the evidence, because three things move at once.

• The words change meaning. Mass in Newton is conserved and independent of speed. Mass in Einstein is not. The same word, two different quantities, so a sentence with "mass" in it is not the same claim in both.
• The list of good questions changes. Asking why a body keeps moving is a serious question before Newton and a confused one after him.
• The standards change. What counts as a satisfying explanation is set by the paradigm, so each side scores itself well by its own rules.

Kuhn's own picture for it is the duck and the rabbit. One set of lines, two ways of seeing, and no third way of seeing it that is neutral between them. You can flip between them and you cannot hold both at once, and nothing in the drawing itself settles which one is correct. A scientist who moves to a new paradigm does not get a better view of the same thing; the thing looks different.

What Kuhn did not say, and spent thirty years insisting he had not said, is that theory choice is irrational or that anything goes. His position is that there are good reasons, that they are real, and that they do not add up to a proof, so two careful people can weigh them differently. He hated being called a relativist.

Kuhn moves the unit of analysis from the method to the community. Before him, the question was what rule separates science from everything else, which is the Vienna Circle question. After him, the question is what a trained group of people accepts, why they accept it, and what would make them stop. That is a question about people, and people can be studied by sociologists.

Which is exactly where the course goes next. Week 4 is the social construction of scientific knowledge, and the field that grew out of the door Kuhn opened is the one whose textbook this chapter comes from. Sismondo is writing an introduction to science and technology studies, and this chapter is the origin story of the discipline he is introducing.

A seventeenth-century town under plague. It is closed and cut into quarters and streets, each under an official who may not leave for the duration. Everyone is shut in their house on pain of death.

Every inhabitant is registered by name, age and sex regardless of station. Copies of the register go to three places: the quarter's intendant, the town hall, and the official who calls the roll daily at each window. Each person has to appear and answer to their name. Everything observed on the rounds is written down and passed up the chain: deaths, illnesses, complaints, irregularities. Magistrates control all medical treatment, and no other practitioner may treat, no apothecary prepare medicine and no confessor visit a sick person without written permission.

Five or six days in, the houses are purified one at a time. Everyone leaves. Furniture is raised off the floor, windows, doors and even keyholes are sealed with wax, and perfume is burned inside. The men who did the work are searched on the way out, to prove they carry nothing they did not bring in.

Foucault's summary is the sentence to underline. An enclosed, segmented space, observed at every point, every individual fixed in place, the slightest movement supervised, all events recorded, and an uninterrupted work of writing linking the centre to the edges. He calls it a compact model of the disciplinary mechanism.

Discipline is a technical term for him, and it names a way of exercising power that works by sorting: dividing a crowd into individuals, assigning each one a place, a timetable and a record, and then watching. Its characteristic operation is analysis, in the literal sense of breaking a thing into parts. The plague is a mixture, bodies and diseases and rumours confused together, and the answer to a mixture is to separate it into countable individuals.

The contrast he draws next is the most quotable thing in the essay. Literature has always dreamed of plague as carnival: laws suspended, prohibitions lifted, masks off, bodies mingling, everyone abandoning the identity they were assigned. There is also a political dream of the plague, and it is the exact reverse of the carnival. Regulation reaching into the smallest details of daily life, and the assignment to every individual of a true name, a true place, a true body, a true disease.

Two historical models, two techniques. The leper was excluded, in one massive binary cut, exile, and then left to a mass in which telling one person from another was pointless. The plague victim was included and sorted: multiple separations, individualising distribution, surveillance in depth. Modern institutions inherit the second technique while still using the first's language of casting out.

In 1787 Jeremy Bentham designed a building: a ring of cells around a central tower. Each cell runs the full depth of the building and has a window at the outer wall and one at the inner, so the occupant appears from the tower as a perfectly legible silhouette, backlit. The tower has blinds and angled partitions, so its interior stays dark.

The consequences follow mechanically:

• The inmate is seen and never sees. He can never verify whether anyone is in the tower at this moment.
• So he behaves at all times as if he were being watched, since the cost of being wrong is too high. Visibility itself becomes the trap.
• The prisoner takes over his own surveillance. The power no longer needs to be applied to him, because he is applying it to himself.
• Which makes it automatic, continuous and cheap. Fewer guards, no chains, no violence, no audience. Bentham's phrase for the effect is power of mind over mind.

Foucault insists the Panopticon must not be read as a dream building. It is a diagram: a mechanism of power reduced to its ideal form, stripped of any particular use, a figure of political technology that can and must be detached from any specific application.

Which is why he lists its uses in a single breath: reform prisoners, treat patients, instruct schoolchildren, confine the insane, supervise workers, put beggars to work. Wherever a task has to be imposed on a multiplicity of people, the schema applies. He also notes Bentham's own sales pitch, quoted for its enthusiasm: morals reformed, health preserved, industry invigorated, instruction diffused, public burdens lightened, economy set upon a rock, all by a simple idea in architecture. A reformer proposed this as reform.

And the final point is the subtle one. This power is not a weight pressing from outside on a function it interferes with. It is woven into the function so intimately that it increases that function's efficiency while multiplying its own points of contact. The school teaches better because of it. That is what makes it hard to argue with.

Look at what the plague town and the Panopticon are both made of: registers, roll calls, case notes, timetables, examinations, files. Every one of those is simultaneously an instrument of knowledge and an instrument of control. You find out what a person is like by the same act that fixes them in place.

Foucault's historical claim is that the human sciences were born inside exactly these institutions: psychiatry, criminology, the study of childhood and education. Prisons, clinics, asylums and schools were the first places with captive populations that could be measured over time. So knowledge about people did not arrive from outside and get applied to institutions. It was produced by them. Once that is on the table, "whose knowledge counts" stops being a philosophical curiosity and becomes a political question, which is where Weeks 4 through 9 of the course begin.