The Structure of Scientific Revolutions
I read the book yesterday. Here’s what I wrote on goodreads:
“I’m not rating this, but I’ll note that ‘it’s an interesting model.’
I’d only really learned (…heard?) about Kuhn’s ideas through cultural osmosis (and/or perhaps a brief snippet of his work in HS? Maybe. I honestly can’t remember if we read Kuhn back then…). It’s worth actually reading the book, and I should probably have done that a long time ago.”
I was thinking about just quoting extensively from the work in this post in order to make clear what the book is about, but I’m not sure this is actually the best way to proceed. I know some readers of this blog have already read Kuhn, so it may in some sense be more useful if I say a little bit about what I think about the things he’s said, rather than focusing only on what he’s said in the work. I’ve tried to make this the sort of post that can be read and enjoyed both by people who have not read Kuhn, and by people who have, though I may not have been successful. That said, I have felt it necessary to include at least a few quotes from the work along the way in the following, in order not to misrepresent Kuhn too much.
So anyway, ‘the general model’ Kuhn has of science is one where there are three states of science. ‘Normal science’ is perhaps the most common state (this is actually not completely clear as I don’t think he ever explicitly says as much (I may be wrong), and the inclusion of concepts like ‘mini-revolutions’ (the ‘revolutions can happen on many levels’-part) makes things even less clear, but I don’t think this is an unreasonable interpretation), where scientists in a given field has adopted a given paradigm and work and tinker with stuff within that paradigm, exploring all the nooks and crannies: “‘normal science’ means 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 it further practice.” Exactly what a paradigm is is still a bit unclear to me, as he seems to me to be using the term in a lot of different ways (“One sympathetic reader, who shares my conviction that ‘paradigm’ names the central philosophical elements of the book, prepared a partial analytic index and concluded that the term is used in at least twenty-two different ways.” – a quote from the postscript).
So there’s ‘normal science’, where everything is sort of proceeding according to plan. And then there are two other states: A state of crisis, and a state of revolution. A crisis state is a state which comes about when the scientists working in their nooks and crannies gradually come to realize that perhaps the model of the world they’ve been using (‘paradigm’) may not be quite right. Something is off, the model has problems explaining some of the results – so they start questioning some of the defining assumptions. During a crisis scientists become less constrained by the paradigm when looking at the world, research becomes in some sense more random; a lot of new ideas pop up as to how to deal with the problem(s), and at some point a scientific revolution resolves the crisis – a new model replaces the old one, and the scientists can go back to doing ‘normal science’ work, which is now defined by the new paradigm rather than the old one. Young people and/or people not too closely affiliated with the old model/paradigm are, Kuhn argues, more likely to come up with the new idea that will resolve the problem which caused the crisis, and young people and new people in the field are more likely than their older colleagues to ‘convert’ to the new way of thinking. Such dynamics are actually, he adds, part of what keeps ‘normal science’ going and makes it able to proceed in the manner it does; scientists are skeptical people, and if scientists were to question the basic assumptions of the field they’re working in all the time, they’d never be able to specialize in the way they do, exploring all the nooks and crannies; they’d be spending all their time arguing about the basics instead. It should be noted that crises don’t always lead to a resolution; sometimes the crisis can be resolved without it. He also argues that sometimes a revolution can take place without a major crisis, though the existence of such crises he seems to think important to his overall thesis. Crises and revolutions need not be the result of annoying data that does not fit – they may also be the result of e.g. technological advances, like the development of new tools and technology which can e.g. enable scientists to see things they did not use to be able to see. Sometimes the theory upon which a new paradigm is based was presented much earlier, during the ‘normal science’ phase, but nobody took the theory seriously back then because the problems that lead to crisis had not really manifested at that time.
Scientists make progress when they’re doing normal science, in the sense that they tend to learn a lot of new stuff about the world during these phases. But revolutions can both overturn some of that progress (‘that was not the right way to think about these things’), and it can lead to further progress and new knowledge. An important thing to note here is that how paradigms change is in part a sociological process; part of what leads to change is the popularity of different models. Kuhn argues that scientists tend to prefer new paradigms which solves many of the same problems the old paradigm did, as well as some of those troublesome problems which lead to the crisis – so it’s not like revolutions will necessarily lead people back to square one, with all the scientific progress made during the preceding ‘normal science’ period wiped out. But there are some problems. Textbooks, Kuhn argues, are written by the winners (i.e. the people who picked the right paradigm and get to write textbooks), and so they will often deliberately and systematically downplay the differences between the scientists working in the field now and the scientists working in the field – or what came before it (the fact that normal science is conducted at all is a sign of maturity of a field, Kuhn notes) – in the past, painting a picture of gradual, cumulative progress in the field (gigantum humeris insidentes) which perhaps is not the right way to think about what has actually happened. Sometimes a revolution will make scientists stop asking questions they used to ask, without any answer being provided by the new paradigm; there are costs as well as benefits associated with the dramatical change that takes place during scientific revolutions:
“In the process the community will sustain losses. Often some old problems must be banished. Frequently, in addition, revolution narrows the scope of the community’s professional concerns, increases the extent of its specialization, and attenuates its communication with other groups, both scientific and lay. Though science surely grows in depth, it may not grow in breadth as well. If it does so, that breadth is manifest mainly in the proliferation of scientific specialties, not in the scope of any single specialty alone. Yet despite these and other losses to the individual communities, the nature of such communities provides a virtual guarantee that both the list of problems solved by science and the precision of individual problem-solutions will grow and grow. At least, the nature of the community provides such a guarantee if there is any way at all in which it can be provided. What better criterion than the decision of the scientific group could there be?”
I quote this part also to focus in on an area where I am in disagreement with Kuhn – this relates to his implicit assumption that scientific paradigms (whatever that term may mean) are decided by scientists alone. Certainly this is not the case to the extent that the scientific paradigms equal the rules of the game for conducting science. This is actually one of several major problems I have with the model. Doing science requires money, and people who pay for the stuff will have their own ideas about what you can get away with asking questions about. What the people paying for the stuff have allowed scientists to investigate has changed over time, but some things have changed more than others and what might be termed ‘the broader cultural dimension’ seems important to me; those variables may play a very important role in deciding where science and scientists may or may not go, and although the book deals with sociological stuff in quite a bit of detail, the exclusion of broader cultural and political factors in the model is ‘a bit’ of a problem to me. Scientists are certainly not unconstrained today by such ‘external factors’, and/but most scientists alive today will not face anywhere near the same kinds of constraints on their research as their forebears living 300 years ago did – religion is but one of several elephants in the room (and that one is still really important in some parts of the world, though the role it plays may have changed).
Another big problem is how to test a model like this. Kuhn doesn’t try. He only talks about anecdotes; specific instances, examples which according to him illustrates a broader point. I’m not sure his model is completely stupid, but there are alternative ways to think about these things, including mental models with variables omitted from his model which likely lead to a better appreciation of the processes involved. Money and politics, culture/religion, coalition building and the dynamics of negotiation, things like that. How do institutions fit into all of this? These things have very important effects on how science is conducted, and the (near-)exclusion of them in a model of how to conceptualize the scientific process at least somewhat inspired by sociology and related stuff seems more than a bit odd to me. I’m also not completely clear on why this model is even useful, what it adds. You can presumably approximate pretty much any developmental process by some punctuated equilibrium model like this – it seems to me to be a bit like doing a Taylor expansion, if you add enough terms it’ll look plausible, especially if you add ‘crises’ as well to the model to explain the cases where no clear trend is observable. Stable development is normal science, discontinuities are revolutions, high-variance areas are crises; framed that way you suddenly realize that it’s very convenient indeed for Kuhn that crises don’t always lead to revolutions and that revolutions need not be preceded by crises – if those requirements were imposed on the other hand, the underling data-generating-process would at least be somewhat constrained by the model (though how to actually measure ‘progress’ and ‘variance’ are still questions in need of an answer). I know that the model outlined would not explain a set of completely randomly generated numbers, but in this context I think it would do quite well – even if it’s arguable if it has actually explained anything at all. Add to the model imprecise language – 22 definitions… – and the observation that the model builder seems to be cherry-picking examples to make specific points, what you end up with is, well…
The book was sort of interesting, but, yeah… I feel slightly tempted to revise my goodreads review after having written this post, but I’m not sure I will – it was worth reading the book and I probably should have done it a long time ago, even if only to learn what all the fuss was about (it’s my impression, which may be faulty, that this one is (‘considered to be’) one of the must-reads in this genre). Some of the hypotheses derived from the model seem perhaps to be more testable than others (‘young people are more likely to spark important development in a field’), but even in those cases things get messy (‘what do you mean by ‘important’ and who is to decide that? ‘how young?’). A problem with the model which I have not yet mentioned is incidentally that his model of how interactions between fields and the scientists in those fields take place and proceed to me seems to leave a lot to be desired; the model is very ‘field-centric’. How different fields (which are not about to combine into one), and the people working in them, interact with each other may be yet another very important variable not explored in the model.
As a historical narrative about a few specific important scientific events in the past, Kuhn’s account probably isn’t bad (and it has some interesting observations related to the history of science which I did not know). As ‘a general model of how science works’, well…
No comments yet.