Walmsley: You contrast the phenomenological – how you manipulate the world and how you want to make the world to be – with the theoretical explanations that we might give these things. In How the Laws of Physics Lie, you seem to attack the theoretical because it didn’t really fit very well with how you could manipulate and alter the world. A lot of people and you yourself at the time described this as an attack on realism in the philosophy of science.
Cartwright: I never described it as an attack on realism, I don’t think. I have described it as an attack on the truth of certain very abstractly formulated principles in physics. I do find that people expect things from these abstract principles that I don’t think the abstract principles can produce. So for instance, it is expected that these abstract principles really do contain within them descriptions of what happens and counterfactually what would happen in the world and that the only problem with them for the purposes of application is that the vocabulary is very abstract. So you have these very abstract descriptions and if you want to build a laser you have to know how to arrange the materials in front of you, and the standard view really does think that when you get the materials in the right arrangement, there’s an abstract description of them and that abstract description feeds into the abstract laws, and the laws really do describe at that abstract level what happens. So the problem for using these abstract principles is one of knowing how to get down from the abstract level to the material, more concrete level. It’s just a matter of knowing how to match up the descriptions.
That seems to me an entirely mistaken view of how these principles come into use. They’re not the way I’ve seen them put to use in laser engineering classes or in articles in laser engineering physics or in laser engineering labs in silicon valley. It seems to me it’s an entirely wrong picture and in so far as we subscribe to it then we’re putting our money in the wrong place.
Walmsley: In the introduction to your latest book, The Dappled World, you’ve been talking about bridging principles and the fact that models can only accommodate certain precise situations and it’s only when we get these precise situations that we have any sort of reason to suppose that the models are actually matching what’s going on in reality, and outside that scope there’s no real reason to suppose that the abstract theory has any kind of applicability. The obvious question to ask is why suppose that these things don’t hold out of the areas that we can’t model very well? We have a situation where the abstract principles work quite well because we have good linking, bridging principles from the world to the abstract principles …
Cartwright: …And vice-versa
Walmsley: …and we don’t have good bridging principles in other situations. But that doesn’t necessarily imply that the abstract laws don’t hold, it just implies that we don’t have good bridging principles.
Cartwright: I agree, it doesn’t necessarily show that there are no such links. It doesn’t necessarily show that these abstract principles don’t apply outside of the narrow range of situations that can be fitted to the models fairly closely, but the question is why you should believe that they do. It’s a very weak criticism to say that a line of argument doesn’t necessarily show that something can’t happen.
Basically, I think it’s a matter of how big an induction you want to make. When we make inductions we normally have grounds for the scope of the induction, and where we don’t have grounds we don’t want to trust our conclusions very much. In this case I think I have got grounds for the scope the limited induction I make because I see again and again these very specific kinds of models where their abstract terms are interpreted in a handful of very specific ways. Those are the ones that have proven to work. Those are the ones that I’m willing to make my induction to. To make a broader induction, it seems to me, a kind of leap of faith.
Walmsley: Can I ask your view then on capacities, because it seems to me that you take away the general laws, but in another one of your books, Nature’s Capacities and their Measurement, you say that laws are just capacities, and these capacities are things that are stable and apply in different situations. I just want to understand what the difference is between capacities and laws, why it is you think that the capacities apply across a range of situations, whereas the laws are only applicable as far as the models work.
Cartwright: One of the things about laws, no matter how you try to recast them, and even the way I recast them, is that they are still fairly rigorous, and we use the information in those laws in certain prescribed ways, articulated ways, in order to make derivations. I looked for the scope of where one can do that kind of exact science, where for what you’ve articulated the things you are interested in can be derived, because that’s a model that has been important to the structure, methods, and nature of natural and social science – the idea that there’s some ideal of exact science and things should be entirely articulated so you don’t have to think anymore about your applications. Whatever warrants you gave for your theory you automatically have a warrant for what you want to do with your theory. It’s kind of a mechanical transfer of warrants. I think that’s been a value, I think it’s a very narrow value and an odd one when you think about it. It’s a value that wants to eliminate judgement, which always seems to me to be very odd.
But that aside, when I think in terms of the language of laws, I’m thinking in terms of this model of exact science which I was brought up with, where you get these very precise predictions that can be drawn and that in principle any situation could have its treatment derived in that precise way.
There I think we have very good evidence that our successful systematic treatments are limited in a very special way. They’re limited to very specific kinds of situations, to just the ones we learn the models for. So my basic idea is if I want to say where this process works, I look at what its proponents think to be its greatest successes and they’re repeatedly restricted to the kinds of cases where there are canonical models.
Now that limits the scope of laws. I also notice now that there are a lot of cases where one uses, as a post-case explanation or even a prediction, in a more qualitative way, something like the law of gravity. Or take the example I use of a pin. I have a pin which is caught in the floorboards, and I want to get the pin out. Now we have no exact law for that. We have no model for that. We can’t use science in this exact way. We can’t use our knowledge as the paradigm of exact science. We nevertheless believe that magnets attract pins and we believe that that belief is encoded in our knowledge of magnetism, and if it’s not too costly, we’ll go and get a magnet to try and get it out and that sometimes succeeds. It’s pretty clear I’ve got no problem with causal inference – it’s pretty clear that in many cases when the pin sticks to the magnet or rises up to the magnet it was the magnet that did it. We have millions of cases like that. What the opposition thinks those cases show is a very dramatic conclusion. The opposition takes it that those cases show that there are abstract, exact laws somehow in God’s mind, or God’s physics, or God’s engineering text, that there is a real model for this situation with abstract descriptions which are really true for the situation and that once you apply those abstract descriptions, the whole thing fits into a model for which one can make precise predictions. A view that takes cases like that and says, “But look, we see that the laws of magnetism apply everywhere.”
I look at it a little more closely, and I don’t see that we use the law of magnetism there. What we use is a much cruder, more qualitative claim that magnets attract pins.
So the question is, what about the scope of capacities versus the scope of laws? I think the answer was somewhat too complicated. It’s really rather that if you think of laws in terms of those things with which we are at least in principle able to make precise predictions, then there are a lot of cases where we use the information in some sense in the laws, where we have no idea one way or another whether we can in principle make precise predictions. That’s the point – we don’t know. The person who is very optimistic about the scope of laws assumes that even though we don’t know the abstract description and we don’t use the abstract description on this occasion, it must come under this abstract description. Well, as I’ve said, we don’t know it, we don’t use it and this situation doesn’t fit in any apparent way the models which we do use to bring things under an abstract description. So why do you believe that it must?
The language of capacities seems to me, as I develop it, much closer to the way that we actually reason without making these additional terrifically strong metaphysical assumptions. It’s weaker notion than this notion of exact law and it allows us to account for what we successfully do without having to have as big a metaphysical leap as the person who believes everything has an abstract description which brings it under the laws of physics. As I said the reason that I’m opposed to big metaphysical leaps is because sometimes they cost you.
Walmsley: It strikes me that you’re always on the back foot, because it seems that science has progressed, or things change in science, and there are more and more precise descriptions, and more and more comes under these abstract laws. Compared to the seventeenth century when this sort of precise science started with the Newtonian revolution, we could cover very little then and we can cover a lot more now, and we’ve only just got started.
Cartwright: That’s true. If you look at how in fact how we do cover it, there are a lot of things to note. We have invented new theories to cover new things. We’ve got electromagnetism. Newtonians thought you could do everything with Newtonian physics. There was a big Newtonian movement which lasted for a very long time, in medicine in the study of electricity and in the study of everything. We have not learned that Newton’s laws cover everything. What we have learned is that there’s a bunch of phenomena that we know how to model in a certain way, that we can use certain abstract principles of electromagnetism to study. We’ve also learned in certain very specific kinds of models how to bring the two together. We’ve learned how to correct Newtonian mechanics, in certain kinds of cases when we say relativistic corrections. We’ve learned certain kinds of cases where we’d want to use the general theory of relativity. So I’m not opposed to the idea that we’ve learned more.
I like to think of the great physicist, Lord Kelvin, who hated Newtonian physics. He believed in electromagnetic ether. He thought of the world as floppy, elastic, continuous and that’s not the kind of world that’s well treated by Newtonian mechanics. His view was to chuck out Newtonian mechanics and try and do everything with his theory. My view is that Newtonian mechanics works very well for just the kinds of things which Kelvin said didn’t make up the bulk of the world. It works very well for rigid rods, compact masses, springs etc. and there’s no reason to deny that it works very well there. It doesn’t mean that it’s going to work for elastic fluids. If you have a very good theory for elastic fluids, it doesn’t mean that you’re going to be able to take over the study of compact masses, rigid rods etc.
Walmsley: I think this is leading to the your view in your latest book, in which you describe a “patchwork of laws”, a “dappled world”. You postulate that it might have been the case that God decided to create patches of order here and there, bits and pieces of things we can sort out, and these are the sorts of things we have good evidence to believe in. You seem to be saying that lots of the parts of the world, parts we can’t model, are subject to happenstance. There’s no law that covers these things at all.
Cartwright: Maybe. I think you can produce no good empirical or philosophical reasons to settle the issue and to support your view that it is subject to law. There are no good arguments that make it more likely that way than any other way.
If you independently knew everything was law-governed, there are lots and lots of ways of reconciling that with the mess we see in most cases in the resistance things have to coming under laws, and in particular coming under a particular favoured set of laws. You could always reconcile it. The question is you don’t know that. You have to argue to that position, and the evidence of the world around you points much more immediately and directly to a dappled world than to this leap that behind the muddle of appearances there’s a universal order. My picture is that probably real nature is not the same as apparent nature, there are pockets of order there are pockets of disorder. And about any particular problem I would like to settle that the actual, on the ground, promising research programmes.
Walmsley: Would you not concede though that there is a possibly a heuristic value in assuming there’s kind of order behind various appearances. One example that springs to mind is the unification of Keplerian planetary motions and Galilean theories abut falling bodies. They were put together under Newton’s theories. There you had what were considered to be massively diverse phenomena at the time and they didn’t seem to be related at all in any way shape or form. But Newton found some underlying order there and if he hadn’t been interested in looking for some unified theory then we wouldn’t have had a huge leap forward in scientific understanding. So isn’t there then at least heuristic value in assuming order.
Cartwright: No, I don’t think there is a heuristic value. I think there’s a heuristic value in looking for what’s right and sometimes it will turn out that unification is right and then there was a good value in hunting for it, and when it turns out that unification is false, then a lot of money and a lot of time and a lot of effort could have been put into another programme.
We have certain theories that are kind of take-over theories: genetics, various kinds of quantum mechanics, Darwinians’ approach to the social sciences, game theory in economics – these are take-over programmes and people like take-over programmes because people tend to believe in unification, that there is going to be one theory that accounts for everything and that everything is subject to order and that moreover is a simple order. I think that very often these take-over theories get used when they ought not, they get funding when they ought not. When you’re thinking about whether to fund a project, you need to look in great detail at the research programme proposed and the promise of it and that means at looking at how good the ideas are. I think that we’ve got good reason to think that often these take-over theories get an extra dollop of support just because they are take-over theories, and that seems to me dreadful and that’s a disaster because it’s substituting a badly established philosophical view for positive evidence. It’s leading you to bet on something rather than something else, or invest more money in it than in something when you don’t really have reason to do that. In the book I say that I think that women are dying of breast cancer because we’ve put too much money into the gene programme. I also think that money goes into the wrong bits of physics..
Walmsley: One of the things you’ve been emphasising is practical application. Would you attribute this to your work on Otto Neurath which has been very prominent in your work lately?
Cartwright: I think I share with Otto Neurath a view about what the point about doing philosophy is. That philosophy is not to be driven by the history of philosophy. That’s the context within which one does philosophy, so you’re trying to think about and resolve variable questions in the history of philosophy that have been raised by previous philosophers across the centuries. That’s the context within which you locate what you take to be a philosophical problem. I have a quite different view that the context for doing philosophy, say philosophy of science, is to help solve and participate in contemporary scientific problems; the context for moral philosophy is to participate in the understanding and solution of contemporary moral problems. That’s the way in which Neurath approached philosophy and when I discovered him it was a great treat for me.