increpatio
|
|
« Reply #19 on: 21:00:42, 16-09-2007 » |
|
So trying to take an active interest in figuring out what all this "science" stuff is about (not being a scientist myself), I read some books by Kuhn and Popper; first "The poverty of historicism" by Popper, which contains a rather concise statement of his views on the scientific method, followed by "The Structure of Scientific Revolutions" by Kuhn. I just purchased, but have not yet read "The Logic of Scientific Discovery" by Popper. I can't claim the following to be much more than an assemblage of paragraphs: I have tried to arrange them as much as I can, but don't have the time to rearrange this into a cohesive essay.waffle waffle waffle
So I'm quite sure I am passingly familiar with most of the tools in the scientific repertoire: theorising, experimentation, induction, &c.. Viewed from this perspective, my main interest is in what new tools Popper and Kuhn introduced to the scientific method.
Popper's clear new idea is that of falsifiability: something that to me does have a feel different to any of the existing concepts though, of course, it does relate quite closely to the sceptical tradition. I don't think that it does encompass the entire framework of scientific deduction. For instance, if somebody is looking for the presence of water on mars in some reasonable way, I would consider him a scientist. Here his chief goal is to confirm this rather singular statement rather than attempt falsify it. However, Popper does not consider this science, but rather historical.
I think that this distinction stems from his belief of falsifiability as being characteristic of science. He says that historians can *use* various theories to describe things, but they use them "as a rules, without even being aware of them". That is to say that they do not try to test *anything*, but merely wish to gain some comprehensible description of what's going on and what, from their perspective, is going on by applying these laws.
There are distinctions to be made, but I do not think that denying these various activities the title of "scientific" is a reasonable thing to do.
As opposed to this distinction between historical work (a lot of data-gathering could be considered historic) and scientific work that Popper makes, Kuhn has his own one, that of the difference between pre-revolutionary
Kuhn has his own categorisation: of "pre-revolutionary" and "post-revolutionary" science; in the first people have problems, they're trying to find a good theory; lots of incompatible proposals are put forward, &c., in the second we have "regular science" (="post-revolutionary science"), where people in a particular are working exclusively with one theory retesting it, and investigating its implications and, probably, eventually coming up with some issues that can't be resolved.
There are some parallels here between these two views, but the divides aren't really the same at all.
Another classification that I haven't seen either mention is be the experimental/theoretical science divide. In the former people are looking for basic laws and relationships from data and, in the latter, people are working with existing analytic models seeing what they can do with them. New theories can emerge from the latter when two models seem to be incompatible: take special relativity for instance as a theoretical reconciling of electrodynamics and Newtonian physics. Of course, once put forward, there had to be experiments proposed that would allow people to distinguish it from Newtonian physics (It's fully compatible, I think, with electrodynamics), and blah blah blah. And nowadays there are theorists trying to reconcile the standard model of particle physics with general relativity: even though there's no data yet that shows that either model is *wrong* in any convincing way.
Kuhn's main claim to novelty is that of the "Paradigm": it's not quite as novel as Popper's, and I do find the term (as I do with much of his terminology) a little odd (he actually uses the term "paradigm "to mean several different things, but the initial one that prompted him to the term describes the small number of important examples/experiments that scientists give that explain or exhibit any scientific theory, often picked so that they exhibit how the theory is different from prior or 'competing' theories (so statistical physicists talk of heat engines, field theorists of scattering experiments, relativists talk of diffraction of light by the sun or about gravitational waves, and so on).
I personally do not like the term "revolution"; this is not to say that I believe in the constant progress of science, of a process of constant, compatible incrementation of knowledge. I might have idly entertained such ideas before reading Kuhn however, but no longer.
I think that I regard Kuhn's work as being much more scholarly than Popper, much more reasoned and cohesive, though, at the same time, neither as effective or as inventive.
On one topic where I find myself disagreeing with both, and where they agree is that scientific concepts are not chiefly the product of investigation: that investigation is simply impossible to carry out unless one has decided what it is, exactly, that one feels wants to investigate, and conversely if one decides that one's theorising were misled, one's experimental data might count for very little indeed (the chief example Kuhn gives being that of the discovery of X-rays by people experimenting with cathode tubes; the admission of these forms of light that were not taken into account previously and could have interfered with many results obtained before then and so a lot of retesting was called for).
Popper, when talking about social sciences, seems not to be talking too deeply or convincingly about their matters. He says that the predictions cannot be both "exact and detailed". But that's true of a lot of (all?) physical phenomena as well; they can only make exact and detailed predictions in simple circumstances, otherwise statistical (i.e. not detailed) predictions will be the order of the day for chances are that owing to chaos blah blah exact predictions will be of bugger-all use.
More interesting, statistical results in sociology indicating that, surely, there *must* be some more fundamental laws that can be found? But this not yet known to be true of QM, and most physicists are relatively accepting of the statistical nature of QM.
In some sense I prefer Kuhn's view of scientific research: that of the evolution of theories: that is to say that theories are not trying to *go* towards some perfect theory, but rather are trying to better existing ones. When popper says that "it is an important postulate of scientific method that we should search for laws with an unlimited realm of validity" I understand what he's saying to some extent, but to some extent I would rather more sympathise with a statement along the lines of "we should search for laws that have a greater realm of tested validity than the current laws".
The idea of falsifiability does derive a little from the frequentist approach in statistics. The alternative, Bayesianism, can work with the idea of confirmation just fine. Popper's main issue is that if one wants to confirm, say, Newton's laws, one might just swing a pendulum and make measurements. The frequentist method of induction would accept this as providing new evidence for Newton's laws each time it's done, and so it needs this extra impetus, that of falsification, to ensure that it doesn't just get stuck doing the same thing again and again. Bayesian inference, however, takes in to account subjective viewpoints, and how "surprised" a person would be if a result were true: this means that repeating the same experiment for the millionth time is unlikely to prove anything once people are familiar with and much expect the result.
In relation to this falsification malarky, as Kuhn points out, never (according to Kuhn) has the falsification on a theory resulted in its rejection: it is only rejected if a better theory is found. That's not to say that such experiments are fruitless: they point out where work needs to be done. This might require the adoption of a new theory, or merely a different explanation by the existing one, or maybe a mild adjustment of the theory (take general relativity and cosmological constant). Popper has a lot to say in response to this, but I have not read his views just yet.
|