August 23, 2004

Towards a Critical Theory of Physics

Via Three Toed Sloth, I see that there is something of a fuss brewing over the anthropic principle's role in cosmology following the distribution of a paper criticising it as inhernetly unscientific.

The argument is made on Popperian grounds, and as Cosma knows I have some fairly serious problems with the Popperian approach. Indeed, Smolin's resumé of the grounds for defining science as identical to falsifiability point to my problems with it.

This point is so basic to how science works that it is perhaps worthwhile taking a moment to review the rationale for it. Few working scientists will disagree that an approach can be considered scientific only to the extent that it requires experts who are initially in disagreement about the status of a theory to resolve their disagreements - to the fullest extent possible - by rational argument from common evidence. As Popper emphasizes, science is the only approach to knowledge whose historical record shows over and over again that consensus was reached among well trained people as a result of rational argument from the evidence. But - and this is Popper's key point - this has only been possible because proposed theories have been required to be falsifiable. The reason is that the situation is asymmetric: confirmation of a prediction of theory does not show that the theory is true, but falsification of a prediction can show it is false.

If a theory is not falsifiable, there is the very real possibility that experts may find themselves in permanent disagreement about it, with no possibility that they may resolve their disagreement rationally by consideration of evidence. The point is that to be part of science, X-theorists have to do more than convince other X-theorists that X-theory is true. They have to convince all the other well trained scientists who up till now have been skeptical. If they don't aspire to do this, by rational arguments from the evidence, then by Popper's definition, they are not doing science. Hence, to prevent the progress of science from grounding to a halt, which is to say to preserve what makes science generally successful, scientists have an ethical imperative to consider only falsifiable theories as possible explanations of natural phenomena.

But this emphasis on coming to a consensus by rational argument is precisely what weighs against a Popperian metatheory of science.

The ugly truth is that science is full of arguments that were never resolved by falsification, consensus or rational argument. The ultimate decision maker in the hard sciences is graduate students. Arguments are resolved when old physicists die without convincing any grad students to continue to work on their theories. Graduate students in the hard sciences need to understand that their profs need them desperately, because it is only through grad students that their work has a future.

This is a methodologically individualist metatheory of science, in constrast to thinking that scientists as a group get together every now and then and hash out their differences, or that there is some sort of courtroom procedure for introducing evidence and making arguments, before some impartial jury. The people who ultimately need to come to an agreement about some scientific theory are the very ones least likely to do so.

It would be difficult for a grad student in physics in this day and age to contest a well established theory like relativity. First, they are unlikely to find anyone else that contested it. Second, they would be bucking peer pressure in an enormous way, so much so that it would likely end their careers. Thirdly, they have nothing to lose by buying into relativity. These are all social causes. These social causes in turn, have roots in old experiments, in problems that relativity solves but that no other alternative has been able to touch, and in all the engineering that you can do with relativity. Relativity is deeply rooted in physics as a culture. Scientific theories win over their opponents when the next generation no longer sees alternative positions as viable. Members of the next generation simply do not come to that conclusion in a vacuum.

Scientists tend to see this as an anti-rationalist conception of science - which is true - and as abnegating their work and methods - which is false. Just because I suspect that science is not really about methods or rational argument doesn't mean that I want to throw it away. Awkward, expensive, clusmy and sometimes just plain bizarre, it still produces desireable goods. If the price of microwave burritos is having to listen to the odd gripe about cosmology, I'm inclined to view it as a small price to pay.

I have no particular position on most issues of cosmology. Personally, I've never much liked the anthropic principle either, but not on scientific grounds. It always struck me as a way to throw up your hands and say "this is unknowable." But, when the validity of a widely-held scientific theory starts to depend on arguments from methodology, I have to start asking how it differs from critical theory? Part of what makes critical theory so frequently unreadable is the need to lay out your political assumptions, theoretical biases and methodological principles out before or at least alongside your argument, because either you or your audience believes that your conclusions are not separable from them. This paper seems to me to start moving in that very direction with its initial citation and justification of falsifiablity.

The level of argument - and of bile - that the anthropic principle brings up is reminiscent of some arcane argument in post-structuralism. Look at the arguments deployed, at how tentative they are: "The hypothesis of early universe inflation gives a plausible explanation of several observed features of our universe, such as its homogeneity and uniformity [...]", "According to the singularity theorem of Penrose, proved on general assumptions [...]", "We also have reasonable, if not yet compelling, theoretical evidence that black holes bounce [...]" Honest to God, this could be linguistics!

For someone whose life and work are just as remote from cosmological theorising as from the disputes between Searle and Derrida, there is a strong temptation to think this is all a lot of hot air about nothing. I remember reading a text on evolution that basically said that efforts to explain the origins of life before the 1860's are best left forgotten. My hope is that someday in the future, cosmology textbooks will say something similar, only I'm pretty sure the cut-off date they use will be after 2004.

Posted 2004/08/23 12:47 (Mon) | TrackBack

> Look at the arguments deployed, at how tentative they are:
> "The hypothesis of early universe inflation gives a plausible
> explanation of several observed features of our universe, such
> as its homogeneity and uniformity [...]", "According to the
> singularity theorem of Penrose, proved on general
> assumptions [...]", "We also have reasonable, if not yet
> compelling, theoretical evidence that black holes bounce [...]"
> Honest to God, this could be linguistics!
I dunno, Scott. It really doesn't sound that different to me from what I see when I go back and read old physics papers from the 1950s --- especially from the more mathematically rigorous sub-communities, who naturally find it harder to get results which are both powerful and established to their satisfaction. (Look at, say, Streater and Wightman's PCT, Spin and Statistics, and All That.) So I don't, prima facie, see a big change here. Obviously that's not how our textbooks talk, now or then...

Posted by: Cosma at August 23, 2004 21:07

Plus ça change, plus c'est la même chose. My undergrad advisor was a defector from an early 60's era crystallography community with a fetish for mathematical rigour that wore off. So, I was aware that none of this was new. Reading the griping about quantum theory in the 40's suggests that there is little new under the sun.

It's just really easy, when you're not an actual practicing physicist and you don't see a lot immediate applications, to step back and say, "Whatever, wake me up in a generation, after the current crop of grad students have tenure, and tell me if anything came of it."

Posted by: Scott Martens at August 23, 2004 21:59

Shorter Scott: Kuhn beats Popper any day.

Which he does, at that.

Posted by: des at August 24, 2004 17:00

The way I think about the Kuhn / Popper thing is to run this thought experiment in my head.

Imagine, one day, we come across a long-lost tribe who have had no contact with our scientific institutions, have no word for "grad-student", have a cosmology entirely different from ours, without atoms or relativity etc.

Yet they have a caste of people who's job is to find out about the world. And who operate more or less in a Popperian manner. Ie. they make hypotheses, invent experiments which test those hypotheses, and if those experiments falsify them, throw them away and come up with something new. Their reasoning is as impecably "rational" as ours.

Would you call this caste "scientists" and what they are doing "science"?

Posted by: phil jones at August 27, 2004 8:09

Phil, no, I wouldn't. I suspect that no experiment they do would ever falsify their belief that the sun turns around the earth, that dense objects fall faster than light ones, or that disease is caused by moral failure. Their rationalism might well confirm - to them - all of those hypotheses and more.

Science is a whole lot more than hypothesis, experiment and the application of reason. It involves a lot of very vague, half-formulated notions like Occam's Razor, a sense of aesthetics, the rejection of dualism and the belief in material causes, and an unsupported, unprovable, unfalsifable belief that causes can be known and explained in a manner which can usefully inform our actions. Actually disconfirming a thesis through experiment approaches the impossible, as Quine pointed out. Real world experiments often involve the rejection of a selection of the data and acts of interpretation that do not flow automatically from the results. Rather, there comes a point where alternative explanations to some theory become more convincing, more useful, or more productive. What makes those explanations more convincing, useful or productive is very cultural in nature. While other societies have often had castes who were dedicated to the discovery of new knowledge and who possessed a certain authority in making some claims about the world, the particular value system and institutional structure we associate with modern science is far more defining of it than either rationalism or falsificationism.

The final validation of scientific theories is that they be sufficiently useful as a guide to action that they become convincing at least to scientists if not to everyone.

Consider, for example, Isaac Newton and his contemporaries on whom we have retroactively bestowed the label of "scientists". They did not put forward or use a particularly falsificationist methodology. Newton, IIRC, believed that he had deduced universal laws of motion from observations. Was he a scientist?

Science is a particular cultural tradition. It is not constant and has changed over time in response to external and internal conditions. Other cultural traditions might, in principle, work as well or better at providing good things, but I have yet to see one that has done as well at providing me with the things I want and at informing my actions.

The word "science" gets attached to a lot of claims these days because it is such a powerful word. Popperian demarcationism has its uses as an arguing strategy, but as a definition of science it falls short. Science is what scientists do when they say they are doing science. Creation science, Hindu science, Islamic science, New Age science - these things aren't science because in some key respect they very strongly violate the cultural values of scientists. Teaching real science instead of those things is an ideological choice to promote the cultural values of scientists. It is a choice which I wholeheartedly support, but it is no less an ideological choice because of it. If some other ideology could do the same work, and produce the same or better results, I might change my ideological commitments. If this remote tribe of yours could convince me that their science produced better results - results that I prefer to those of mainstream science at any rate - I might be induced to convert to their knowledge system. I might call it science, and the practitioners scientists, in order to make that choice more palatable to others. But, that would be tactical choice to label it as such. It wouldn't be science.

I'm not actually a Kuhnian per se. "Paradigm" is a much abused word, especially in linguistics where it has been used to cover up vast ills. Sometimes, the problems a theory creates, the things it doesn't and can't explain, aren't enough to lead scientists to abandon a theory that is productive and useable. This is right and proper and as it should be. But, I object most strongly to Kuhn's all too frequent use in support of theories which produce few useable results and become dubious guides to action. It's much to easy to reject challenges to bad theory by yelling "paradigm shift!" at the top of your lungs until your opponents go away.

Posted by: Scott Martens at August 27, 2004 10:57


It sounds to me you're denying my premise when you say "I suspect that no experiment they do would ever falsify their belief that the sun turns around the earth."

That's not the way I set up the situation. This tribe don't have to be "primitive". Which is what I think you are infering. Let's say they're Martians or whatever. If they come up with an experiment that shows the sun doesn't go round the earth, they'll certainly accept it.

I think the more general form of my original question is this. Is science a "historical particular" ie. a lineage or tradition of ideas and practices that arose in Europe at a particular time (like Latinate languages or common law.) Or is it a "type" of activity which could be recognised as such whenever and wherever it arises. And which could arise many times independently?

Sounds like you're tending towards the first when you say "Science is a particular cultural tradition." But do you really mean that?

Posted by: phil jones at August 27, 2004 19:24

In this post, you fail to explain Popper's philosophy properly and at the same time illustrates his ignorance of the history of science.

Popper addresses the logical question of what people must do in order to learn about the world. He concludes that they must (1) notice problems with their current theories, (2) propose solutions to those problems, (3) criticise those proposals, (4) tentatively accept one of these theories and (5) notice problems with their new theories. Hew summed this up by saying that science progresses through conjectures and refutations. Any of of the stages may itself involve a round of conjectures and refutations. In Popper's view scientific theories are those theories that are susceptible to criticism using observations.

For example, Maxwell's theory of electromagnetism said that atoms should not exist. Bohr proposed a model of the atom in which electrons acted like waves and so only occupied some orbits around atoms. Bohr's model turned out to be wrong as it was too simple in a variety of ways and was refuted by experimentalists. At this point two theories were proposed, one by Heisenberg the other by Schrodinger. In the non-relativistic regime they turned out to have the same predictions and could be put in mathematical correspondence with one another - the resulting theory is non-relativistic quantum mechanics. In the relativistic regime the Schrodinger theory doesn't work and we have to use the Heisenberg version. Of course, quantum theory has its own problems, such as the lack of a viable theory of quantum gravity.

Now, some commentators say that Newton believed he had deduced his ideas from experiments. Obviously this is false because Newton's theories covered infinitely many cases that he would never experience. So his theory was certainly not logically implied by his observations. However, he was impeccably Popperian in what he actually did. He came up with his theory of gravity and tried to fit it to observations of the moon and failed. When he failed he discarded the theory. However, when he found out that better observations of the moon were availble he used them to test his theory and it passed. The earlier observations were themselves conjectures that were later refuted.

Science is about explanations and this helps to explain many of the features that puzzle some people, such as Occam's Razor. If we have two viable explanations of a given phenomenon and one is simpler than the other then accepting the more complicated explanation leaves these added complications unexplained and this is bad, thus Occam's rule that we should not multiply entities, that is explanations, beyond necessity.

The other big objection above is that Scott Martens says that scientists accept that the world is comprehensible and that this is unjustifiable. The main reason for this confidence is realism - the belief that there is an objective world. Compare realism to solipsism. Whether you are a realist or a solipsisty you have to accept that there is a large chunk of the sensations you experience that does not do what you want it to: people who drive badly, people who do not accept arguments, the weather and so on. So solipsism just says that your mind is a very big thing that you do not entirely control and are not entirely aware of, which makes it equivalent to realism in every important respect except that it labels the world as part of the solipsist's mind. So realism is better than solipsism at explaining what we are aware of. Scientists do not provide foundations for their belief in realism or anything else, realism and current scientific theories are just the best theories people can come up with at the time. Nor do scientists have to justify the standards by which they judge theories, they just have to be prepared to alter them in the face of criticism. This does not mean they change their minds every time they come across a new criticism as they may criticise that criticism, just that they should think that their theories might be wrong. Since scientists almost all think that there is a real world, they think that can leanr by trying to compare their theories with the real world.

Some scientists do not alter their theories in the face of devastating criticism, such as the chaps who started the cold fusion debacle. That just proves that no means of learning works for people who refuse to use it.

I should like to end by saying that anyone who wants to know what Popper said should read books like "Conjectures and Refutations" by Popper, "The Fabric of Reality" by David Deutsch and "The Retreat to Commitment" by W. W. Bartley III. Incidentally, I do not recommend Popper's "Logic of Scientific Discovery" as it has been superseded by his later work.

Posted by: Alan Forrester at January 3, 2006 21:00
Post a comment

Remember personal info?