Science vs "Science"

[ineffabelle]

The reason why physics is a pretty solid discipline is not because physicists are particularly smarter than anyone else, they just have a much better set of mechanisms for checking their stupidity than most fields do.
The thing that makes science ... well, scientific, is the position of maximum possible scepticism that it takes toward all knowledge.  If scientists start losing that scepticism, they become less and less scientific.

(Socio-biologists, for example, seem to be particularly prone to making claims way beyond what the evidence would allow)

Comments

[polyanarch.livejournal.com]

Let's not even talk about climatologists.

[firepower.livejournal.com]

No, please. Talk about climatologists.

[puellavulnerata.livejournal.com]

I'm not quite sure that's true, though. In more applied areas of physics, yes, but the present situation in theoretical physics is that testing new theories experimentally is quite difficult, requiring large, expensive machines useful for nothing else. The rate of progress is limited by that difficulty, and the main motivation for theoretical progress is inadequacies or inconsistencies of the theories themselves rather than inconsistency with experiment.

There's an interesting thing about physics, though, which doesn't hold in most other fields of science: the (metaphorical) Kolmogorov complexities of the systems it describes tend to be smaller than they first appear. A very complex system with many components turns out to obey a simpler law which can be described much more concisely than the superficially visible behavior of the system itself. The corollary to this is that the space of possible descriptive theories is much smaller than the space of possible apparent system behaviors, so if the system behavior differs from your theory by just a little bit, you can't just make some small tweak to the theory to account for it. It often turns out, rather, that your theory will be just an approximation within its particular domain to a very different underlying theory. Thus, one can often make very non-trivial predictions from a new theory obtained purely by considerations of theoretical consistency, and actually be right. The vastly richer logical and mathematical structure of theories in physics gives them much greater predictive power than theories in most other sciences. So, by comparison, things like socio-biology are at best, as Rutherford put it, stamp collecting.

For example, consider general relativity. The Michaelson-Morley experiment debunks aether, and motivates special relativity as a new way to describe electromagnetism consistently. Special relativity turns out to be inconsistent with Newtonian gravity, so Einstein spends the next ten years constructing a new theory of gravity, motivated purely by logical and theoretical considerations: it needed to reduce to Newtonian gravity and special relativity in the appropriate limits, and describe space-time in terms of a pseudo-Riemannian manifold. After doing a whole lot of math, he comes up with general relativity, and makes a whole series of very unexpected quantitative predictions: bending of light, black holes, gravitational radiation, gravito-magnetism, gravitational time dilation, and so forth. They all turn out to be exactly correct as far as we've been able to tell in the subsequent 94 years. This sort of thing never happens in any other science.

[ineffabelle]

By mechanisms I didn't necessarily mean physical mechanism, but that they have a stronger Techne for filtering nonsense out.

[ineffabelle]

Ooo. Actually what you mentioned about Kolmogorov complexities relates completely to what I was getting at. See, when dealing with Biology or, worse yet, Psychology or Sociology, you are operating on a level where things cannot be "nailed down" so to speak, empirically. Measurements, whether direct or even in theory are already insanely complex on the first level. You can't really tell what you're looking at. Nothing can be clearly defined. This leaves either a prohibitively complex set of equations or "educated" guesswork.

[puellavulnerata.livejournal.com]

Yeah. When the space of theories is nearly as large as the space of possible observations, it's hard to tell the difference between a predictive theory and making up just-so stories ex post facto.

[expandonthat.livejournal.com]

I only understood half of what you wrote there...