L'astronomie montre que la science est reproductible en dehors du laboratoire

Article du mois : janvier 2009

par
W.T. « Tom » Bridgman 

Objet :    | L'astronomie comme science « non prouvable »
Date :       | 29 janv. 2009
Message-ID : | 0001HW.C59BBAC9001CB174B019F96F@news.verizon.net

Astronomy is predominantly an observational science. We only know about objects when some effect from them arrives at the Earth (or near the Earth and detected by satellites). I've had some creationists claim that this makes astronomical knowledge 'unprovable' and therefore not worth considering as 'real' science.

First of all, science doesn't 'prove' anything. Those who like to use this term try to use it in an 'absolutist' way, such as with mathematical proofs, when their actual model is the much weaker standard of 'proof' used in the legal system. The weakness of legal 'proof' is demonstrated by the fact that the legal system usually has a system of appeals, except perhaps in totalitarian states. Those who distort the definition of 'proof' often like to mangle the definitions of scientific 'theory' in much the same way. To avoid diverging too much on this side issue, those interested should read:
Back to Astronomy...

We learn about the distant cosmos by two primary methods:
  1. photons arrivant d'objets lointains
  2. particules à haute énergie (rayons cosmiques) arrivant d'objets lointains
For both of these methods, the information we measure includes their flux (number of particles per second per unit area), their energy (wavelength or frequency), direction and sometimes polarization.

Our ability to interpret what is going on 'out there' depends on one basic assumption: that it is physically consistent.

One characteristic of this physical consistency is our ability to predict some characteristic(s) of the system's behavior based on physical principles that can be expressed in mathematical form. That this mathematical process works as well as it does has been the subject of ongoing philosophical debates, such as described in « L'efficacité inouïe des mathématiques dans les sciences naturelles ».

How do we test this physical consistency? The primary method is that the objects we observe behave the way we predict. For example, we can predict the locations of planets in the solar system with excellent precision, decades in advance. This capability is used in everything from eclipse prediction to interplanetary probes.

Occasionally, we encounter situations where our predictions appear to break down. To maintain physical consistency, there are three possibilities scientists generally consider:
  1. certaines approximations utilisées dans le calcul mathématique se situent en dehors de leur domaine de validité et introduisent des erreurs plus importantes dans le calcul. La résolution de ces problèmes peut nécessiter des ressources informatiques accrues, qui peuvent ou non être disponibles ;
  2. il existe une physique supplémentaire connue, non incluse dans les calculs originaux, qui devrait l'être ;
  3. la possibilité la plus excitante de toutes : une nouvelle physique est détectée.
The history of science is loaded with examples of 'great problems' where the solutions were found through this method. Sometimes, finding these solutions takes a few years, but the process has been known to take decades. Notice that 'supernatural intervention' is not on the list. To illustrate this process, I will focus on two historical examples:

Neutrinos solaires
Initial measurements of solar neutrinos were one-third the expected fluxes based on nuclear reactions required to produce the observed luminosity of the Sun. Some creationists tried to claim that this was evidence that the Sun was not powered by fusion and could therefore not be billions of years old (see Preuves d'un jeune Soleil, by Keith Davies). It would take over three decades, but real scientists would continue perusing the 'naturalistic' solutions, examining possibilities (a), (b), and (c). A number of refinements were made pursuing possiblities (a) and (b) but they did not improve the agreement in a significant way. Eventually, theories and experiments began to favor option (c), culminating in some experiments to actually test the hypothesis of neutrino oscillations, where the neutrinos oscillate between different 'flavors' when traveling through matter. Eventually experiments detected the mu- and tau- neutrinos from the Sun, which were created by oscillations of the electron-neutrinos created in solar fusion reactions. This was achieved by the Observatoire des neutrinos de Sudbury. An Earth-based experiment measured neutrino oscillations more directly by passing neutrinos through the Earth to a detector ( K2K Expérience d'oscillation de neutrinos à longue base officielle). Also check out the Page ultime sur le neutrino.

La théorie de la gravité
Newtonian gravity was a great success in predicting orbits of solar system objects and even discovering the planet Neptune based on deviations from the predicted orbits. However, the orbital deviations of the planet Mercury proved harder to explain as many searches for a perturbing planet met with failure. The solution would eventually be found by Albert Einstein in 1915, during the development of his Théorie générale de la relativité. General Relativity was tested through a number of astronomical observations before it was possible to test in Earth-based laboratories. Today, general relativistic effects must be included when computing the signal propagation times in the Système de positionnement global. Also check out Relativité dans le système de positionnement global by Neil Ashby which illustrates the details of how this is done.

I have described a number of similar astronomical discoveries in my paper, « Le Cosmos dans votre poche. Comment la science cosmologique est devenue une technologie terrestre I », available at the Cornell Preprint server.

Also important in scientific testing is the question of reproducibility. We cannot 'reproduce' astronomical observations in the laboratory sense. However, there are other ways to solve the reproducibility question, such as:
  1. Nous testons en observant d'autres objets similaires, comme dans le cas des supernovae et des sursauts gamma ;
  2. Nous pouvons effectuer des observations plus détaillées du même objet, peut-être à de nouvelles longueurs d'onde, avec des résolutions angulaires et temporelles plus élevées, accumuler des ensembles de données sur des bases plus longues, ou examiner d'autres propriétés observables telles que la polarisation des photons ;
  3. Si l'idée implique de nouveaux processus atomiques ou des particules, nous pouvons tenter de les reproduire ou de les détecter dans des expériences contrôlées sur Terre (c'est-à-dire l'astrophysique de laboratoire).
All of these provide feedback that can reinforce or invalidate our interpretation of events distant in time and space.

All knowledge is accumulated indirectly, even when conducted in laboratory equipment. The constituents of atoms have never been seen, only inferred from many experiments and observations. We construct mathematical models of these particles that can reliably reproduce measurements, both past and future.

I can't prove that we aren't all brains in a vat, providing biochemical energy for some supercomplex machine, and our 'reality' is just a very sophisticated VR program, such as in the movie, La Matrice.

There is no 'proof' of the theory of gravity. There is no 'proof' of quantum mechanics, nuclear physics, atomic physics, or any other science. There is only an overwhelming amount of physical evidence that it works. While they work today, tomorrow, we may have new experiments that hint at something beyond our 'standard models' of these phenomena.

Yet we build microelectronic circuits, nuclear reactors, and launch satellites into space and to other planets, using these 'unproven theories'. These same 'unproven theories', taken together, give us the great age (over 13 billion years) of the Universe. In spite of creationists' denials, these 'unproven theories' have made modern technologies possible.

So to those who wish to argue with me that astronomy is 'unproven' as a science, I insist that you provide me with PROOF, not just evidence, of the reality of electrons, protons, and neutrons. To make things more interesting, perhaps I should insist on PROOF that these subatomic particles are not, say, magical pixies that just happen to behave in ways we observe but which could change their behavior at any time. If you can't prove this, why are you using any microelectronics technology?

I'll close with an applicable quote:
De plus, le terme « fait » ne signifie pas « certitude absolue ». Les preuves finales de la logique et des mathématiques découlent de manière déductive de prémisses énoncées et n'atteignent la certitude que parce qu'elles ne concernent pas le monde empirique. Les évolutionnistes ne font aucune revendication de vérité perpétuelle, bien que les créationnistes le fassent souvent (et attaquent ensuite notre style d'argumentation, qu'ils favorisent eux-mêmes). En science, le terme « fait » ne peut signifier que « confirmé à un tel degré qu'il serait perverse de refuser un assentiment provisoire ». Je suppose que les pommes pourraient commencer à monter demain, mais cette possibilité ne mérite pas un temps égal dans les salles de cours de physique. -- S.J. Gould, « L'évolution comme fait et théorie »

[Retour aux articles du mois de 2009]