Sunday, July 13, 2014

Falsifying Scientific Models

This post covers a set of general issues about falsification of scientific models and is also meant to be a follow-up to the claims by Bruno Suric in this comment.

I've written on the lame claim about the original Eddington observations of gravitational deflection (see Relativity Denial: The 1919 Solar Eclipse).  Supporters of this claim seem oblivious to the fact that this was far from the only time in history these observations could be done.  The observations have been repeated, and improved upon, in the 90+ years since.  The Hipparcos data has measured the deflection to an accuracy of 0.3%, far more than enough to exclude the Newtonian model of light deflection.  The fact that there were some possible conflicts in the observations suggests future tests must improve the experimental controls to remove the ambiguity.

The history of science is filled with examples where Mr. Suric's interpretation of falsification of a scientific model would lead one down an erroneous path.

Consider another example.  In late-1700s It was found that the orbit of Uranus did not match that expected from Newtonian gravitation when including the perturbations due to the other known planets.  Did that falsify Newtonian gravitation?


While the possibility that Newtonian gravitation did not apply in the outer reaches of the solar system was certainly a possibility, it was not the only possible solution of the problem.

Because one of the other assumptions in the calculation was the motion was the other (known) planets in the solar system.  What if THAT assumption were incorrect?

It turned out to be the case, and the discovery of the planet Neptune was the result of testing that hypothesis  (see Wikipedia: Discovery of Neptune).

This happens quite often in science as many complex models are built with many underlying assumptions.  When the model undergoes rigorous testing, those assumptions are also being tested.

Falsifying the Standard Solar Model With the Solar Neutrino Problem?

Few modern models are the result of only one assumption.  They are usually the result of  a combination of many assumptions, all tested to some finite level of precision.  You might make the assumption that a component of the model is still valid beyond the tested range of precision, and that is often not an unreasonable assumption.  Nature is surprisingly consistent in many of these cases, but Nature is under no obligation to agree with extreme extensions of empirically-tested natural 'laws'.

Mr. Suric claims that the solar neutrino problem should have falsified the entire model of the Sun.  But this demonstrates a lack of understanding of how complex scientific models are made and tested.

What we call the Standard Solar Model has evolved over time as our understanding of the underlying fundamental physics and computational techniques improved (Wikipedia).  The first solar models were computed with pencil-and-paper math and slide rules, integrating the structure equations with large distance steps which had the chance of computation as well as transcription errors.

The ability of computers to do many of the computations repeatedly and tirelessly with smaller distance steps and more accuracy enabled more accurate physics to be included.  Today, some researchers write their own stellar structure codes or run one of the publicly available versions (see references below) that can run on modern personal computers.  I've run some of these models on my laptop.

In the 1960s, at the time the first solar neutrino flux estimates were made, there were a number of other assumptions that went into the standard solar model (see Helium Content and Neutrino Fluxes in Solar Models).  These assumptions were ALSO being tested.
ALL of these assumptions, and more, were being tested in the Standard Solar Model.  After the first solar neutrino measurements revealed a neutrino count significantly less than expected (about 1/3 of the expected value), the other assumptions in the Standard Solar Model were re-examined (see Search for Neutrinos from the Sun).

Other less conventional ideas were also explored.  Could there be a black hole at the center of the sun, accreting and applying the missing energy and reducing the neutrino count?  (see Solar models of low neutrino-counting rate - The central black hole).

An examination of all the alternatives eventually narrowed the problem down to the possibility that the neutrino oscillated between the different 'flavors' requiring the neutrino to have a mass, albeit very small.  We still cannot measure the mass of the neutrino directly, but there are experiments that can put limits on the mass.

In my early undergraduate days, in the late 1970s, this idea as a solution to the SN problem was under discussion.  i worked with a professor who did weak interaction theory (which often involves neutrinos), who, at the time, regarded the neutrino essentially massless.  This is a fundamental assumption in the Standard Model of Particle Physics.

What a difference twenty+ years of technology improvements makes...

Eventually, neutrino detectors were built with sufficient sensitivity that they could detect the other 'flavors' of neutrinos, the mu and tau neutrinos.  With these, we successfully detected these extra neutrinos from the Sun.  Experiments were also conducted using nuclear reactors with known neutrino productions rates, sending neutrinos through the Earth, to be received by a remote detector.  This experiment provided an earth-based test of neutrino oscillations where we had a controlled source (T2K Experiment).   Neutrinos with mass are not part of the Standard Model of Particle Physics (Wikipedia, Neutrinos in the Standard Model).

So the Nobel prize awarded for this work (NobelPrize: Bahcall).  This is ridiculed by Mr. Suric, but have we seen a prediction of the neutrino flux, as well as the stellar structure, from the model HE advocates.  If so, where?  How well does his model hold up under examination with the newer experimental techniques like helioseismology (Wikipedia)?

With these new experiments, a number of cranks have had to resort to handwaving attempts to discredit the experiments (see Tim Thompson's Rebuttal to Donald Scott on the topic).  Yet legitimate researchers continue to use the results of these experiments to build even more refined experiments that not only verify but improve on the precision and extend the results.

That is what REAL science does, and results of previous experiments are subject to repeated testing and retesting. 

Meanwhile pseudoscience is still making excuses.  After all, where is THEIR computation of the predicted neutrino flux of the Sun computed from first principles in THEIR model?  We've seen NO publication of THEIR predictions with how they were obtained, yet they continue to scream their model is more successful!

As I note above, the problem with the Standard Solar Model was solved requiring a revision in the Standard Model of particle physics.  But does that mean all physicists must stop using the Standard Model for nuclear and particle physics calculations?


Because for many other areas in computing reaction rates, etc. the Standard Model works just fine.  It only becomes an issue near the limits of applicability of the model, which we now know is when the neutrino mass may be a factor.

Galileo said that the acceleration of gravity was constant.

When Newton suggested that it varied with distance as 1/r^2, did we stop using a constant for the acceleration of gravity? 


Because even Newton's theory demonstrated that gravity deviated from a constant at the surface of the Earth by an amount so small as to be irrelevant for most practical applications near the surface of the Earth.

Similarly, we still use the Newtonian model of gravitation for planning rocket launches and travel through most of the solar system, because even Einstein's theory gives the same predictions as the Newtonian theory to a precision smaller than the errors produced by aerodynamic and other engineering uncertainties.

Will the Standard Model of Particle Physics eventually be replaced?

Almost certainly!

But you can bet that whatever theory replaces it will generate predictions close, or identical to the current Standard Model in the areas where the Standard Model works well today.

Ad Absurdum
"The theory of the round Earth cannot explain the existence of mountains.  Therefore the model must be discarded.  This is evidence that the Earth is flat."
Most people would regard this statement as ridiculous (except perhaps the late Charles Johnson and his supporters).

Yet this is exactly the pattern of 'logic' invoked by many pseudo-scientists.

Charles K. Johnson & the International Flat-Earth Society
Stellar Structure & Evolution Codes

No comments: