[Martin Taylor 2008.12.15.13.39]
[From Bill Powers (2008.12.15.1324 MST)]
There are different levels of "laws of nature" (which are, of course, all human-made). ... The second law of thermodynamics is a summary of observations, not a proposed mechanism. Nobody has ever, to my knowledge, proposed a model to explain why entropy increases.
The name "Boltzmann" springs to mind. After Boltzmann, the question goes the other way. It becomes one that applies to pretty well all phenomena that Kelvin would have agreed to belong to "Science". Why is it that the laws of mathematics so phenomenally well apply to observations of the natural world? To take an extreme example, how was it possible to predict and observe a few flashes deep underground, more or less coincidentally with the observation of a new star in the sky?
Boltzmann changed the question from "Why does this abstraction we denote as "entropy" always increase" to "Under what circumstances might we ever expect to see entropy increase", and answered it with "when the system under observation has very few degrees of freedom or is subject to external influences". The reasoning is as purely mathematical as was Einstein's in postulating an equivalence between mass and energy, thereby changing the concept of the conservation of energy by including mass as a form of energy.
Now, Boltzmann obviously didn't propose a "model" in the sense of simulating in a computer the behaviour of myriads of independent interacting entities, but instead made the mathematical argument (almost a tautology) that it is more likely for the system to move from a low-probability state to a high-probability one than the reverse. If you choose not to call this a "model", you are at liberty to make that choice, and you would thereby justify your claim. However, if you do restrict your use of "model" to such a simulation, then you have to run your model for all possible initial states, and when you have observed all the results, you still would not have an explanation. I will take Boltzmann over that as "Science" any day.
Martin