You misunderstand what I said on this: the kinematic model of temperature is science, because it works with the repeatable, since statistically the individual movements of molecules can be averaged: it is the individual movements of the molecules that are unpredictable.[quote=“glipsnort, post:73, topic:5784”]
I can observe that genetic variants that confer lactase persistence increased in frequency in humans much more quickly than can be explained by genetic drift, that this happened multiple times, and that it only happened in populations that were practicing herding.
[/quote]
This really (once more) makes my point: lactose tolerance is almost a model system in its simplicity, and population genetics can therefore predict it well. Without reading up on it specifically, I assume that the tolerance depends on one or a very few alleles, which were fortuitously (and, perhaps inexplicably: contingency waves its hand for attention) present in some proportion of the pre-pastoral population - presumably a sufficiently high proporion to make drinking animal milk customary, rather than taboo: populations don’t keep on trying to drink stuff that makes them ill in the hope of a mutation arising at some stage.
A population geneticist could hardly imagine a simpler case: a single, entirely new, food source appears amongst humans (whose love of manipulating the world makes them almost the only species one can envisage trying such a thing), which because of the advantages of pastoralism spreads steadily across the world. The genetics are, I assume, simple - certainly the phenotypic trait is. Pastoralism, clinical tolerance and allele frequency are all easily tracked both now and from historical records, and the scientific population genetics model works. Likewise, in other restricted cases such as genetic bacterial resistance, Lenski’s E. coli citrate tolerance etc. One can also add that the unusually vast interbreeding population of H sapiens not only assists the maths, but also buffers other evolutionary change.
But in the case of macroevolution in the wild, multiply up the number of genes (and non-coding elements) all undergoing unpredictable change at once, their mutual interactions, the wide range of phenotype changes they mediate, the additional mechanisms like neutral change - and in many cases the complexities of understanding what role any gene really plays in a trait, and what role any trait plays in survival (one common and much studied, and disputed, example - the stripes of the zebra). And then multiply up all the myriads of factors in an ever changing environment. And factor in the likelihood that the organism creates its own environmental niche. And then add in the difficulty that, for past events, much or most of that information is unavailable.
What one is left with is a mass of contingencies which cannot possibly be known individually, and which exhibit far more variables than the movements of individual molecules in a gas. Remember, firstly, that the original claim I challenged was that natural selection is not random, and I still argue that it is, as far as the scientific definition of randomness is concerned, that is, “unpredictability”.
Secondly the principle I suggested was that randomness (and hence contingency, which is random because unrepeatable and irregular) is the point at which science must cease to comment about efficient causation, because “unpredictability” is not an efficient cause, but an admission of ignorance of the causal chain.
James Clerk Maxwell, the father of statistical science, was well aware of this, separating the science that could be done statistically from the unknowable contingencies, which he placed in the realm of divine providence rather than science, by saying:
Would it not be more profound and feasible to determine the general constraints within which the deity must act than to track each event the divine will enacts?
So you see, I am not completely at odds with a historical understanding of the philosophy and theology of science in raising this.