The folly of defining randomness randomly

After 190 or so posts  in the BioLogos thread to which I refer in the last two columns, I’m still not convinced that my central point has been answered in the to-and-fro about the definition or modelling of natural selection.

My provocative contribution was intended to focus attention on whether “contingency” in nature can truly be said to be a subject for science proper, which is the study of patterns of regularity in nature. Contingency, by its very definition, is what is left over once the patterns appear to run out. I introduced this in relation to natural selection in my comment, replying to GD:

Is natural selection a regularity in nature, or a near-infinite series of mostly incalculable contingencies?

If the former, its results ought to be predictable according to scientifically determinable laws (as opposed to laws that predict how a feature or gene will behave, assuming it has been selected.)

If selection cannot be reduced to laws (ie it is not a lawlike process), then it would seem science can only (in the main) observe it in the same way that one can observe historical events, such as the results of wars or elections, without being able to formulate laws of history that explain reliably who will win any particular struggle and why.

After much discussion, various scientifically-orientated contributors denied that natural selection is purely contingent (and therefore it is amenable to the maths of population genetics), whilst admitting that, in practice, the systems involved are in many cases too complex and chaotic to be accurately modelled, or for predictions to be made. This to my mind simply confirms my point – that what is not, in reality, transparent to science cannot simply be included in science as if it were transparent.

All along, natural selection was merely an example: the issue which interested me was where science was willing to admit the limits of its epistemology, or what it can know (“science”, of course, being a synonym for “knowledge”). It’s not clear to me how one can claim knowledge of that about which one admits ignorance, such as living systems exhibiting organised complexity, nor how one can make truth claims about what actually lies beyond ones means of study.

We can focus on this by remembering, again, what science actually means by “randomness”, and how easy it is to forget that definition. Here is how Futuyma defined it:

[S]cientists use chance, or randomness, to mean that when physical causes can result in any of several outcomes, we cannot predict what the outcome will be in any particular case. (Futuyma 2005: 225)

In other words, as I have frequently been saying in this discussion, scientific “randomness” is about ignorance of causation – epistemology – and therefore can simply have nothing to say about the cause, or lack of it, of what it calls “random”. Whatever is random – what cannot be predicted – therefore lies outside scientific epistemology. Let me remind you that this definition of “randomness” also matches pretty closely that of “contingency”, for what is contingent, rather than being a regularity in nature, is unpredictable. Contingency is randomness.

The failure to appreciate this is exemplified, I think, in another BioLogos thread in which a “troubled soul”, Emily, asks (more than once) how evolution can be God’s work if it is random.

A M Wolfe seeks to reassure her by a reiteration of the common argument (even Dawkins uses it) that evolution is not random, because natural selection is not random. In the light of my discussion above, see if you can understand exactly how the word “random” is being used in the discussion. He writes:

Mutations are random, as best we can discern, but which mutations survive is not random, because the environment naturally tends to favor certain sorts of things and not others.
…
This means that the overall pattern that emerges over time will be one of life adapting amazingly well to new environments. This overall pattern is not random! But it uses a random algorithm to get there. Does that make sense?

Now the answer is intended to explain the science scientifically, but “randomness” is used in some non-scientific manner, which in the end gives no real idea of what God might or might not be doing in evolution.

If Futuyma’s definition is in mind, “Mutations are random, as best we can discern” means “As far as we know, we do not know what causes mutations”. That sounds a bit odd, and I can only suspect that what was really intended was not an epistemological, scientific understanding of randomness, but an ontological one: “As far as we know, there is no real cause.” And that is to attribute knowledge to science that it cannot possess.

Incidentally, “random algorithm” is itself an interesting concept – human so-called “random number generators” either use complex algorithms that cover the tracks of their mathematical patterns well, or deterministically use numbers generated from quantum events believed to be ontologically random. The algorithms themselves are anything but random – how could they be when the definition of an algorithm is “a process or set of rules to be followed in calculations or other problem-solving operations”?

In point of fact, there are clues that mutations do have some lawlike causes – which is one reason more pedantic biologists insist that mutations are only “random with respect to fitness”: that we can discern no pattern linking mutations to function. But even on that more precise formulation, the “randomness” means no more than “scientists can see no pattern”. As far back as Asa Gray, and as far forward as James Shapiro, teleological patterns behind variation have been suspected.

Turning, then, to the supposed non-randomness of natural selection that can only mean, speaking to Futuma’s scientific definition, that the patterns of natural selection are (unlike mutations) predictable and lawlike, because “the environment naturally tends to favor certain sorts of things and not others.” But this comes in the wake of the first long thread I mention above, in which it was conceded that the details of natural selection in the “wild” environment are, in practice, unknowable.

Perhaps one might reply that natural selection in theory predicts the survival of the fittest. But that, once more, is putting the theoretical horse before the lawlike data. To Darwin and his generation, natural selection was a totally efficient and constantly attentive overseer, ensuring that all living organisms were moving ever closer to whatever was optimal to prosper in their environment. What that optimality might be in any given situation might be contingent, but the principle of natural selection itself was absolute.

Only extreme adaptationists believe that now. Reading around the literature, it’s clear that several variations of neutral theory are, at the very least, jostling for space with adaptive selectionism – and in many cases, the role of selection in evolution is very much downplayed. The only uncontested area seems to be that of purifying selection – that drastically bad stuff dies.  A consistent principle, but not an explanation of life.

It is held under neutral theory that many favourable mutations never become fixed in populations and are lost, and that many mildly-deleterious ones do. As a result, there are many claiming that organisms are, for the most part, only rather poorly adapted to their environment: enough for survival, but very far from optimization. The environment may well tend to favour certain things (just as the structure of DNA undoubtedly favours certain mutations more than others), but other forces in the environment may overturn those “favours” to a greater or lesser extent, making the environment, like mutations, unpredictable with regard to fitness.

And remember, folks, “unpredictable” means “random”, in science.

The discernment of overall patterns does not alter this randomness. Evolution is sometimes compared or contrasted with weather. It’s certainly true that there are patterns in weather which make some prediction possible, whether that be the proverbial wisdom that “red sky at night, shepherd’s delight” or the truth that here in Britain the prevailing wind is south-westerly. But we all know that forecasting is limited by the fact that weather is a chaotic phenomenon. As Wikipedia (sadly similarly confusing epistemological/scientific and ontological randomness) says:

Chaotic systems are predictable for a while and then ‘appear’ to become random.

No – epistemologically they do become random, and that’s the only kind of randomness that science can talk about. The weather is, on Futuyma’s definition, random because we can’t predict it, whatever we may believe about the ultimate compliance of chaotic systems with known laws of nature. It’s comparable to the fact that a dice throw is scientifcally random, even though we think we know full well that full measurement of all the forces involved would give us the outcome every time. Since we don’t, and can never, have that knowledge, we’re still ignorant, and the dice throw is always “random”.

Likewise, since there are both chaotic and “organised complex” (to use Hayek’s term) systems interacting in natural selection, any assumptions about its “lawfulness” are just that – assumptions and beliefs. As far as knowledge, rather than idealised Darwinian theory, goes, natural selection is no more or less random than mutations. The key being “we don’t know why, and can’t predict how, certain mutations happen; and we don’t know why, and can’t predict how, certain environments will act selectively.”

Now in saying this, please understand I’m not trying to deny science, though I am trying to stop science wrongly making statements about ontological randomness and making it seem, wrongly, that science presents any problem to theology regarding randomness. Why should it be a problem to anyone’s faith, even in principle, that there are contingencies whose causes scientists can’t fathom?

My point is that, as far as reassuring someone like Emily on the BioLogos thread goes, the idea that “mutations are random but selection is not” lacks any substantive content, as well as being dubious scientifically.

What needs to be said, instead, is that where we find epistemological randomness in nature – that is, where we fail to find laws to explain events – science has hit its epistemological limits. As Christians we ought, instead, to be saying that wherever we find contingency in evolution – whether in mutation or selection or both – we are witnessing God’s free choices. And wherever we see lawlike processes, we are seeing God’s faithful consistency. That is true for both evolution and the weather, for God’s providence is sovereign over both.

All the stuff about whether evolution is random or not is totally irrelevant – but we ought to be pointing out in no uncertain terms that the pretensions of non-believing (or even believing) scientists that it is relevant are based on the confusion between the scientific definition of randomness, which is wholly epistemological, and an ontological definition of randomness as “having no definite cause”, which is nothing to do with science anyway.

Repeat after me:
(A) Scientific laws are God’s regular actions in nature.
(B) Contingencies are God’s irregular actions in nature.
(C) “Random” in science, is purely about epistemological limitations, of whatever kind.

None of this denies, nor conversely requires, secondary causes, whose powers are a metaphysical matter, but one taken as a working assumption in science. Science has plenty to say about (A), but can only observe and record (B). It should have its wrist soundly slapped every time it forgets (C).