Skip to content

God and Evolution 5: The problem of chance

Last updated on 24 May 2013

Many religious thinkers hold that chance is the enemy of God. God is omniscient in many theisms, and so if chance occurs, and chance is unpredictable even for God, then the reality of chance means that God does not exist. This doesn’t apply, of course, to gods that are limited in their knowledge. The Bible seems to imply that God is not omniscient in some passages (e.g., Genesis 3:9, “But the Lord God called to the man, “Where are you?”), but in many others the use of casting lots is seen as a way of determining God’s will.

The meaning of a word like “chance” along with similar words like “random”, “luck”, “contingent”, “accident” and “probable” are more complicated than ordinary use seems to imply. For now, we can distinguish several meanings:

  • Uncaused
  • Small likelihood
  • Unpredictable
  • Not necessary.

These meanings are often mixed into each other, so that a problem caused by one is taken to imply the others. If some event in the world is uncaused (such as a quantum particle appearing), it might in fact be of moderate likelihood and be both predictable to some extent and a necessary event given the laws of physics. Likewise an individual event might be of small likelihood and unpredictable, but be caused by those laws and so be (physically) necessary, and so on.

Evolution is not, in itself, a matter of physics, although most would agree that it happens on physical properties. So, of course, does everything else that science studies. The origins and nature of the universe are a problem for theists whether evolution happens or not, so we can leave these issues to the cosmologists and theologians, like Keith Ward, whose God, Chance and Necessity (1996) unfortunately conflates cosmology and evolution as theist issues. Nevertheless, chance and necessity are issues we must deal with.

Since Jacques Monod, the molecular biologist, published Chance and Necessity in 1972, the question of the role of chance in evolution has been widely debated. Monod, along with more recent authors like Dawkins, Hawking, and others, argued that the universe was a blind mix of chance and necessity with no purpose. Necessity is, in effect, the operation of natural law. Chance is its absence. However, the previous understanding of natural selection by evolutionary biologists is that natural selection operates on “blind chance”, but is in fact a reducer of chance. Natural selection is a probability raiser. Theodosius Dobzhansky was a leading evolutionary biologist and thinker in the middle of the 20th century, and also a Russian Orthodox believer. He wrote in his The Biology of Ultimate Concern (1967):

Natural selection is a chance process … only in the sense that most genotypes [particular genetic instances – JSW] have not absolute but only relative advantages or disadvantages compared to others. … Otherwise natural selection is an anti-chance agency. It makes adaptive sense out of the relative chaos of the countless combinations of mutant genes. And it does so without having a will, intention, or foresight. [60]

Mutations are chance, according to the consensus of biologists, but they merely provide the material on which natural selection operates, winnowing out those that are relatively less successful in a given population. What sort of chance is a mutation?

Mutations are clearly not uncaused. They are due to chemicals that interfere with the process of replication, that is, copying, the DNA of an organism when cells divide (often causing cancers in body cells). Or they are caused by occasional radiation, such as from cosmic rays emitted by collapsing stars or the background radiation of the earth itself. Put a Geiger counter to a house brick, and you will get a reading. Radiation is all around us. Mutations are the result of natural laws.

But what is chancy about mutations is that they cannot be predicted, and this is because they are “accidents”, which happen in ways nobody can determine ahead of time. Arguably, if we were all knowing we’d be able to predict not only when such events would occur, but exactly what mutations would occur. Since evolutionary biology deals with populations, though, and not individuals, it can only predict that mutations will occur with some frequency. Every individual carries a number of mutations. For mutations to genes to be passed on to progeny, they have to occur in the sex cells, the gametes, or else they will expire with the death of the organism. On average, a human being carries around 100–200 mutations.

Most mutations are not bad. Mostly they occur in what is called (slightly misleadingly) “junk DNA”, the vast majority of genes that have no function in the development of the organism. These just sit there generation to generation doing nothing much, being copies. Occasionally, though, the “stop codon”, or the full stop that sits between genes and junk or between genes that are functional, is itself deleted by a mutation event, and so this junk DNA gets “read”.

In this way these neutral mutations can persist for quite a while, and even spread through a population, before they become subject to selection.

Also, many mutations are “silent”, which means the change didn’t alter the product of the gene. Genes are formed from sequences of three “letters” or codons. Each codon “codes for” a particular protein part (an amino acid), but several codons can code for the same amino acid, so changing one base pair (the “letters” G, A, T, and C) may not affect the outcome.

Of those that are read off and produce proteins, often the difference is so low in functionality that the mutated gene makes no real change in the fitness of the organism. But natural selection, under the right conditions, can take even very small difference in relative fitness and drive a mutated gene to either a balance in the population (where an increase in the numbers of organisms that have the mutated gene would not lead to an increase in the overall fitness of the population) or to what biologists call “fixation”, where it takes over all alternatives at that place in the genome.

And some are just a lot fitter. This is another way of saying in that population, and in that environment, that mutated gene is an advantage to any organism that has it relative to the alternatives. In these cases, if the conditions are right, that gene will become fixed in the population.

Okay, so with all that behind us, what is the role of chance here? We already know that mutations are unpredictable. Are they accidents in the metaphysical sense? Are they undetermined or beyond God’s ability to predict them? If “undetermined” means uncaused, then no, they aren’t undetermined, but are they the result of physical processes God could not foresee, for example, quantum decay in atoms that cause radioactivity? Possibly. It depends on how you reconcile not just evolutionary biology, but the whole of modern physics with your theology. And such “unpredictable” events may in fact be predictable at scales we can’t access, but presumably God could. One physicist, Nobel winner Gerard t’Hooft in the Netherlands, has hypothesised that quantum events can be caused by deterministic processes smaller than our limit of resolution, the Planck Limit. If so, we could never predict them, but we could study, as we do, their outcomes using statistical methods. If this were so, then while quantum accidents are a problem for physicists, they presumably would not be for God.

But what if these accidents were unpredictable by God? What if chance is an irreducible fact of existence? This is a question for the theologians of your chosen faith community to resolve. For now I will make only a few comments.

The earliest Proto-Indoeuropean views of God had him or the gods in general striving against chaos. Chaos (KHAOS) was one of the first protogenoi or primeval gods in ancient Greek religion, which the later gods opposed to create order. In the earliest parts of the Hebrew Torah and Tanakh, the Christian Old Testament, God imposes order upon the “deep” (tihom). An earlier and possibly influential Babylonian myth, the Enûma Elish, has Tiamat, the sea god, as an agent of chaos opposed by Marduk who creates the universe as it now is. There are many such theogonies (origin stories of the gods and the world) in the Indo-European and Semitic traditions.

Chaos as now understood by modern science is less disturbing. It is founded on the idea that very slight differences in starting conditions can lead, quite deterministically, to very different outcomes. Imagine a flat pinball machine on which completely rigid balls bounce around against pegs. A minute difference in the starting angle can multiply each time it hits a round pin so that you almost cannot duplicate a sequence no matter how hard you try. Although each bounce is quite exactly the result of the angle of attack, these minor differences will cause different pins to be hit, so that the final point of the ball will be unpredictably different.

So chaotic behaviour is not necessarily the outcome of chance, but instead of these contingent differences acting in determined ways. This leads us to necessity. If physical laws are fully determined, then God can predict what to us look like, and are in practical operation, random events. Chance (as we understand it) and necessity need not be in opposition at this level. God’s necessity is our chance sometimes. It depends on how much knowledge, and how precisely we know it, that we have.

So in the end chance might boil down to our limitations. Of course if God is limited by chance, then all bets are off for the theist. But that is not raised by evolutionary biology. Instead it is a deep issue in theology, to be sorted in each tradition as the authorities and intellectuals of that tradition see fit.

Each tradition has its own ways of dealing with these issues, but a final point is worth making. In the so-called Abrahamic religions – Judaism, Christianity, Islam and their offshoots – chance is seen usually as the basis of atheism. In fact, the word for atheism in the Jewish Talmud is apikoros, a transliteration of the name of a Greek philosopher who came slightly after Plato, Epicurus. Epicurus is sometimes seen as a hedonistic atheist who challenges all religion, but the reality is somewhat more subtle. Although he certainly did not think that the gods desired our worship, he did not deny their existence. Instead, his deities were perfect, and so being perfect would only contemplate their own perfection. Consequently our belief or worship was of no consequence to them. This is the philosophical tradition in Greek thought at its limit.

Epicurus is also famous for his “Swerve” (clinamen in Lucretius’ Latin). In his theogony, Epicurus proposed that at the beginning, the atoms were sleeting through the void in parallel, and an uncaused event led to one particle (or more) hitting another, causing a chain reaction in which the atoms eventually combined at random to form the physical world. With a suitable reinterpretation in modern terms, this is not so unlike the theory of the big bang. What happens after the swerve, or big bang, unfolds according to necessity (that is, the laws of nature).

Thereafter theists began to attack Epicureanism as atheism, hence the Talmudic interpretation (although I doubt the rabbis had ever actually read Epicurus, any more than most of the later Christian theologians). Charles Hodge had applied this to Darwin in the 1870s and more recently so too have Benjamin Wiker and William Dembski  in their book Moral Darwinism: How We Became Hedonists (2002). This is very unfair on Epicurus. He was not a hedonist, but thought that since pleasure was all there was to life, it should be enjoyed, but in moderation, or else it became a lack of pleasure. Epicurus’ theory was written up as a poem by the Roman Lucretius in the first century, De Rerum Natura (On the nature of things), which was rediscovered during the Renaissance and which influenced the development of modern thought. Most of Epicurus’ own writings are lost.

The sense of indeterminacy in Epicurus’ thinking seems to have been one of unpredictability, not a lack of cause, but it is unclear.


  1. Theology aside, I find discussions of change in biological evolution to be tricky and unsatisfactory. It’s easy enough to say that using an informal notion of probability (your unpredictability), but how do you make a formal estimate? Has anyone done it, or made a proposal?

    • Not in general. As I understand it, such calculations involve all kinds of background assumptions that are not general and are a little bit subjective in their importance.

  2. Whoops! I meant “chance” not “change.” Do people have anything more or less exact in mind when they say that life is improbable?

  3. Food for thought:
    One definition of omniscience is knowledge of all reality.
    1. Given this definition of omniscience and traditional compatibilism, then God possesses knowledge of every specific detail about the future, which is called “exhaustive definite foreknowledge.”
    2. Given this definition of omniscience and a world with genuine contingencies, then God does not have exhaustive definite foreknowledge, which is called “open theism.”

  4. Excellent!

    You have articulated well my objections to Jerry Coyne’s naïve determinism. Besides making his advocacy of “science” and “reason” unintelligble, his assertion that our thinking is nothing but colliding sub-atomic particles sets up exactly the conditions that an omniscient god could use to both “guide” evolution and, deliberately or not, avoid our attention

  5. Very interesting reversal, snatching “everything happens for a reason” from the teleologists.

    I also like you way you play up the tension between epistemology and ontology here, by opposing chaos (the inability to discern causes) against chance (the actual absence of causes.) We don’t know why a uranium atom decays they way it does; each radiated alpha particle has no apparent individual cause (indeed, no conceivable cause, given that each atom is theoretically identical, but does not behave identically)–but stochastically the decay is lawful and predictable. Einstein and Bohm and others of course theorized “hidden” variables to explain the lawfulness of the decay, but my (admittedly modest) understanding is that the question is far from settled. (If the matter should become settled in the future, I wonder if we will be treated to some new realm of indeterminism, just as we were 100 years ago.)

    I’m inclined to think this is the true meaning of the creation myths you cite in this post. Chaos is the world pre-dating rational analysis–the dawn of the human. It is “deep” because the easiest things to explain are those closest at hand. Our rational perception is very linear–it can only really focus on one thing at a time. Despite our best efforts, chaos lurks around every dark corner. Its deification points to this eternal quality. (In Julian Jaynes’ great analogy, the rational mind is like a flashlight that that concludes no spot of darkness exists, because it can never locate one. Everything is shines upon is lit.)

    And if you find buddhist ontology at all compelling, chaos is the “real” world, because it can be beheld “all at once” (though not without difficulty), whereas rational analysis relies upon mediation through signs and abstractions. To the buddhists, chaos is sunyata, emptiness–not non-existence, but an absence of ideas about that existence. In this version of indeterminacy, the terms are reversed, so that instead of chaos reflecting our ignorance of causes and relations, rather it is the insight that the analyzed world is a schema of projected ideas.

    • A more sophisticated version of the drunk looking for his car keys under the lamp where the light is better. I like it.

    • Not a single Christian friend of mine disagrees with me when I say: “Everything happens for a reason. But sometimes the reason is that one is stubborn and makes bad decisions.” 🙂

  6. Roger Shrubber Roger Shrubber

    The discussion of mutation bothers me some, mostly for biochemical reasons but perhaps the correction helps with the nature of random chance. This is far too technical for your purposes but perhaps there’s something to be distilled from it.

    Mutations are clearly not uncaused. They are due to chemicals that interfere with the process of replication, that is, copying, the DNA of an organism when cells divide (often causing cancers in body cells). Or they are caused by occasional radiation, such as from cosmic rays emitted by collapsing stars or the background radiation of the earth itself.

    Firstly, the role of radiation is generally overstated. The primary cause of point mutations is reproduction fidelity or a lack thereof. A useful way to think of things is that the fundamental chemical reaction of an elongating DNA strand is the 3′ -OH group of the existing DNA strand forming a bond with the innermost phosphate group of the next deoxynucleotide triphosphate, and releasing pyrophosphate. The reaction with dATP, dTTP, dGTP or dCTP are all possible and all highly favorable reactions according to established thermodynamics. Outside of the context of synthesizing a complementary strand a simple “elongase” would randomly add dNTPs of all types based on whichever ones diffused on by and were held long enough to happen to react. If there was twice as much dATP as dTTP, the random polymer made would predictably have twice as much A as T but the precise sequence would be random. And it would be random because the diffision of dNTPs is a chaotic dance, somewhat like perpetual motion of billiard balls. And if there’s any indeterminacy anywhere, it would be predicted to have propagated it’s novelty to the molecular motion and altered the otherwise predictable if unfathomable molecular mosh pit of diffusion.

    The trick that DNA replicating enzymes have played to get the “correct” reaction to be favored is two-fold. The correct dNTP is capable of forming the well understood Watson-Crick hydrogen-bond pairing so that opens the door to bias which dNTPs react. Depending on specifics of an individual enzyme, the binding energy differences between correct and incorrectly paired dNTPs vary. The differences in binding energies changes the half-life with which the different types of dNTPs hang out in an enzymes active site. Again, this means you can enhance the chance of the correctly paired dNTP reacting. . But here’s an important lesson, the reaction does not happen immediately upon the right type showing up. Instead, hundreds of jostling molecular collisions have dNTPs dancing in and out of the correct binding site. The dNTPs are not queued up waiting for the master of ceremonies to tell them they look like the right sort of fellow, it’s their turn, get in their and give up your pyrophospate. No, it’s the chaos of billiard balls in an eternal dance diffusing about. And while the enzyme has biased things in favor of the right dNTP, the wrong sort diffuse into the active site as well and they still have the potential to react. This is partly because the differences in how well something binds to an enzymes does not establish an absolute clock on how long it will remain bound but instead establishes the average half-life with a distribution that itself is driven by that random molecular dance and so now and then even the wrong dNTP sticks around long enough to react and grow the DNA polymer.

    So DNA replication is only biased in favor of building a matched complementary strand, something we overlook because the bias is so very strong. But that bias from what is almost certainly non-deterministic molecular motion is imperfect and randomness leaks through.

Comments are closed.