The Blind Watchmaker
by Richard Dawkins
Richard Dawkins starts his book by recognising that biological organisms are complex e.g. 'We animals are the most complicated things in the known universe' (p1) and, as such, they ‘cry out for an explanation’ (p xiii) for their existence. I agree.
Because of their complexity and evident fit for purpose they appear to be designed. But, says Dawkins, we shouldn’t be persuaded by just this subjective perception of design – which he calls the ‘Argument from Personal Incredulity’ (p38) – into concluding that they actually have been designed. I also agree with him on this.
Dawkins’ proposed solution is twofold:
- break down the complexity into small manageable steps i.e. steps that are small enough to have a reasonable chance of occurring (e.g. by random mutation); and
- have a mechanism that can identify when a small step is advantageous, and retain it so that it can be used as a stepping-stone to further improvement; which of course is natural selection.
This is a recurring theme throughout his book, summed up towards the end as follows:
To 'tame' chance means to break down the very improbable into less improbable small components arranged in a series. No matter how improbable it is that an X could have arisen from a Y in a single step, it is always possible to conceive of a series of infinitesimally graded intermediates between them. However improbable a large-scale change may be, smaller changes are less improbable. And provided we postulate a sufficiently large series of sufficiently finely graded intermediates, we shall be able to derive anything from anything else, without invoking astronomical improbabilities. We are allowed to do this only if there has been sufficient time to fit all the intermediates in. And also only if there is a mechanism for guiding each step in some particular direction, otherwise the sequence of steps will career off on an endless random walk.
It is the contention of the Darwinian world-view that both these provisos are met, and that slow, gradual, cumulative Natural Selection is the ultimate explanation for our existence. (p317)
In this passage Dawkins identifies two provisos or conditions for the success of his solution:
- a guiding mechanism, and
- enough time;
and I shall comment on these below.
However, there is a more fundamental obstacle to his proposed solution:
Steps cannot be infinitesimally small – Genes limit how small a step can be
In the above passage, Dawkins says “no matter how improbable it is that an X could have arisen from a Y in a single step, it is always possible to conceive of a series of infinitesimally graded intermediates between them” (my emphasis). And he expands on this in his discussion about the evolution of eyes:
Is there a continuous series of Xs connecting the modern human eye to a state with no eye at all?
It seems to me clear that the answer has to be yes, provided only that we allow ourselves a sufficiently large series of Xs. You might feel that 1000 Xs is ample, but if you need more steps to make the total transition plausible in your mind, simply allow yourself to assume 10,000 Xs. And if 10,000 is not enough allow yourself 100,000, and so on. (p78)
And he even envisages a much larger number of intermediates:
Given, say, a hundred million Xs, we should be able to construct a plausible series of tiny gradations linking a human eye to just about anything! (p78, my emphasis).
And, not only is it possible to conceive of all the intermediates he needs, but his ‘feeling’ is that they all will be available and functional when required.
My feeling is that, provided the difference between neighbouring intermediates in our series leading to the eye is sufficiently small, the necessary mutations are almost bound to be forthcoming. (p79 emphasis in original.)
Being able to divide a large transition into exceedingly small steps is foundational to his thesis for how an unlikely end-product could have evolved naturally from a simple starting point. Unfortunately, because it seems clear to him that "the answer has to be Yes", he does not examine whether or not this premiss is actually true. It’s a pity that he didn’t subject his feelings to a bit of objective scientific scrutiny. If he had, he might have realised that this premiss on which his whole thesis is built – of infinitesimally graded intermediates – is false.
His premiss fails at the level of molecular biology.
First, genes are discrete entities. Dawkins may well argue that half an eye or a rudimentary wing is better than none; but this certainly doesn’t apply to genes. Half (or even much more than half) a gene does not work at all.
Second, genes cannot arise through a series of small changes. As mentioned in obstacles to new genes, not only is there a minimum size for a protein of about 70 amino acids, but most of the sequence has to be right in order for it to work at all. Taken together, these criteria defy a progressive evolutionary origin of proteins from a short or non-specific amino acid sequence.
In other words, genes set a lower limit to how small biological progress can be divided up. That would not necessarily be an obstacle if genes themselves could arise readily or by small degrees; but they can’t.
Dawkins’ model for the evolution of genes
So this leads me to look at Dawkins’ illustration for how he imagines genes could have evolved (p45-50). It’s well known, but here’s a summary.
He compares the amino acid sequence of a protein (such as haemoglobin) with a sequence of letters in a phrase: he selects ‘METHINKS IT IS LIKE A WEASEL’ (28 characters, including spaces) and aims to find this by random typing. Of course, it is practically impossible that the whole phrase would ever be typed at random (in one go).
But he envisages that a random sequence of 28 characters is typed and there is some way of (i) recognising whether any of the letters are correct, and (ii) if so, fixing them so that they do not change. The sequence is then selectively copied – keeping the right letters fixed, but at each of the remaining places another character is selected at random. If any of these are ’right’ then these too are recognised and fixed, and the selective copying repeated. Not surprisingly, with this sort of process it does not take many rounds of selective copying (generally a few 10s) to end up with the right sequence.
As a general illustration of how cumulative selection might work it is acceptable. But as a model, or even an indication, of how (genes for) proteins might have evolved – which is clearly his intention – it is totally fallacious and misleading.
- Most obviously, although he says repeatedly that natural selection has no foresight, the mechanism he is using here relies on foresight; which he recognises (p50) but still seems to think it is a valid illustration of cumulative progress guided by natural selection.
- Perhaps he resorts to using foresight because he is only too aware that natural selection would be incapable of guiding the evolution of a gene. For natural selection to be applicable to his model it would be necessary to assume that even having just one amino acid right would be advantageous (why else would it be recognised as correct?). Whereas we know that the vast majority of the sequence has to be right before there is any function at all. But, of course, getting a large part of a sequence right presents the sort of daunting improbability that he is trying to circumvent. Unfortunately, rather than recognising this plain biochemical fact, he presents a model based on foresight, perhaps hoping that his readers won’t notice the deception.
- Which leads to another criticism of his model – that, as already indicated, the sequence he uses is much too short to represent evolution of a protein; because for that there is a minimum length of about 70 amino acids (see proteins need to fold).
- Finally, his model completely ignores the essential prerequisite that the length of DNA (where this supposed evolution is taking place) needs to be recognised as a gene (see obstacles to new genes); which requires control sequence(s), which again are all or nothing in order to work.
Natural selection can favour only what IS useful (to the organism there and then), not what is potentially useful. So there is no way ‘right’ letters can be recognised by comparison with a goal (as Dawkins does); they can be identified and favoured only if they confer an actual benefit.
Indeed, the very idea of aiming for a gene/protein with a particular function (such as haemoglobin) is totally contrived. The organism has no way of 'knowing' that the target gene/protein will be useful.
So, for several reasons, his notion of being able to break down evolutionary progress into steps as small as he wishes breaks down at the level of genes.
The implications — and will Dawkins be true to his word?
The fact that genes cannot arise in a gradual evolutionary way undermines and invalidates all evolutionary scenarios that consider only the level of morphology.
In chapter 4 ‘Making tracks through animal space’ Dawkins shows – at least to his satisfaction – how eyes, lungs, wings and ears might have evolved via part-formed structures, each of which offered at least some advantage over its predecessor that natural selection could favour and hence guide the overall process. He concludes:
It is thoroughly believable that every organ or apparatus that we actually see is the product of a smooth trajectory through animal space, a trajectory in which every intermediate stage assisted survival and reproduction. (p90-1)
The trouble is, his scenarios are entirely limited to a description and evaluation at the morphological level, and fail to take on board that (many) new genes must be required in the course of evolving such organs.
(This is a common error in putative evolutionary scenarios, not only by Dawkins - see Neo-Darwinism. Because existing genes allow a virtually continuous variation in many characteristics, it has lulled biologists into thinking that the continuum they observe can be extrapolated indefinitely. To the extent that, although they know that ultimately new genes would be required, there seems to be a tacit assumption (as Dawkins makes here) that new genes will arise as required and their action blend seamlessly into that of genes that were already available. Perhaps this belief arose from the days of the ‘synthetic theory’ of evolution – see the quotation from Fisher – but is no longer tenable.)
Continuing from the preceding quotation, Dawkins writes:
Wherever we see an X in a real live animal, where X is some organ too complex to have arisen by chance in a single step, then according to the theory of evolution by natural selection it must be the case that a fraction of an X is better than no X at all, and two fractions of an X must be better than one, and a whole X must be better than nine-tenths of an X. I have no trouble at all in accepting that these statements are true of eyes … and all other examples trotted out in anti-evolution propaganda. (p91)
Presumably Dawkins thinks he’s on safe ground. But the truth of the matter is that where X is a biological macromolecule, such as a protein, then it is not true that a fraction of this is better than no X at all; because a fraction of a protein will be completely useless (and arguably worse than no X at all because of the waste in producing a useless protein). Hence, based on his statement that ‘according to the theory of evolution by natural selection it must be the case that a fraction of an X is better than no X at all’ it should be concluded that evolution by natural selection is not true, at least for macromolecules such as proteins.
And let’s take it a bit further. He then quotes Darwin:
If it could be demonstrated that any complex organ existed which could not possibly have been formed by numerous, successive, slight modifications, my theory would absolutely break down. (p91 )
It is understandable that Darwin’s comprehension of organs and their possible evolution was limited to morphology, because in the mid 19th century there was no knowledge of molecular biology. But today’s biologists have no such excuse.
Nevertheless, full of confidence in (the truth of) his solution of cumulative selection to the problem of complexity in biology, Dawkins boldly asserts:
I do not believe that such a case will ever be found. If it is – and it'll have to be a really complex organ, and, as we'll see in later chapters, you have to be sophisticated by what you mean by 'slight' – I shall cease to believe in Darwinism. (p91, original emphasis)
Well, the truth of the matter is that every organ fails this evolutionary test. No organ could arise by numerous, successive, slight modifications, if for no other reason than every organ is absolutely dependent on proteins, none of which could have arisen in this way.
Further, the more we find out about the extraordinary complexity of molecular machinery within cells, such as the cytoskeleton (see box), it is increasingly evident that neither could cells have arisen by a series of slight modifications.
Dawkins is right: organs are ‘really complex’ – at the cellular and molecular levels – far too complex to have arisen in an opportunistic evolutionary way. So, in the light of what we now know about biology, the challenge to Dawkins is – will you be true to your word and abandon your faith in Darwinism?
Now, as promised above, I shall comment briefly on the two provisos that Dawkins identified.
Throughout his book Dawkins is anxious that his readers fully understand that natural selection:
- is not random (unlike mutations);
- that it has no foresight, but only favours variations that are of immediate survival/reproductive advantage; and
- that through repeated application of (ii) it has the power to guide significant overall change through a series of small changes.
I completely agree with him on these.
The trouble is, he makes the (all-too-common) error of thinking that, because natural selection definitely occurs, this means the whole of the evolutionary account must be true.
Examples of his invalid extrapolation from the role of selection (from existing genes) to large scale evolution (which would require new genes) are when he compares the time taken for industrial melanism to spread through populations of the peppered moth, and for the domestic breeding of dogs, with the evolution of eyes and echolocation (p40).
And the examples he gives of the supposed power of selection fall into two categories:
- Guiding the evolution of organs such as eyes, wings, ears etc. – via his hypothetical part-formed intermediates.
- Guiding the ‘evolution’ of genes, and his biomorphs – both of which rely on foresight – so they do not illustrate natural selection at all.
His biomorphs present an interesting comparison with the real world.
His biomorph computer programs (p51-74) are intended to illustrate embryonic development. At bottom there are several routines which he calls genes, each of which affects a particular aspect of the resulting biomorph (such as degree of branching or length of branch), their particular action being determined by the value of a number that is associated with each routine. Dawkins allows the program to ‘mutate’ by which he means that the value associated with each ‘gene’ is allowed to change – and this affects the shape of the resulting biomorph. After allowing various ‘mutations’ he inspects the resulting biomorphs, selects the one that looks like it’s heading in the right direction for his intended objective, and uses this one for another round of ‘mutation’. And so on.
What I found particularly noteworthy about this exercise was the way in which mutations were allowed to change only the value associated with each routine.
On one hand, even allowing only this degree of freedom permits an extremely large number of variations and seemingly endless variety in the resulting shapes.
On the other, he could not allow random mutations to the code of his program because he knows that that would not have resulted in additional variation and potential for new shapes, but almost certainly in the program simply not working at all. But that’s the level of mutation that would be required for the evolution of new forms.
So it seems to me that this is a useful illustration of embryonic development. Different alleles (gene variations) result in change to the resulting adult (which is why we get morphological variations), but radical changes to the genes directing embryological development simply result in deformities if not fatal.And this underlines (i) the extreme difficulty of trying to evolve actual new genes (you don't get new functions by making random changes to existing code), and (ii) that embryonic development cannot be tinkered with blindly, so (iii) that where apparently ‘homologous’ structures are formed by different embryological routes, this contradicts that apparent homology, and shows they did not derive from a common ancestor. (See Homology - evidence for or against evolution?)
One final point on this. Dawkins writes:
… the ancestor is a tiny dot. Although the ancestor's body is a dot, like a bacterium in the primeval slime, hidden inside it is the potential for branching in exactly the pattern of the central tree of Figure 3. (p57)
If he had compared the dot of his first biomorph with a fertilised egg this would have been reasonable: The dot has within it the potential – by expressing the functions in his computer program – of developing into the various biomorph shapes; and this can be compared with the embryonic development through expressing genetic routines. But to imply that the development of his dot compares with the evolution of a bacterium into the tree of life – which would require innumerable new genes – is totally false and misleading.
Dawkins’ models fail not so much for lack of time but because they are fundamentally flawed. It is not possible to break down evolutionary progress into the infinitesimal steps that his thesis requires, so any theoretical time required to proceed through a series of such fictitious steps is irrelevant. Nevertheless it is worth noting that he shows a remarkable lack of understanding about the significance of time when it comes to evolution.
Dawkins says that ‘we have no intuitive grasp of the immensities of time available for evolutionary change' (p39). He may be right about that, but his intuition is to exaggerate the significance of the time available, for example:
The large numbers proverbially furnished by astronomy, and the large timespans characteristic of geology, combine to turn topsy-turvy our everyday estimates of what is expected and what is miraculous.
Actually, they don’t. A little objectivity shows that even taking into account the whole of the universe and it’s lifetime hardly makes a dent in the daunting odds of trying to find specific biological proteins (see The odds against new proteins). (In The God Delusion he suggests life might have arisen on one in a billion planets ; but there is no justification for this figure; so presumably it’s just his intuition.)
In the light of which, Dawkins’ approach in the Watchmaker is right: for evolution to have any hope of working then it’s necessary to break down the astronomical odds stacked against it. The trouble (for Dawkins) is that this approach doesn’t work.
Of more concern, coming from a professional biologist writing about evolution, is his apparent complete ignorance of population genetics. Repeatedly he assumes that an advance can be made good every generation e.g. in his model of gene evolution, his biomorphs, and his description for the evolution of the eye. But this is naively unrealistic.
He makes several false assumptions:
- That a favourable variation will arise in every generation – he gives no consideration to actual mutation rate.
- That the first time a favourable variation arises it is automatically identified and retained (fixed) by the population; whereas in reality there is only a very small chance of this (see The odds against new proteins).
- That a favourable variation is assimilated immediately i.e. spreads in a single generation throughout (at least a large proportion of) the population so that it can act immediately as a stepping-stone to further improvement; whereas in fact it takes many (at least 100s of) generations to spread through a population sufficiently for this to happen (again, see The odds against new proteins).
This is particularly relevant to his scenarios where he envisages a very long series of very small changes, most of which would necessarily confer only a very small advantage over its predecessor. Bearing in mind that the % chance of fixation is only twice the % improvement in survival/reproduction, it must mean that the vast majority of his intermediates would need to recur independently very many times (at least 10s, probably 100s) in order to have an overall reasonable chance of being fixed.
In fact, with the large number of finely-graded intermediates he envisages, the selective advantage offered by many/most of them would be so small as to approach that of a neutral variation, for which the chance of fixation is only 1/2N (where N is the effective population size), so each of these steps would need to recur innumerable times before they had a good chance of becoming fixed.
The time taken for a variation to spread through the population depends on its selective advantage – the smaller the advantage, the longer the time. So, again, a consequence of the very small selective advantages between his proposed finely-graded intermediates is that it will take much longer for the variations to spread through the population.
Was he really ignorant of these issues when he wrote Watchmaker; or did he just choose to ignore inconvenient facts (and rely on the ignorance on this subject of many of his readers)?
As indicated above, Dawkins' omissions in this area are completely overshadowed by the fundamental flaw in his thesis. Nevertheless, it means that even if his innumerable intermediates were possible, his scenarios would probably still fail for lack of time. And his total disregard for the realities of population genetics further demonstrates that his whole approach is not so much based on science, but on wishful thinking.
Notes display in the main text when the cursor is on the Note number.
1. Richard Dawkins, The Blind Watchmaker with Appendix, Penguin Books, 1991; first published 1986.
2. Here X stands for an intermediate rather than a destination as in the preceding quote.
3. Ronald Fisher, The Genetical Theory of Natural Selection, Revised edition (1958), Dove Publications Inc. (First published 1929), p103. Note that I am not suggesting Fisher did not appreciate the difference between evolution due solely to segregating existing genes and that requiring new genes. He was writing at a time when little was known about the molecular nature of genes or biochemistry.
4. Charles Darwin, On the origin of Species, Penguin Classics (1968), p219; (first published 1859).
5. See Wikipedia article 'Cytoskeleton'.
6. Dawkins, The God Delusion, Bantam (2006), p138.