Skip to content

Repost: What actually *is* design?

Last updated on 18 Sep 2017

This is a repost from my old site, with corrections. I will be back properly online next week, I hope.

I’ve been wondering of late what it is that is explained when something is called “designed”.

The older design theorists had no such trouble – Aquinas, for example, noted that

… things which lack intelligence, such as natural bodies, act for an end, and this is evident from their acting always, or nearly always, in the same way, so as to obtain the best result. Hence it is plain that not fortuitously, but designedly, do they achieve their end. [Summa, 1Q2.iii]

This is a wonder of spare and elegant prose, as always with Aquinas. Modern readers may miss the “for an end” and the meaning of “fortuitously”. Something has an end if it is for that reason it does whatever it does. And something is fortuitous if it does not happen by necessity. These are old, Aristotelian, categories of metaphysics. It is my opinion that they were developed by Aristotle and his successors as an attempt at science, not grand philosophy. You can explain things with an end/efficient/material/formal cause distinction. And so Aristotle and crew explained their world by positing that they had natures which included an end, and which had necessary properties referred to as “essences”, and “accidental” properties which were fortuitous and unnecessary.

Anyway, much of the … err… necessity for these categoricals is gone from science. We can have things develop from egg to adult without “ends”. In effect we explain this mostly in what Aristotle would call “material” (the properties of the matter from which things are made) and “efficient” (the actual motive forces) causes. Ends are not required in the paradigmatic cases of living things.

Aquinas continued, presenting the locus classicus of the Argument from Design:

Now, whatever lacks intelligence cannot move towards an end, unless it be directed by some being endowed with knowledge and intelligence; as the arrow is shot to its mark by the archer. Therefore some intelligent being exists by whom all natural things are directed to their end; and this being we call God.

Once we lose the need for an ends-metaphysics, a teleology or entelechy (that is, a general external view of ends being imposed, or a specific view of ends being innate, respectively), we lose the force of this argument, but it is a pretty good attempt at the time to explain all the physical and in particular biological phenomena we see around us. [Incidentally, Aristotle restricted ends to living things; his successors were not so spare.]

But we no longer need it. We no longer need it in the matter of development, which is explained now as the outworking of the physical properties of the constituent elements of living organisms in a given environment (internal and external always need to be included here – chicken eggs don’t develop well in space). And we no longer need it in the matter of how organisms came to be as they are – natural selection and the rest of evolutionary theory obviates that. So Aquinas quick and easy conclusion carries no formal force to us – there is an alternative.

But the modern intelligent design-theorists, the IDevotees, as I have called them, are not arguing to the existence of a designer, a point noted by that somewhat notorious philosopher Anthony Flew in his book Darwinism. Now they are arguing from a designer to an explanation of the properties of living things. Somehow, “design” is an explanation of why bacteria have flagella, why we have hemoglobin, and so on. So what is “design” that it explains anything?

The magical designer theory

Let’s ask two questions here. One: What is design? Two: What is an explanation, such that design explains something?

To the first: when some person designs an object, they typically have several features as designers. They have a plan or intention, they have the competence to make that plan work, and they have the tools to implement the plan.

Plans don’t generally pop up out of thin air. The most inspired of people have usually, if not always, spent years learning and pondering, trying out variations, and so on. As Pasteur said, fortune favours the prepared mind. Artists train for years in older styles and techniques before they can make the break and move to novel ones of their own devising. Even what looks like random splatter, as in Jackson Pollock’s Blue Poles, is the end result of years of experimentation. Plans get developed by inheritance of prior technique and goals, and by trial and error.

So, too, the competence. Anyone who trained in any art or skill knows that you don’t do it easily at first, or well, but that you need to practice, identify your mistakes (or more likely have them identified for you by one who has already achieved proficiency), and correct them. Competence is the outcome of trial and error.

Tools and techniques in the manufacture (the word once meant “the doing of the hands”) of designed objects are themselves the end result of a long period of cultural and technological development. One cannot make, say, a Toledo blade sword, without the knowledge of what materials to use, how to assemble and treat them, and the tools to cast and sharpen them. This is itself the end result of a process of, you guessed, trial and error, at the level of the tradition, guild or community.

So ordinary design, the design we do know something about, and which features in our explanations of artefacts, is the end result of trial and error processes, which the American psychologist Donald Campbell called “variation and selective retention”. He did this in direct parallel to a process of natural selection, and he added one more adjective – “blind”. The variation cannot, ex hypothesi, be directed, because we do not know ahead of time what will work. So, we work blind, applying what we used before and making, deliberately or not, small changes to our techniques which may, or may not, work out. We make these variations in many ways. We try them out in our mind’s eye, and reject those that seem wrong to us (based on the past). Sometimes this is not useful. We may discard approaches that could have worked very well indeed. But we will never know this, and the “space” of all possible variations is so vast that we just cannot try them all out.

Then we try the seemingly viable variations out in practice. Often we will start, and stop when it seems to be going wrong. This may also mean we miss out on good tricks. Again, the space of all variations is vast. We could also waste our lives trying everything. Then, if they work in a small scale, we will attempt to see if they work well in the wild, so to speak. If people find them useful swords or pleasing paintings. Then, we have advanced our tradition beyond where it once was. Thus is progress made.

Could it be that genius can sidestep this trial and error? I doubt it. Even the greatest of geniuses have, in their fields, studied. They may have a more creative mind, or be especially well-adapted to making bold moves that turn out to work, but before the event, they are indistinguishable from the many creative and bold folk who will eventually fail.

So I am going to make a declaration here; a principle of design, if you will. Call it Wilkins’ Design Desideratum: any successful design is the outcome of past experience and trials.

Design and explanation

Let’s think for a second about explaining how things happen or why.

When we give an explanation of some thing – philosophers refer to the thing to be explained as the explanandum – what is it we are doing? There are a number of accounts in play, but in science, I believe the deductive-nomological (DN) model or “covering law” model and its variants is the best account. Another account is the causal mechanism account in which something is explained if a causal mechanism is offered that is sufficient to cause the explanandum – I have no major objections to this, but I think it can be accommodated in the DN view. A third, as given in the above link, is the “unificationist” account, which is pretty well a “winner takes all” view that owes a lot to evolutionary epistemology, so let’s leave it to one side for fear of question-begging.

Under the ND model, explanation is given when something like the following is successfully offered:

M. Model, law or generalisation
C. Initial and boundary conditions
E. Thing (phenomenon or process) to be explained

The “sum” here is a relationship whereby M + C make E likely, or determinate. If M, C and E are all statements representing the states of affairs, then the outcome is to make E true or very likely to be true. It is usually put, since explanation is an act of knowledge, as an inferential relation. Hempel’s original version had it as a deductive relationship, but probabilistic and statistical accounts were very soon forthcoming, from him and others.

Okay, enough arcane stuff. Let’s consider how we might explain the existence in bats of a membrane (the patagium) between their phalanges (finger bones) that enable them to fly. Evolutionary explanations would run like this:

M. In environment E, if Trait T is fitness-enhancing, T will become fixed or widespread at an equilibrium in the population P
C. [Bat ancestors were in E, had membraneous inter-phalangeal T and T was fitness-enhancing in P]
E. Modern bats have T

Notice something here – the conditions C are not available to us. We do not know what conditions bats were in, and we do not know whether in fact having membranes between the phalanges increased fitness. This is something we infer. In fact, it is something we predict. If we ever find a basal bat fossil, we expect it will have partial membranes and will be, on one account anyway, an arboreal creature, like sugarglider possums and flying squirrels are today.

An explanation allows us to make predictions. If the model M has exceptions, then we are able to refine or extend the model to enhance our knowledge and explanatory ability. This process is called scientific progress (or “learning”). For example, if we find that there are animals that routinely have a fitness-enhancing trait that do not end up at fixation or equilibrium.

Now, how would we explain an E with a design model? Let’s take two cases – one where we know that a thing was designed, but not how (the Toledo steel used in the blades), and one in which we know how a thing might be designed, but not that it as.

Design: A tale of two knives

Let me present you with two objects. One is a Toledo blade, made from Wootz steel (which, it turns out, was actually made in south India by families who guarded the secret of its manufacture. Even today we know its composition but not the way it was manufactured, exactly, although there are enthusiasts who are reinventing techniques, possibly not the same ones, to make equivalent steel. Neal Stephenson’s Baroque Cycle trilogy is based partly on this).

The other is a stone knife, or seems to be, made from chert around 6,000 years ago.

What can we infer about the objects? Well, it may seem obvious that they both are made by design or intent. But it is possible that a suitably primitive stone knife is indistinguishable from an accidental flake of chert caused by another falling rock hitting a cliff face. There is no seemingly plausible account of the Toledo sword being made by unintentional processes.

Why can we reliably infer design in the case of the sword and not the stone knife? Let us consider some reasons:

1. We have a historical record of Toledo blades, but not of stone knives. Well, yes, this is true, but if we came across a Toledo blade for the first time, with no prior record or knowledge of them, Damascene steel, or even human warfare, we’d still likely conclude it was made by technologically sophisticated manufacturers, and that it was designed to do its job – either by the sword maker or by a tradition of sword making to which the maker was heir. It doesn’t matter if the designer is a single individual or a tradition here.

2. The complexity of the structure is not producible by any known natural (i.e., non-artificial) process, while it is known to be the sort of thing humans make. Complexity alone is not sufficient as an indicator of design, despite the assertions of the IDevotees. Many natural things are complex (in particular, the microstructure of minerals and crystals). But as we uncover more about the natural world, we get an increasingly clearer idea of what the natural world can produce, and very surprising it is too. The usual argument for ID is based on an argument from incredulity, as Dawkins calls it, or a failure of imagination.

3. We know the function of swords in human culture. We also know the function of stones with sharp edges, but unless there is some sort of complexity (flaking patterns on the edges) that could only feasibly be produced by humans, we can’t tell if it is a natural or an artificial product.

Are you starting to see a pattern here? We can tell if it is designed only if we know the sorts (classes) of things that designers produce, and the classes of things produced in the natural world without design, and the instance in question is in one but not the other.

Doing without design

We now move from the boring part – intelligent design and the various other attempts to revivify the argument from design. Let us instead look at design in evolution itself.

It is commonly said that design, in the guise of adaptation, is a core element of evolutionary theory. I want to make a somewhat radical suggestion – let’s lose all mention of it unless we are, as Bacon allowed, talking about human motivations. There is no design in biology. Adaptation is not producing “design-like” or “designoid” objects. There is only ever biological design in a single place – between the ears of observers of biology, whether they are specialists or naive.

Design is context-sensitive to the interests of the designer, not the interests of the observer. We project ourselves onto the natural world all the time – we’ve done this since the earliest recorded times. Plato did it. Aristotle did it less (and the Neo-Platonists Plotinus and Porphyry did it more). But that doesn’t mean the world is under any obligation to behave the way we think, or like us. Anthropomorphism is a Bad Thing.

Thinkers, like Dawkins or Dennett, who say that evolution produces “designoids”, or “information”, or “function”, or “purpose”, take ascriptions from the human (“intentional”) domain and apply them to the non-human, and non-cognitive domain. We say that a screwdriver has a function because we use it for one. The function of a screwdriver for a manufacturer or retailer is to increase profitability. But the function of an aspect of a living organism is, simply put, whatever the model used to describe and explain it ascribes. When you look at the number of models that can apply to, say, a structural element of a cell like an actin filament or an mRNAse, each one ascribes a function, a purpose or a “design”.

Design and its cognate terms are projections. The question is not whether they are projections, but in what circumstances it increases our knowledge of the things we ascribe them to. In other words, when is the function of a model properly applied to the thing modelled?

The answer is a purely epistemic one: when the model successfully describes the thing. Actin filaments have a role in the development of, say, William’s Syndrome. So the “function” of actin has a role in the normal development of glial cells and neurons. But actin also is a key element in the structural integrity of single celled bacteria. It has a “function” in maintaining and changing shape. It also plays a role in muscle contraction, and cell motility, and so on. Each of these is, in the appropriate (that is, the applicable) context, part of its function. But although it might play a role in sharpening pencils, and in the relevant model it would have that “function”, it actually doesn’t.*

Designs are abstractions based on how we model things. Functions and goals are likewise. Information is a property of the way a physical structure (usually the primary sequence of a polymer) is symbolically described – you get a different information content when you symbolically describe the primary sequence of a stretch of DNA as “GTAC” than if you described the hydrogen bonds or the numbers of protons or the energy shells of the atoms. It suits us to use the nucleotide abbreviations; but it may also mislead us. Sometimes the weak and strong bonding points on a primary sequence are more important than the sequence itself, biologically speaking, for these will affect folding, expression, and error-correction (this last being an abstract way to speak of mismatch repair).

So I want to go the whole deflationary hog. Let’s stop talking about these abstract properties of biology, and just talk about the biology. What shall we miss? Nothing, since function and goal is only a metaphorical way to express actual physical properties and quantities. Of course, we may continue to use these terms on the understanding that they “stand for” real things, and that the metaphors are only for convenience. Would that it were that simple. We intentional animals keep falling into the trap of mistaking the periphrases for the facts.

So I want not just a desert ontology, as Quine proposed, but a rigidly physical one, in biology. Evolution is not a process of producing information, goals, or functions. It is a process of physical systems that make more or less accurate copies of themselves in environments of more or less stable thermodynamic and spatial properties. There’s nothing that can be called “information” that doesn’t involve, in my opinion, a clear analogue to a Shannon system of sender-channel-receiver. Perhaps cell–cell signalling qualifies. Certainly, neural systems do. Genes? Not on your, or any other organism’s, life.

Think of the confusion that would be avoided. Think of how we would be able to see clearly that we aren’t talking about computers, programs, goals, intentions, purposes or plans. Just living things…

Bet nobody agrees. [Late note: Nobody does. Well some did at the talk I gave in July 2006, but they were not the main respondents. I’m not surprised. What did surprise me was that the respondents took physical reduction to mean eliminationism, for instance, it was taken to imply that I thought that there were no information-processing systems in the world, which is absurd, or that humans have no intentions, which may be absurd with some further argument.]

Explaining Rosemary’s Garden Design

When we design something, we typically make use of prior knowledge and techniques, and of the materials we have used or which hold promise for us on that basis. So, when we explain ordinary design we do this by packing a lot of information into the general model of design processes.

If I explain why Rosemary has uprooted a garden bed and replanted it with native plants, even though hitherto they had always planted English roses, and had given no indication of why she had done this or that she was about to, my explanation taps into a large body of default and even tacit knowledge of gardening practices, all of which is summed up in the general premise “Gardeners design gardens” or something similar.

I do not need to have a fully elaborated model of what either Rosemary is likely to do – that is, I don’t need a complete or sophisticated model of Rosemary’s predispositions to behave – nor do I need to explain why Gardeners occasionally uproot English roses to plant Australian native plants. My explanation is sufficient, so far as it goes, if I note a general tendency of Australian gardeners to plant Australian native plants, so long as it is understood that there are in fact reasons for this (they are more drought-tolerant, require less weeding, and are in fashion because people are coming to appreciate the aesthetics of the Australian flora).

Such a general set of implicit rules makes the outcome “Rosemary replanted her garden with Australian natives” explicable, and indeed, more likely. Moreover, the explanation taps into that tacit knowledge base to assert something about Rosemary – that she is affected by that general set of considerations. To assert that Rosemary’s Garden is the product of design is to assert something about her community, culture and the traditions of gardening.

Now if Rosemary attacked her garden with a hoe, leaving it in a state we would call disarray, we can only know that this is, in fact, disarray in contrast to some kind of Garden-dadaism or Floral Cubism, by making reference to the traditions of design and the expected variations that would count as acceptably Gardening. Maybe Rosemary just had a bad day, or a fight with her Significant Other.

Explanations fall in what Alan Garfinkel once called a “contrast space” set up by the variables of the question the explanation seeks to address. “Is Rosemary’s Garden designed?” specifies a set of alternatives, and these together make a space of possible “yes/no” answers. Garfinkel illustrated this with the famous [and mythical] question asked of bank robber Willy Sutton: a priest asked him why he robbed banks. Sutton answered, “That’s where the money is”.

The priest was expecting a moral answer – Sutton robs banks because he needs the money or because he can’t get a job. But Sutton has no moral issue here; it’s purely a matter of practicalities. He robs banks rather than, say, druggists, because there’s more money there. Sutton has a different contrast space than the priest. Explanations are relative to the contrast space required.When I explain that Rosemary’s Garden is the product of design, there’s a whole set of contrasts I am relying upon.

But the way the Intelligent Design crowd employ both the notions of “design” (to be a rarified rather than an ordinary notion), and the notion of “explanation”, there is no actual contrast space here – all is fluid. All that matters for them is “saving the theology”, not “saving the phenomena”. So long as something like a God can be made to seem sensible, all is done that need be done. This is not science, and it’s not even compatible with science. It is, and remains, epistemological despair, epistemic nihilism, Knownothingness.

Contrast that with the traditional natural theology that preceded Darwin. It had an honest intent – to uncover the nature of the Deity from the nature of his works. Or contrast it with the reaction to evolution of English Catholicism. The leader of the Oxford Movement, Cardinal John Henry Newman noted that evolutionary biology was not opposed to God or creation, and wrote:

I have not insisted on the argument from design, because I am writing for the 19th century, by which, as represented by its philosophers, design is not admitted as proved. And to tell the truth, though I should not wish to preach on the subject, for 40 years I have been unable to see the logical force of the argument myself. I believe in design because I believe in God, not in a God because I believe in design. [Quoted in Ruse 2003, 72. This is a good, if simplified, introduction to the role design has played in Christian thought in the English speaking world, though I should like to have seem more of John Ray, the founder of natural theology, and of those who followed him in the 18th century.]

In Germany though, Catholics were less sanguine about evolution due to the aggressive evangelical materialism of Ernst Haeckel. Still, nobody thought that design proved God, or protected theism from science. Newman was happy for science to continue without reference to God. He saw that in science the contrast space of explanations did not include final causes, except, as Francis Bacon noted long ago, with reference to human behaviour:

It is a correct position that “true knowledge is knowledge by causes.” And causes again are not improperly distributed into four kinds: the material, the formal, the efficient, and the final. But of these the final cause rather corrupts than advances the sciences, except such as have to do with human action. [Francis Bacon, Novum Organon, aphorism 3]

Bacon knew that to assert that design, or teleology, applies to the natural world is a case of what Freud came to call projection – asserting that something natural is like human choice. I will thus finish here with a quotation from the leading critic of design talk in biology, Susan Oyama:

Powerful and protean and far from being banished from secular science, the argument from design is ubiquitous. Perhaps because we are creatures whose existence and survival depend on our ability to discern regularities in our surroundings and in turn leave our mark, our design, on them, we tend to infer prior design or intent from observed regularity. We formulate, that is, a descriptive rule, which is a form of knowledge, and infer from it a prescriptive rule, which is separate from the processes we see and controls them.

That’s enough about design. It’s really rather boring. Explanation is much more interesting, after all…

* I owe this complaint to my friend Clem Stanyon, who was working on these things at the time.


Forms of explanation: rethinking the questions in social theory. New Haven, Conn: Yale University Press.

Oyama, Susan. 1985. The ontogeny of information: developmental systems and evolution. Cambridge: Cambridge University Press.

Ruse, Michael. 2003. Darwin and design: does evolution have a purpose? Cambridge, MA: Harvard University Press.


  1. Michael Fisher Michael Fisher

    Hi John

    I’m stuck in the thicket of your long words & I would enjoy/appreciate a synopsis because I have the feeling that I have missed the point of your post: What actually *is* design?

    I think you’re saying the word design is best avoided (along with I suppose purpose, plan, intent, etc.) in relation to blind evolutionary processes.

    Is that it ?


    • John S. Wilkins John S. Wilkins

      The headline: design exists only in the heads of designers, and so far as we know they are all human (or Caledonian crows, etc.). Evolution has no design and no functions otherwise.

      • Brian Brian

        Are you suggesting that you’re smarter than Aquinas, Paley or Dembski? 😉

        • John S. Wilkins John S. Wilkins

          Not Aquinas or Paley, no. But I’m more right.

  2. Wow, really in depth.

    “Let’s stop talking about these abstract properties of biology, and just talk about the biology. ”

    I agree with this whole heartedly.

  3. Really excellent post. Long, but worth the effort.

  4. MKR MKR

    Of course, we may continue to use these terms on the understanding that they “stand for” real things, and that the metaphors are only for convenience. Would that it were that simple. We intentional animals keep falling into the trap of mistaking the periphrases for the facts.

    I take the last sentence to be alluding to the mistake of taking the so-described “design” characteristics of living things as evidence of a designer. But there is another, quite different mistake that fits the same description. Where the scientifically ill-informed may reason, “Living things show design, therefore they are caused by a designer,” many of the scientifically well-informed reason, “Living things show design, but this is no evidence of a designer; therefore, design does not imply a designer.” To be more precise, they use the vocabulary of design–such terms as not only ‘design’ but also ‘engineering’, ‘problem’, ‘solution’, ‘makeshift’, and so on–and in so doing claim to be speaking literally, while at the same time they deny that such talk implies a designer–or perhaps they would say, as if thereby making a distinction, “an intentional designer.” At least, this is a position that I seem to encounter repeatedly in discussions on the Web. It seems to me to show a confusion in the use of language. If the word ‘design’ is used without the implication of an intentional and intelligent agent then it is not being used literally. One cannot say, “Oh, I am using it literally–just not in that sense,” unless one has an explanation of this supposed other sense to offer; and I have not heard one.

    A final point about the use of words: I could not find the word ‘periphrase’ in my Merriam-Webster Dictionary, but I did find ‘periphrasis’, which is glossed there as “use of a longer phrasing in place of a possible shorter form of expression.” (It has the plural form ‘periphrases’, but the stress is on the second syllable and the ‘-es’ ending is pronounced ‘eez’ rather than ‘iz’.) Assuming that this is equivalent to what you mean, surely descriptions of structures of living things in intentional terms tend to be shorter, not longer, than descriptions in purely physical terms. So if either class of description is periphrastic with respect to the other, surely it is the physical, not the intentional one that is so.

Comments are closed.