You know, people think it must all be very easy, creating.
They think you just have to move on the face of the waters and
wave your hands a bit. It’s not like that at all. —Terry Pratchett, Eric
"Life looks designed" is a common refrain among a variety of creationists. The claim is intuitively appealing because we have experience with design. For most people, that is the only way they know for making a functional machine. Since design is the only explanation they can imagine, they naturally consider it the best explanation. To this extent, "looks designed" is just an argument from ignorance. But many creationists further claim that this appearance of design is objective, can be (and, some say, has been) demonstrated scientifically, and therefore is suitable for teaching in public schools (for example, Dembski 2001a). The little evidence they present, though, is maddeningly vague. In most cases, the supposed evidence for design consists simply of pointing to various examples from natural history and saying, "Look, can’t you see it?" Typically, this is accompanied by the usual creationist attacks on evolution and the claim, implicit or explicit, that design is the only alternative. Often there are vague analogies with human artifacts such as watches or writing, but never with objective standards of comparison. In design theory, "looks designed" has been left to the imagination of the believer.
When done properly, though, the "looks designed" method, or the method of analogy, is an effective method for detecting design. In fact, it is almost always how we recognize design in our daily lives. We learn through direct experience that some things are designed — by seeing the things made — or through testimony of the designers themselves. Most artifacts, though, we recognize as designed because they look like things that we already know are designed.
Analogy is used in science, especially in fields such as archaeology and forensics, to distinguish design from non-design. For example, archaeologists can tell whether a flint was broken deliberately or naturally because flints known to be worked by humans differ from naturally broken flints in features such as fracture angle (Cole and others 1978). SETI researchers, in searching for non-human design, use analogy by assuming human-like properties of extraterrestrials — namely, an interest in communicating and a desire to do so efficiently. And analogy is explicitly accepted, even promoted, by some creationists as a valid method of determining design (Moreland 1994; Thaxton 2001). Analogy to known design should be one way to detect design that evolutionists and creationists can agree upon.
Of course, the analogy method can only provide comparisons with designs produced by humans, since those are the only designs with which we have significant experience. Other design arguments suffer a fatal weakness: Without knowing anything about a designer, we cannot say anything about what to expect from one (Hume 1779; Sober 2003). Detecting a certain pattern does not indicate a designer until it can be demonstrated that the designer produces
such a pattern, and this task would seem to be impossible when dealing with potentially supernatural designers. By assuming at least some commonality between humans and the unknown designer, we can avoid that problem. The analogy argument, despite the weakness of its assumption of human-like designs, is one design argument which leads somewhere other than in circles. To use this method, though, we must first say what design looks like.
Determining what design looks like is no trivial matter. A communications satellite, a drainage ditch, OPEC, a mathematical proof, a jelly bean, false teeth, a limerick, the controlled burn of a forest, and shampoo have little in common, but all are designed. Probably no single criterion can ever describe them all. Still, design does have some properties that are fairly general. I examine some of these properties below and consider how they compare with what we see in life. I also consider other properties that creationists claim as indications of design. There are some similarities and some differences between life and design, but as we shall see, even the similarities argue against design as a scientific theory.
Probably the most obvious aspect of designed things is an intermediate level of structural order. Unfortunately, this sort of structure is difficult to characterize quantitatively, but its quality is apparent. Almost all designs have an arrangement that is neither very regular nor very random, but instead is between those extremes. There are exceptions, of course; a brick wall is highly ordered, and a stew is very disordered. Most designs, however, are neither uniform nor random, neither regular nor chaotic. Such an intermediate level of structure arises as a consequence of design. Objects that are too highly ordered are limited in their applications by their simplicity. Objects that are too chaotic are generally more expensive to produce, or their disorder keeps them from fitting and functioning well with other designs.
An intermediate level of structure plainly exists also in life. It is probably the most important characteristic people have in mind when they say that life looks designed. It is related to concepts of information, so it may have inspired some creationist arguments about information theory. Since there is no commonly accepted word for this property, and since it is hard to characterize, it is not surprising that creationist claims about design are vague and ill-formed. Despite the lack of rigorous description, though, we can be fairly confident that having an intermediate level of structure is an important quality shared by both design and life.
This is not enough to conclude that life looks designed, though, because an intermediate level of structure can arise naturally, too. Such structure can be found in molecules, cave formations, the Northern Lights, and Jupiter’s atmosphere, to give just a few examples. Structure arises spontaneously from a variety of processes; in fact, it takes only a couple of seconds for structure to appear in a candle flame. With regard to life, there is evidence that structure not only can
arise naturally from ordinary processes, but perhaps should be expected from it (Kauffman 1993; Adami and others 2000).
An underappreciated aspect of design is simplicity. Although many people associate design with complexity, almost all designs aim for maximum simplicity. (Complexity is another concept whose exact meaning is hard to pin down. As I use it here, greater complexity indicates that something is generally harder to understand; simplicity, of course, is the opposite.) Simplicity is important in design because simple designs are easier to invent, easier to implement, easier to modify, and usually easier to use. A good design is a simple design.
Of course, most designs require a certain amount of complexity. A home computer, for example, would not be able to do much if it consisted of nothing more than a solid block of silicon. (Although an advanced civilization could reputedly do a lot with a rectangular black obelisk.) It is in such seemingly complex designs, though, that the principle of simplicity is most important. A computer is actually a fairly simple arrangement of components — CPU, memory, various peripherals, and wires connecting them — with fairly simple interfaces among the components. Each of the components, in turn, is a simple arrangement of sub-components, which may themselves consist of smaller sub-components, and so on until the simplest level is reached. In this way, each component, at whatever level, can be treated as a separate, almost independent unit, making it relatively easy to understand. Without such a simply-connected modular structure, each piece would have the potential to affect any other piece, and considering all the possible interactions would be impractical to say the least.
Simplicity is not what we see in life. Although most life has modular structure — that is, organisms made up of organs made up of cells made up of organelles — the complexity of life is far greater than we see in design. The individual parts are still very complex, the interfaces between parts are very complex, and individual parts can usually directly affect a large number of other parts. This complexity is compounded by the fact that organisms change a great deal over their lifetimes. After decades of work, biologists have scarcely begun to understand how a human body works, much less how all the various organisms in an ecosystem work and interact. A good illustration of the complexity of life is the difficulty of designing a drug with no unwanted side-effects. But I need not elaborate; creationists themselves cite complexity as one of the hallmarks of life. Nothwithstanding disagreement over its source and significance, the complexity of life is another things that evolutionists and creationists can agree upon.
Although simplicity is a goal, complexity can still enter into design in some ways. One way that complexity enters into design is through the process of modification. If a change is made that renders part of a system obsolete, it is often easier to leave in some or all of the old parts, which then add unnecessary complexity to the design. Modification also adds complexity when changes are jury-rigged onto the existing structure rather than incorporated into the fundamental design. For example, some fixes to the Y2K bug involved checking the 2-digit year and trying to determine which century was intended, rather than the simpler and more correct, but much harder to do retroactively, fix of using 4-digit years. Such complexity is not necessarily bad design, either, since a frequent requirement of design is to get a working product out quickly, even if it is not as elegant as possible. Such complexity seems to appear in life, too, in the form of vestigial and jury-rigged features such as the appendix and the panda’s thumb. Evolutionists cite these as examples of poor design, which they may be from the standpoint of an omnipotent creator, but they are traits that life shares in common with our experience of design.
In summary, although creationists frequently cite complexity as evidence of design, simplicity would be the real evidence. Complexity can enter design through careless modification, but again such complexity can often be recognized as such, as with jury-rigged or vestigial parts. Besides, such complexity is what we expect from evolution.
Finally, design can become complex through evolutionary algorithms, which use repeated cycles of reproduction of initially random designs, selection from among them, and slight modifications and recombinations of the results (Davidson 1997). Such a design procedure does not need to minimize complexity because it always treats the design as a whole. The final design is extremely difficult to understand, but there is no need to understand it. The use of such a design method by humans is still in its infancy, but if it becomes widespread, we may then be justified in saying that life’s complexity looks designed. Of course, at that point "designed" and "evolved" become synonyms.
One of the defining features of life is that life reproduces itself. This is very different from designed things, which, with very few exceptions, are designed so that their production is separate from their other functions. A separate manufacturing process offers extreme benefits of efficiency for the simple reason that a manufacturing plant does not need to be built into each artifact. The few designed things that do reproduce themselves, such as computer viruses, can do so only because the production process and necessary resources are trivially cheap. And even the self-replicating human designs differ from life in that they do not go through the growth and development that living things experience before they can reproduce.
Let us suppose, along with Paley (1802, ch 2), that someone on a heath found a watch that "possessed the unexpected property of producing, in the course of its movement, another watch like itself." Paley said, "The first effect would be to increase his admiration of the contrivance, and his conviction of the consummate skill of the contriver." But would it? Such a watch, even with today’s technology, would be far too large to wear. Even if it were small enough, it would still be far larger than necessary. What’s more, the watch would need some way of obtaining raw materials, which would mean either the watch leaves its owner from time to time, or it manipulates its owner to bring it and the materials together. We could certainly admire the consummate skill of the contriver, but our admiration of the contrivance would be severely mitigated by the unnecessary impositions that reproduction would require. Reproduction may find some uses in design; for example, a self-reproducing factory for ordinary watches could conceivably produce an endless supply of useful watches with little requirement for labor. However, there is also a demand for non-reproductive manufacturing of designed items. Almost all designs that people are familiar with today would be useless if they had to include the capability of reproduction.
Repair of designed objects also has to come from the outside. The same economies that keep reproduction out of design also prohibit self-repair. Life forms, in contrast, include the ability to repair minor and in some cases extreme damage. This difference between life and design is so familiar that I need not go into further detail.
Form and Function
Another aspect of design is that form tends to follow function. A designer looking for a component to perform a particular function will, when possible, use an existing design rather than inventing a new one. When a useful innovation is introduced, it quickly gets applied to a wide variety of uses. This leads to the property that similar parts fulfill a common function even on very different products. For example, zippers of essentially the same design are found on clothing, tents, luggage, and other things. The same basic engine design can be found on motorcycles, motor boats, and lawnmowers. Some parts, such as screws, resistors, and software libraries, are even standardized so as to make it easy to use them in a wide variety of applications.
Life, in contrast, shows much less connection between form and function. Different taxa achieve similar functions with very different forms. For example, bats, birds, insects, and pterosaurs all have quite different wing anatomies. In different groups of insects, various forms of hearing organs are found in at least 11 different places on the body (Yack and Fullard 1993; Hoy and Robert 1996). And similar forms in life do not imply similar function. A human hand, a bat’s wing, a mole’s paw, a dog’s paw, and a whale’s flipper all have the same basic bone structure, despite their different functions of grasping, flying, digging, running, and swimming.
This difference between life and design is most apparent in the fact that life arranges naturally into a nested hierarchy, but design does not. With life forms, taxa defined by major features fall either entirely inside or entirely outside other taxa. This property led to the familiar hierarchical classification begun by Linnaeus. The hierarchy is not perfect, but it is a natural
hierarchy in that there are enough common traits to make most of the groupings obvious. With designed things, on the other hand, overlap is the norm. Although it is possible to form a nested hierarchy of designed things (indeed, it is possible to arrange any
set of different objects in a nested hierarchy), there is no natural
nested hierarchy. Consider sports, for example. There are lots of different features one could consider in classifying various sports: team sports, sports played on a rectangular field, sports played with a ball, and so on. However one classifies them, though, the groups overlap. The category of sports itself overlaps with other categories such as combat, art, and fitness. No obvious classification scheme presents itself. In fact, the only classification scheme that is commonly used with designed things generally is alphabetical order.
Trial and Error
Creationists seem to think of design as a single event that is done quickly and is over with. Even those creationists who see creation spread over time seem to envision many individual creation events. Real design, however, is a process. Designs are rarely completed in one attempt. They must be tested and modified to account for unforeseen consequences. Testing is done at many stages, from the first conception to field tests of the final product. Entire industries are devoted to the testing of structures, vehicles, computer systems, and other designs. All of these tests (if they are effective) result in information that guides the subsequent design. Furthermore, designers draw upon the experience of previous designers. When an architect designs a simple bridge or building, the process may seem straightforward, but that design is based on an education that comes from literally centuries of trial and error by earlier architects and builders (Petroski 1982).
This last point raises another observable property of design. Because designs are so often built upon previous designs, designs evolve over time, with new designs appearing as modifications of previous ones. This, of course, is also a property of life, as the fossil record shows. However, because people can intelligently combine a wide variety of innovations and other features, designs can change rapidly over time. Very few human designs have been around for more than a few thousand years, and most do not last nearly that long. Furthermore, the more complex designs are generally the shorter-lived. Although life changes over time, it does not do so nearly as fast as we see in human-driven modifications in design.
Purpose and Function
Creationists often claim that purpose indicates design. But purpose is hard to specify without knowing the designer, and it is often conflated with function. Purpose, as I use it here, is what someone intends a thing to be used for; function is what the thing actually does. The intent is useless for determining design, because it can be whatever anyone proposes, and the same object can, and often does, have different purposes for different people. Purposes often conflict. For example, a lynx’s purpose for a rabbit is likely quite different from that of the rabbit itself. Undesigned things often have purpose. For example, a stone need not be designed for people to give it a purpose as a pounding stone. The designer of an object can design a purpose into it, but others can find their own uses, as any MacGyver
Function also fails to indicate design for many of the same reasons. People can find functions other than what the designer intended. And functions can change in a heartbeat, as when the muscles of the fleeing rabbit become food for the lynx. Most importantly, undesigned things can have function — in fact, we expect function to evolve (see below). In short, purpose and function are too variable and subjective, and do not discriminate designed items from undesigned items.
Dembski proposes to recognize some design through a property he calls complexity-specification. If a pattern is highly improbable and yet matches a specification that was given beforehand, then that pattern has complexity-specification and, he says, must have been designed (Dembski 1999). For example, if I deal a hand of 13 cards that exactly matches an example bridge hand you saw in the newspaper that morning, you can be confident the deal was designed to come out that way. To detect this sort of design, Dembski proposes an "explanatory filter" which, if it rules out regularity (natural law) and chance, finds design as the only alternative (Dembski 1998). But because complexity-specification is defined simply as the lack of known causes, it is nothing more than an argument from ignorance given formal mathematical form. It does not say a thing about the properties of design.
However, it is instructive to consider complexity-specification at greater length anyway. Specification means matching something that was given elsewhere. Complexity (in Dembski’s unorthodox usage) simply means unlikelihood of occurring by chance in its observed configuration. By these definitions, patterns of complex specification can be produced naturally, too, with chance providing complexity and regularity acting selectively to reduce it. Evolution proceeds in large part by random mutations causing variation and natural selection winnowing that variation according to constraints of the environment. The mutations produce a form of complexity, and natural selection acts as a specifier. Since evolution includes complexity (mutations) plus specification (selection), it is only to be expected that evolution would produce complexity-specification in evolved life.
It is because Dembski’s filter fails to consider this combination of regularity and chance acting together that it will inevitably group together the products of evolution with design. Dembski claims that natural selection cannot create complexity-specification, but he only argues against the straw-man of creating it de novo
. Even he admits that natural selection can bring the specification in from the environment (Dembski 2001b). And this, after all, is what natural selection is all about.
Actually detecting the results of specification, though, can be a tricky business. Ideally, we conclude specification when an observation matches a complex pattern that was given earlier. This does not work, though, when the observation comes before we know what we are supposed to match it with. In such cases, the "specification" comes from finding a pattern in part of the object and seeing the same pattern carry through the rest of the object. (This is the general procedure that Dembski suggests. To the best of my knowledge, he has never provided a way of detecting complexity-specification in life that is objective and practical enough for two people to get the same results.) In other words, complexity-specification implies, in practice, some amount of regularity, but not so much that the word "complex" no longer applies. This just describes the intermediate level of structure discussed in a previous section. And since this property originates via both natural processes and design, it cannot be used to distinguish between them.
Another property that has been taken to indicate design is functional integration, or multiple parts working together to produce a particular function or end (Lumsden, quoted in Alters 1995). This property seems intuitively appealing because much design consists of assembling parts to create a particular function. But functional integration may be claimed even when origins are known to be natural. For example, the climate of the Mississippi Basin is determined by the Rocky Mountains, the Gulf of Mexico, trade winds, and other factors. Since the climate is a functional end (it allows an ecology suitable for certain organisms) produced by multiple factors, it fits the definition of functional integration. And in fact this example was used as an argument for design by the 19th-century creationist George Taylor (Morton 2001). Obviously, though, any arrangement of physical factors, whether designed or not, is going to create some kind of climate. Since functional integration arises from non-design, it cannot reliably indicate design.
It may still be argued that functional integration that arises naturally is not necessarily very functional (the inland Antarctic climate is not terribly hospitable) or very integrated (we do not often think of trade winds, mountains, and a gulf as a single unit). Again, however, functional integration is a quality of evolution as well as of design. Evolution cannot proceed without units to reproduce. "Unit" already implies some integration, and reproduction is itself a function. Furthermore, survival entails many additional functions such as finding food and escaping predators. Natural selection would ensure that such functionality and integration are maintained. So functional integration indicates evolution at least as much as it indicates design.
Although it applies not to life but to the universe around it, the fine-tuning argument for design deserves some consideration here. This argument claims that many physical constants and other features of the universe fall in the only narrow range that would allow life to be possible — so many features, in fact, that the combination could not be explained by chance and must be designed (Barrow and Tipler 1986; Ross 1994). Others have shown the problems with this argument (Le Poidevin 1996; Stenger 1997). Of interest here is a prior question, namely whether fine-tuning indicates design in the first place.
Fine-tuning is an aspect of design, of course; the term even comes from engineering. Designing components to mesh well with other components or with the outside environment is a common necessity. However, designers are not entirely stupid. When they fine-tune, they tune the parts that are easy to change. If parts are added later that have not been built yet, they fit the new parts to the existing design, making the fine-tuning of the new parts part of designing them. Fine-tuning is done to malleable parts and parts that come later.
This is very different from the fine-tuning argument from "intelligent design theorists". The physical constants of the universe, to all appearances, are not easily changeable, if they are changeable at all. Life, on the other hand, is extremely adaptable. Furthermore, life appeared much later than the universe and exists in only a minuscule fraction of it. The universe we see is compatible with a universe designed in fine detail to support life as we know it (design theory is compatible with anything), but an argument based on analogy to design
would claim that life is fine-tuned to the universe, not vice versa. The claim that the universe was fine-tuned for life is the very opposite of a design argument.
Table 1 ( p 34) shows a summary of the similarities and differences between life and design. Although there are a number of similarities, the differences are large and important. In particular, life’s growth and reproduction alone are enough, it seems to me, to place life and design in quite separate categories. Life’s complexity and its nested hierarchy of traits are also highly significant differences. The overall conclusion is clear: life looks undesigned.
|Table 1. Similarities and Differences Between Life and Design|
|Intermediate level of structural complexity||Intermediate level of structural complexity
|Modular structure||Modular structure
|Evidence of careless modification (jury-rigging, vestigial parts)||Evidence of careless modification (jury-rigging, vestigial parts)
|Change over time; new forms are modifications of previous forms||Change over time; new forms are modifications of previous forms
|Functional integration||Functional integration
|Blueprints, tools, and other evidence of the design process||No evidence of design process
|Simple organization||Complex organization; intermodular interdependence
|Manufacture||Reproduction, growth, and development
|Generally repaired from outside||Self-healing, at least in part
|Form follows function||Forms follows nested hierarchy
|Rapid change||Slow change|
It bears repeating that the properties of design that I have considered are properties of human
design, and they do not necessarily apply to a supernatural designer. However, human design is the only model of design we have by which to tell what design looks like, to the extent that design can be said to look like anything. If it does not look like this, it does not look designed.
The reader has probably realized by now that most of the aspects of life that look designed are also evidence of its evolution. In the cases of evidence of careless modification and change over time, the connection is explicit. An intermediate level of structural complexity probably arises from the selection and recombination inherent in evolution. Functional integration is not necessarily evidence for evolution but is an essential aspect of it. Modular structure is the only other aspect that design has in common with life that is not also
evidence for evolution, but it is at least consistent with evolution. Even fine-tuning argues for life’s changing to fit the environment.
To the extent that life looks designed
, life looks evolved
. This should not come as a great surprise, because the process of design and the process of evolution share some important commonalities (see also Shanks and Joplin 2000). Both processes build upon what has gone before, and both processes select the "good" features and discard what does not work. There are also important differences, to be sure, but the similarities in process should not be overlooked.
Creationists have been criticized for their misrepresentations of biology and other sciences. Their representation of design is no less faulty. They consider complexity to be a hallmark of design, while simplicity is typically the designer’s aim. They believe that design and chance are mutually exclusive, whereas trial and error is sometimes used in design and, in the long run, is an inevitable and invaluable part of it. Finally, they treat design as an event, when in fact it is a process — a process that itself can be designed. Such misconceptions not only make for flawed theology, they cannot be good for engineering practices, either.
In fact, it would not be an exaggeration to say that "intelligent design theory" is not about design at all. Since most of the people who espouse it seem to view the design as a sudden all-at-once event, their model (not surprisingly) seems to be that of the fiat
creation described in the Bible and Koran, not the extended process that familiar design entails. If creationists want to describe a different mechanism than design, they should use a different label for it. I suggest "decree", which has the advantage of fitting the theological position that underlies their ideas.
In both science and engineering, precise specifications are important. Two hundred years have passed since Paley popularized "intelligent design theory" (Paley 1802), and creationists have not yet satisfactorily clarified what they mean by "design", much less suggested useful tests for detecting it. At best, "intelligent design theory" is undefined and thus wholly useless. At worst, taking the phrase "looks designed" at face value as indicating analogy to human design, "intelligent design theory" is contradicted by the evidence.