Intelligent Design: An Argument from Ignorance Masquerading as Science
Two important qualifications before I begin.
First, modern cosmology does not actually establish that nothing exists outside our universe, nor does it prove that God cannot exist. What it does establish is that our physical theories describe the universe, including spacetime itself, from its earliest accessible state onward, and they provide no observational evidence for a physical realm “outside” it (the eternally-expanding inflaton field is purely hypothetical). This distinction makes a stronger and more defensible bridge to the following argument.
Modern cosmology has done something profoundly unsettling to the old idea of a Creator standing outside the universe and designing it from the outside. In the standard cosmological picture, the universe is not an object sitting inside a larger container called “space.” Space and time are themselves part of the universe. Asking what existed “before” the universe or what lies “outside” it may therefore be like asking what lies north of the North Pole: The question assumes a framework that, at that boundary, no longer exists.
This does not prove that God does not exist; science cannot establish that a supernatural being is impossible. But it does remove the familiar physical picture of God as an external engineer who stands beyond the cosmos, examines the materials available to him, and deliberately assembles them. There is no scientifically established cosmic workshop outside the universe in which such a designer could be located. And that distinction matters enormously when we turn to intelligent design (ID): If the complexity of nature is going to be explained by an intelligent designer, we must first ask what, exactly, that explanation is supposed to mean — and whether invoking a designer actually explains anything at all.
Second, the Earth is an oblate spheroid (a.k.a. round), and approximately 4.54 billion years old based on radiometric dating of meteorites and lunar samples. This is a very, very long time. Close your eyes and count to a million without a break. See you in 11.57 days. Nonstop. When you are done count to a billion. A billion seconds is more than 30 years. The Earth is fantastically old; 4.54 billion years is an enormity of time for chemistry, biochemistry, and molecular biology to invent life.
With that said, here we go.
- “Irreducible complexity” does not imply non-evolution. This is probably the central scientific objection to Michael Behe’s version of ID (look up Kitzmiller v. Dover Area School District).
Behe argues that some biological systems are “irreducibly complex (IC)”: Remove one component and the system ceases to perform its present function. Therefore, he argues, the system could not have evolved through successive selectable modifications, because intermediate stages would supposedly have been nonfunctional.
The problem is that this reasoning confuses: The complete system cannot function if a component is removed with The system could not have evolved incrementally. Those are not equivalent propositions. Evolution does not require every intermediate stage to perform the “same function” as the final system. A single component can perform one function; be duplicated or modified; become incorporated into another system; acquire a new function; lose its original function later. This is co-option/exaptation.
2. Evolution can construct an IC system incrementally. Suppose a final biological machine consists of: “A + B + C + D + E” and removing any one component destroys the “present” function.
ID says A+B+C+D could not have existed before E, because the incomplete system wouldn’t work. But evolutionary theory asks a different question: Did A, B, C and D have functions before E appeared? If they did, selection could favor each stage independently. Eventually E can be incorporated and the resulting system can acquire a new function. Thus: IC is a property of the final architecture; it isn’t evidence that the architecture had to appear all at once. This distinction has been explicitly discussed in the scientific literature.
3. The bacterial flagellum is not an “all-or-nothing” machine. The bacterial flagellum is probably the most famous ID example. Behe argues that its enormous molecular complexity makes it irreducibly complex. But evolutionary biology has identified homologous components and simpler systems associated with flagellar structures. Most importantly, portions of the flagellar apparatus are homologous to components of other bacterial systems, including the Type III secretion system (T3SS), providing precisely the sort of evolutionary relationship that ID says should not exist. The molecular components have evolutionary histories independent of the final machine. That makes an incremental evolutionary history biologically plausible.
4. The blood-clotting cascade. Behe has used the mammalian blood-clotting cascade as an example of irreducible complexity. The argument sounds powerful: Remove one component and clotting fails therefore the complete cascade could not have evolved gradually. But comparative genomics reveals that different vertebrate lineages possess different versions of the clotting system. Some organisms lack components found in mammals. That observation is devastating to the strongest version of the argument.
Think about it. If the system were genuinely irreducible in the evolutionary sense, we would expect organisms without supposedly indispensable components to be impossible. Instead, we find evolutionary variation in the system. This issue was particularly significant during the Dover litigation; even Behe acknowledged that organisms exist without some of the components he had characterized as necessary to the system.
5. “Complexity” itself is not evidence of intelligence. This is a deeper philosophical/scientific problem. ID reasoning says complex biological structures are improbable by chance therefore must have been designed. But evolution isn’t random assembly. Natural selection is a cumulative filtering process. Mutations introduce variation; differential reproduction systematically changes the frequencies of variants. Consequently, extremely complex structures can arise without the entire structure having been assembled randomly. The important distinction is: Random variation ≠ random evolutionary outcome. Natural selection is nonrandom with respect to reproductive consequences. This is one reason the watchmaker analogy doesn’t map cleanly onto evolutionary biology.
6. “Specified complexity” has not produced an independent biological detection method. William Dembski developed the concept of specified complexity as another proposed signature of design. The problem is that a scientific design criterion must do something like this:
Given an unknown biological structure, apply an objective procedure and determine whether it was produced by natural processes or intelligence. ID has never produced a generally successful empirical procedure capable of doing that. The literature has examined Dembski’s framework and concluded that it does not establish the necessity of an intelligent agent and has not produced successful biological models or research programs.
7. ID generally substitutes an unexplained cause for an incomplete explanation. The classic God-of-the-gaps objection, although ID advocates reject the characterization. The logical structure is often (i) Evolutionary biology has not completely explained X. (ii) X is extremely complex. (iii) Therefore, X was designed. But 3 doesn’t follow from 1. At most we don’t yet know. A scientific explanation needs to supply a mechanism that generates predictions. A designer hypothesis does not tell evolutionary biologists what mutations should occur; what genes should be duplicated; what intermediate structures should exist; what phylogenetic distribution should occur; what molecular homologies should be present; when the intervention occurred; how the designer implemented the modification.
That is why critics characterize ID as an explanatory stopping point rather than an alternative causal model.
8. The “designer” hypothesis makes remarkably few risky predictions. Consider evolutionary theory. It predicts that we should find nested genetic relationships; shared pseudogenes; endogenous retroviral insertions; homologous proteins; transitional morphological characteristics; gene duplication; conserved developmental pathways; geographic patterns consistent with common ancestry. And we actually find them.
ID doesn’t generate an equivalent suite of independent predictions. If an organism is complicated it is “Designed.” If it is simple then it is “Designed efficiently.” If two organisms share structures: “The designer reused components.” If they don’t: “The designer used different solutions.” The problem is that “almost any observation can be accommodated after the fact.” A hypothesis that accommodates every possible outcome has very little predictive discrimination. Peer-reviewed analyses have specifically criticized ID for failing to produce testable hypotheses and empirical predictions.
9. The biological world contains enormous evidence of “bad design.” If biological structures were engineered from scratch by an intelligent engineer, we might expect them to display something approaching optimal engineering Instead, evolution produces extraordinary examples of historical constraint and kludge.
Examples include: The recurrent laryngeal nerve. It travels from the brain down into the thorax, loops around a major blood vessel, and then travels back upward to the larynx. That makes little sense as a fresh engineering solution. It makes excellent sense as a consequence of vertebrate evolutionary history.
The vertebrate retina. Photoreceptor cells are positioned behind neural and vascular layers rather than facing incoming light directly. Again, this is understandable as a consequence of developmental and evolutionary history.
Get Kandahar Doc’s stories in your inbox
Join Medium for free to get updates from this writer.
The human appendix: A remnant of a much larger ancestral digestive structure is a straightforward evolutionary interpretation.
The human birth canal. The competing requirements of upright bipedal locomotion, pelvic architecture, and large-brained offspring produce a remarkably constrained system.
Evolutionary theory predicts such compromises. The important point isn’t that these structures are “bad.” They’re often extraordinarily functional. The point is that they look like historically modified systems, not systems designed from scratch. The scientific literature explicitly identifies imperfect adaptations as evidence against some forms of ID.
10. Evolution predicts molecular “junk” and historical leftovers. Evolution predicts that genomes should contain: Pseudogenes; broken genes; duplicated genes; endogenous retroviral insertions; transposable-element remnants; nonfunctional mutations inherited from ancestors. And they do. The strongest evidence comes when apparently useless genetic sequences occur at the same chromosomal locations in related species. That’s exactly what common descent predicts. A designer could theoretically have created the same sequences independently but that makes the designer hypothesis increasingly flexible rather than increasingly predictive.
11. Evolution actually produces new biological functions. One of the implicit assumptions behind some ID arguments is that evolution can modify existing structures but cannot generate genuinely new functional complexity. That proposition is empirically false.
Examples include: gene duplication followed by functional divergence; new regulatory elements; novel enzyme functions; changes in developmental regulatory networks; protein-protein interaction changes; metabolic innovations. The key mechanism is frequently duplication leads to redundancy which better tolerates mutation which brings about divergence which develops as new function. Once a gene is duplicated, one copy can maintain the ancestral function while the other explores sequence space without immediately destroying the original function. This provides a straightforward route toward increased biological complexity.
12. Co-option is ubiquitous in evolution. A particularly important concept is “exaptation.” A structure can acquire a function different from the function for which it originally evolved. Classic examples include: feathers, originally associated with functions other than powered flight; jaw bones and evolutionary transformation into mammalian middle-ear bones; existing proteins recruited into new molecular complexes; existing developmental pathways see deployment in new morphological contexts. This is precisely why looking at a finished structure and asking: “How could this have arisen in its present form?” can be misleading. Evolution doesn’t have to construct the final object from scratch. It can modify, duplicate, rearrange and redeploy existing components. Remember 4.54 billion years.
13. The designer hypothesis doesn’t explain the designer. Suppose we grant: Biological complexity requires an intelligent designer. Then the designer presumably possesses enormously greater: information; intelligence; organizational complexity; causal capability. Why doesn’t that require explanation? The ID response is usually that the designer is outside the domain of the biological question or that the designer’s nature is not being addressed. But scientifically that means ID has moved the unexplained complexity one level upward. This is not necessarily a refutation of the theological proposition that God exists. It is a problem for treating “intelligent designer” as a scientific explanation of biological complexity.”
14. Comparative biology independently predicts common ancestry. Evolution isn’t supported merely because scientists can invent plausible evolutionary stories. There are multiple independent data streams that converge on common descent: comparative anatomy; embryology; paleontology; biogeography; genetics; molecular phylogenetics; endogenous retroviruses; pseudogenes; protein homology; chromosome structure.
These datasets weren’t all constructed to support Darwin’s theory. They independently converge on a hierarchical pattern of biological relationships. That is extremely difficult for a design hypothesis to explain without making the designer’s behavior effectively unconstrained. The National Academies’ synthesis emphasizes precisely this convergence of evidence for evolution.
15. ID has not generated a competing research program. Scientists don’t reject ID simply because they are philosophically committed to Darwin. The question is: What research program has ID generated that competes with evolutionary biology? The answer, after several decades, remains essentially none. There has been no established ID methodology that allows researchers to: Identify a biological structure; demonstrate that natural evolutionary mechanisms cannot produce it; independently detect the designer; determine the design mechanism; generate successful predictions from the design hypothesis.
The Strongest Scientific Case, Distilled.
If I were reducing the peer-reviewed scientific criticism to five propositions, I would put it this way:
“Irreducible complexity prevents evolution.” RESPONSE: A final system can be irreducible while its components evolved separately and were subsequently co-opted.
“Complexity implies intelligence.” RESPONSE: Natural selection is a cumulative, nonrandom filtering process capable of generating complexity.
“Evolution cannot produce new functional information.” RESPONSE: Gene duplication, mutation, recombination and regulatory evolution demonstrably generate new functions.
“Biological systems look engineered.” RESPONSE: Their imperfections, historical constraints, homologies and genomic remnants strongly reflect evolutionary history.
“Where evolution cannot yet explain something, design is the explanation.” RESPONSE: Lack of a current explanation does not constitute evidence for a designer, particularly when the designer hypothesis generates no discriminating predictions. |
And there is an important historical point: Michael Behe himself conceded under oath at Dover that there were no peer-reviewed papers providing detailed experimental or mathematical accounts of how an intelligent designer designed a biological system, and no peer-reviewed papers supporting his specific claims that the flagellum, blood-clotting cascade, or immune system were intelligently designed.
Closing Argument
As noted above, modern cosmology has eliminated the picture of a cosmic workshop outside spacetime in which an engineer could stand and assemble the universe. At the same time, the Earth has existed for roughly 4.54 billion years — an interval so vast that chemistry, biochemistry, and natural selection have had ample opportunity to generate complexity through ordinary physical processes. Against that background, the claim that certain biological systems are “irreducibly complex” and therefore require an intelligent designer does not survive scrutiny.
Irreducible complexity describes the architecture of a finished system; it does not describe the historical path by which that architecture was assembled. Components can evolve for one function, be duplicated or modified, and later be co-opted into a new system whose present function would fail if any part were removed. The bacterial flagellum and the vertebrate clotting cascade, two of the most frequently cited examples, illustrate precisely this pattern. Homologous parts exist in simpler molecular machines; comparative genomics shows that different lineages possess different subsets of the cascade. What looks all-or-nothing in the final product has a stepwise evolutionary history.
More generally, complexity is not evidence of intelligence. Natural selection is a cumulative, non-random filter. Mutation supplies variation; differential reproduction systematically retains variants that increase reproductive success. Over deep time this process generates functional complexity without requiring the entire structure to have been assembled at once. Gene duplication, regulatory evolution, and exaptation routinely produce new functions. The empirical record contains both the predicted molecular leftovers of history (pseudogenes, endogenous retroviruses, shared broken genes) and the predicted compromises of that history (the recurrent laryngeal nerve, the inverted retina, the human birth canal). These features are what we expect from descent with modification; they are difficult to reconcile with design from scratch.
A scientific design hypothesis would need to do more than note that something is complicated. It would need to identify structures that natural processes cannot produce, supply an independent detection method, generate risky predictions, and yield a research program capable of testing those predictions. After decades, intelligent design has done none of these things. It accommodates any observation after the fact, shared features mean the designer reused parts; differences mean the designer chose new solutions; simplicity means efficient design; complexity means sophisticated design, and therefore explains nothing that could not have been explained without it. Where evolutionary biology is incomplete, the absence of a current explanation is not positive evidence for a designer.
The distinction drawn at the outset remains important. Nothing in this critique demonstrates that a supernatural being is impossible. What it does demonstrate is that “intelligent designer” does not function as a scientific explanation of biological complexity. It relocates the unexplained information one level upward, supplies no mechanism, and generates no discriminating predictions. Evolutionary theory, by contrast, continues to unify comparative anatomy, embryology, paleontology, biogeography, genetics, and molecular phylogenetics under a single, testable framework of common descent and natural selection. That framework accounts for both the elegance and the historical contingency of living systems. Until a competing research program can do the same, the scientific case against treating irreducible complexity as evidence for design stands.
