Arguments from Bad Design

Summary

A biological structure can perform useful work and still have a defect.  It can also look awkward before its function is understood.  A serious argument about design must identify the disputed function, establish the cost, and compare plausible alternatives under stated constraints.  Neither a complaint about appearance nor an appeal to an unknown purpose settles the question.

The evidence supports four main conclusions:

• Several familiar complaints overstate what anatomy shows. Retinal cells can guide light rather than merely obstruct it. A removable organ need not be useless. These findings correct particular objections; they do not establish that every biological feature is optimal.
• A tradeoff must be demonstrated. Naming a possible benefit does not show that it compensates for a cost, or that the present balance is best. Experimental challenges to the familiar account of pelvic width show why attractive explanations need tests.
• Ancestry and performance are different questions. A structure can retain useful work while losing an ancestral function. Discovering that work therefore does not, by itself, refute common ancestry. Conversely, finding a defect does not demonstrate a mechanism that constructed the organism.
• Creationist claims must remain answerable to evidence. Original creation followed by damage is a different proposal from present universal perfection. It needs evidence of the original condition and the changes. Unknown purposes cannot serve as an unlimited exemption from testing.

The defensible Christian response is to investigate alleged failures and acknowledge real ones.  A catalogue of awkward-looking parts cannot establish the absence of intelligence.  A catalogue of newly discovered functions cannot establish creation either.

Part 1: What exactly is being alleged?

A defect, an inherited feature, or a harmful purpose?

Five different claims frequently appear under the label bad design.  They require different evidence.

A parasite can be effective at invasion while harming its host.  Its effectiveness does not answer a moral objection about a good Creator.  That question belongs in Predation, Parasites, and Pathogens.  Calling such equipment badly engineered confuses successful performance with a harmful outcome.

A broken component raises a different question from a costly component that works normally.  The distinction matters because a repairable failure can identify damage against a known working baseline.  Explaining a normal arrangement as damage requires evidence that its earlier state differed.  Merely observing a disadvantage cannot supply that history.

The same discipline applies to ancestry.  If two species share a peculiar arrangement, its usefulness does not erase the resemblance.  Its inconvenience does not establish inheritance without comparative evidence.  The historical argument must explain which features correspond, which groups possess them, and why the proposed sequence of changes fits the observations better than alternatives.

The Recurrent Laryngeal Nerve applies this distinction to useful nerve branches, embryonic development, and the different routes found in humans and camelids.

Male Nipples shows both sides reading one structure as shared development, and why the open question is why men lack the removal switch that mice and marsupials use.

Embryonic Leftovers applies the ancestry test to embryos:  recapitulation is dead, but the shared mid-embryo stage is measured, and the design reply has yet to explain why development runs through it.

Bad Design in Biochemistry applies this test below anatomy:  most of the design answers hold, but uric acid is a broken gene, damage rather than design.

Vestigial does not have to mean useless

Darwin explicitly allowed an organ to survive with a reduced set of functions.  In the first edition of On the Origin of Species, he wrote: “Or an organ might be retained for one alone of its former functions.”¹  His claim was about change from an earlier condition, not a rule that every reduced structure must do nothing.

Consequently, discovering a useful function answers this structure is functionless.  It may leave this structure has lost an ancestral function untouched.  That second claim still needs historical support; Darwin's permission does not establish any particular example.  Both the inference from usefulness to independent creation and the inference from small size to evolutionary reduction skip a necessary step.

Even a genuine evolutionary remnant need not be defective.  Retaining a useful attachment or support after another role disappears could improve the organism's present performance.  The history and the engineering verdict should be reported separately.  This prevents a dispute over the word vestigial from taking the place of anatomical investigation.

The Vestigial Structures article shows that the modern vestige claim is a three-part common-ancestry argument (homology, reduction, ancestral distribution) that a functional finding never reaches, and sorts the classic cases accordingly.

Atavisms tests the related claim that ancestral traits reappear:  extra-toed horses, whale hind limbs, dog dewclaws and chicken embryo teeth are genuine reawakenings of retained programs, the popular human cases fail, and even the genuine ones show loss within a lineage rather than how far back it goes.

Body Hair, Nails, and Wrinkled Fingers applies this test to three small features:  hair, follicles and nails have measured uses, while the grip function of wet-finger wrinkling remains unsettled.

Part 2: How to test a criticism

State the task and the cost

A useful criticism begins with a concrete question: how much light reaches a receptor, how much energy movement requires, or how often a passage fails at its normal task?  Calling a route circuitous or a part ugly does not measure any of those outcomes.

The objective cannot be chosen solely to secure the desired answer.  An eye's role in vision gives a defensible reason to examine optical performance.  Survival, development and maintenance can add relevant requirements.  A proposed hidden purpose becomes evidence only when it supplies a testable prediction or an independently supported constraint.  Its bare possibility can show that a conclusion is not logically compulsory; it cannot demonstrate that the conclusion is false.

Performance also depends on conditions.  An arrangement that works well in one environment may fail in another.  A test should specify age, activity, exposure and the outcome measured where those factors matter.  A present benefit may be small, restricted to unusual circumstances, or outweighed by another cost.  Those are questions for measurements rather than adjectives.

Compare a feasible alternative

A proposed improvement should explain what changes and what must continue to work.  Moving a tissue in a drawing is not yet an account of how the altered organism develops, supplies it, connects it and keeps it functioning.  Those requirements matter whether the argument concerns engineering or evolution.

But critics do not have to build a replacement organism that wins on every conceivable axis before identifying a defect.  A demonstrated failure at a central task is evidence.  A practical modification that improves that task without a known compensating loss can strengthen the criticism.  Confidence depends on how well the important constraints have been examined, not on impossible knowledge of every purpose.

The scale of the claim determines the burden.  Establishing that an arrangement loses some light is easier than establishing that a different whole eye would perform better throughout its life.  Establishing that one proposed improvement works is easier than showing that no intelligent agent could ever choose the existing arrangement.  The last claim goes far beyond comparative anatomy.

Likewise, failure to construct a viable redesign does not prove optimality.  It may expose a real constraint, or simply the investigator's limited knowledge.  Defenders of design should say which result they have obtained.

Bird Lungs and Mammal Lungs runs this test on a real, better alternative:  the bird lung wins on oxygen extraction but loses on diving and is not needed at sea level.

The Mitochondrial Genome runs this test on moving 13 genes to the nucleus:  12 of the proteins go to the wrong membrane when made outside, and the one attempt in patients has not been shown to work, though the genome's fast mutation rate remains an unanswered cost.

Price the tradeoff

A tradeoff exists when improving one relevant outcome worsens another.  It does not follow that the current compromise is best.  There may be a different arrangement that improves both, or a better balance between them.  Even when every available alternative has a cost, the size of that cost matters.

For example, the proposition that a thicker protective covering needs more material is intelligible.  It does not determine the best thickness.  That requires evidence about injury, weight, construction and the conditions the organism faces.  Substituting the word tradeoff for those comparisons leaves the original question unanswered.

Tradeoffs also need to fit the alternative actually proposed.  A comparison between an intact system and one stripped of a necessary component tests removal.  It does not test every possible replacement.  Conversely, comparing two living species can establish that two workable arrangements exist, but their other differences may prevent the comparison from isolating the effect of one feature.

These rules apply equally to evolutionary explanations and creationist defences.  An appealing account of why selection favoured a compromise can fail a performance test.  An appealing account of the Creator's purpose can fail the same test.  Neither account becomes empirical merely by describing a possible advantage.

Part 3: Three examples of what the evidence can establish

Retinal cells guide light

Kristian Franze and colleagues demonstrated a function that simple descriptions of retinal obstruction miss.  Their 2007 experiments showed that Müller cells can channel light through retinal tissue.  In an isolated-cell arrangement, placing a cell between optical fibres improved transmission.  With the fibres obliquely aligned, measured transmission increased by as much as a factor of nine compared with the same setup without the cell.²

That experiment establishes light guiding.  It does not measure a ninefold improvement in human eyesight, nor compare a complete vertebrate eye with a cephalopod eye.  The denominator is part of the result.

Amichai Labin and colleagues sharpened the finding in 2014.  Their human-retina model predicted an approximately elevenfold concentration of light into a cone at 560 nanometres.  The baseline was light propagating freely through the retina.  Experiments with isolated guinea-pig retinas supported wavelength-dependent guiding; the human result was computational.³

The legitimate correction is substantial: cells lying before the receptors can contribute to light delivery.  The stronger claim that retinal inversion is therefore the best possible architecture does not follow.  Light concentration is one property of an eye, and neither study compared every feasible retinal arrangement.  A functioning mechanism can mitigate a cost, provide a benefit, or do both.  Its discovery alone does not decide which historical process produced it.

The Inverted Retina examines these experiments alongside photoreceptor maintenance, the blind spot, foveal specialization, and cephalopod and bird comparisons.

Why Eyes Use the Light They Use tests the complaint that vision covers only a narrow band, and finds the band fixed by pigment noise, lens blur and water, with every animal that sees beyond it paying in acuity or noise.

An appendix hypothesis is more than nothing, less than proof

R. Randal Bollinger and colleagues proposed in 2007 that the human appendix shelters beneficial bacteria that can help recolonize the bowel after disruption.  Their evidence included blinded assessment of samples from one deceased donor's intact, unprepared colon.  Bacterial biofilms were densest and most continuous in the appendix, becoming less prominent farther along the bowel.⁴

This is a relevant anatomical observation, not a speculative function invented without evidence.  The location and protected arrangement motivate a testable reservoir hypothesis.  But that paper did not directly follow bacteria from the appendix as they reseeded a patient's bowel after infection.  The observation, the proposed mechanism and its effect on lifetime health are separate evidential steps.

Survival after removal would not disprove an occasional benefit.  Equally, a plausible benefit would not erase appendicitis or establish that the present anatomy has the best benefit-to-risk balance.  The paper itself discusses the organ's vulnerability to disease.  Neither side should infer the whole balance from its preferred half.

This example also illustrates why function and ancestry must be separated.  A demonstrated reservoir role would answer a claim of functionlessness.  Deciding whether the organ developed from an earlier structure requires comparative evidence in addition to that function.  The experiment cannot be made to settle both questions by changing what vestigial means midway through the argument.

Pelvic tradeoffs need testing too

A familiar explanation says that a wider pelvis would ease childbirth but make walking more costly.  Anna Warrener and colleagues tested part of that account in 2015.  One experiment recruited 26 adults for movement, force, metabolic and MRI measurements.  A separate experiment measured metabolism in 15 adults.  They found no significant relationship between pelvic width and the overall energy cost of walking or running in their analyses.⁵

The result challenges a simple claim of a demonstrated energetic penalty.  It does not prove that width has no effect, that every much wider pelvis would work equally well, or that all childbirth constraints disappear.  The participants were fit young adults, and analysis-specific samples were smaller than recruitment totals.

Nicole Grunstra and colleagues argued in 2023 that broader pelvic tradeoffs remain supported.  They discussed reanalyses yielding weak positive width-cost relationships, presented a further model-data reanalysis, and emphasized other functions and the need for larger samples.⁶  The dispute concerns the size and interpretation of particular constraints within evolutionary accounts; neither result dates human origins or establishes a creation chronology.

The lesson is methodological.  A widely repeated compromise is not a substitute for measurement.  Its defenders must establish the relevant penalty, while its critics must respect what a small study can exclude.  Creationist writers should not borrow an uncertain evolutionary tradeoff and present it as settled proof of optimal design.

The Human Body Under Criticism applies these tests to the whole-body complaint and finds that its one full redesign concedes its own costs and has never been priced.

Part 4: What creation and intelligent design actually claim

Design does not logically entail maximum performance

Michael Behe directly rejects that entailment: “Clearly, designers who have the ability to make better designs do not necessarily do so.”⁷  His argument in Darwin's Black Box rests on a claimed pattern of coordinated complexity, not a premise that every organ achieves the best imaginable performance.  The success of that positive argument is a separate question.

Behe's distinction is logically sound.  A designed object can be inefficient, damaged or compromised.  It follows that an imperfection alone cannot disprove every possible involvement of intelligence.  It does not follow that imperfections are irrelevant to every proposed designer.  A claim that a particular agent consistently prioritizes a particular outcome can be tested against repeated avoidable failures at that outcome.

This distinction prevents the defence from becoming empty.  If design advocates cite good performance as evidence of a Creator's intentions, they must explain how poor performance affects that same inference.  They cannot specify intentions when performance is impressive, then declare all intentions unknowable when it disappoints.  The consistent options are a modest design inference with modest predictions, or a more specific account that accepts a greater burden of testing.

Original creation followed by change is a different proposal

Young-earth creationists also distinguish the original creation from the present world.  Warren Shipton's account of harmful biological relationships explicitly includes “gene loss, the expression of front-loaded genes now facilitating pathological life cycles, or other scenarios.”⁸  His proposal includes altered expression and relationships as well as loss.  It would be inaccurate to treat it as a claim that every present harmful feature was originally created in its present role.

Nevertheless, invoking the Fall does not establish a biological reconstruction.  A particular degeneration claim needs an earlier condition, the changes that produced the present one, and evidence connecting them.  A demonstrated loss of function can support part of that account.  It cannot by itself date the loss to the Fall or identify the state of all organisms at creation.

The Christian distinction between a good creation and a damaged world therefore avoids one false premise: creationism need not promise present universal perfection.  It does not supply a universal answer to anatomical objections.  Where the cause of a disadvantage has not been established, the honest description is unresolved.  The parasite overview examines the strengths and limits of those proposals in detail.

Evolution does not predict universal bad engineering

Darwin likewise denied that natural selection must produce “absolute perfection”.¹  His argument was that selection favours relative advantages under existing conditions.  Thus a mixture of effective performance, retained structures and costs is compatible with evolutionary explanations.

This matters when a creationist defence succeeds.  Discovering good performance may defeat a particular claim that a feature is useless or obstructive.  If the evolutionary account also permits that function, the discovery does not automatically favour creation over evolution.  A discriminating argument must identify an observation that the competing accounts expect differently and explain why.

The reverse inference requires equal care.  If anatomy supports inherited constraints, that is positive evidence for a historical relationship under the tested account.  It does not establish that intelligence could never operate through inheritance.  If an organ fails, that is evidence about performance.  It does not demonstrate all the mutations, developmental changes and selective steps needed to construct it.  Evidence should be credited for the question it actually answers.

Objections and responses

1. An all-powerful Creator would have no constraints.  A claim about what God could miraculously do differs from an engineering comparison among organisms operating through the physical processes under discussion.  Within that comparison, material and developmental requirements are relevant.  But invoking physical laws does not prove that the Creator had to choose this particular arrangement.  The defender still owes the connection between the actual constraint and the disputed feature.

2. A critic cannot judge a design without knowing every purpose.  That standard would make even ordinary defect detection impossible.  Known functions permit bounded judgments.  Uncertainty about other purposes should qualify a sweeping verdict, not erase a measured failure.  A newly demonstrated purpose can change the balance; a permanently unspecified one cannot establish it.

3. Finding a function vindicates design.  It vindicates a functional claim to the extent that the experiment supports it.  The retina studies establish light guiding.  They do not identify an agent or show that selection could not retain the property.  Design arguments need additional evidence that bears on the causal question.

4. The accumulation of defects makes the case stronger than any single defect.  It can.  Several independent failures under a specific design prediction can weigh more heavily than one ambiguous example.  A cumulative argument still needs sound entries and a stated prediction.  Repeating the same unsupported assumption across many organs does not create independent evidence.  Nor can strong performance elsewhere cancel a genuine defect without an explanation connecting them.

5. No one has produced a superior replacement, so the criticism fails.  That demand confuses observing a failure with solving it.  A critic can identify a reliable shortfall before discovering its remedy.  A proposed alternative strengthens the case about avoidability; it is not always required to establish that the shortfall exists.  Conversely, merely sketching an alternative does not establish its viability.

6. Degeneration explains any defect that remains.  Only where degeneration has supporting evidence.  The possibility of a damaged original establishes logical compatibility with creation, not the actual history of a particular structure.  Making the explanation cover every outcome without a recoverable baseline removes the biological test.

7. These cautions make any conclusion impossible.  They permit narrower and stronger conclusions.  A measured optical guide, an observed biofilm distribution and a tested locomotor prediction are real findings.  Keeping their denominators and limits intact produces knowledge that survives scrutiny.  Inflating them into universal verdicts weakens their legitimate force.

Issues and unanswered questions

The examples leave substantive questions open.  The retinal studies do not establish which whole-eye arrangement performs best across the relevant tasks.  The appendix microscopy paper motivates a protective-reservoir hypothesis without completing its clinical test or a lifetime cost comparison.  The magnitude and distribution of pelvic costs require broader evidence than one movement study.

A creationist reconstruction also owes more than an explanation of present usefulness.  Where present defects are attributed to later damage, the proposed earlier state and route of change need independent support.  Where a critic treats an awkward arrangement as decisive evidence of ancestry, the comparative and developmental argument must be supplied.  These are distinct research tasks; neither can be completed by relabelling the other.

The resulting verdict is limited but firm.  Appearance alone is a poor guide to biological performance, and imperfection alone does not exclude intelligence.  Real functions and real defects should both be investigated.  The strongest case is the one that states its prediction, accepts a possible failure, and follows the evidence when a familiar explanation does fail.

Sources