Historical Science

Do evolution & design

 use scientific method?

 ( Part 3 of 3 ) 

by Craig Rusbult, Ph.D.

This page builds on the foundation of Historical Science: Is it scientific? (Part 1) by applying general principles to specific types of historical theories that propose naturalistic evolution (in Section 7G) and intelligent design (in Section 7C).


7G. Can evolution be scientific? 

      Some critics of biological evolution (bio-E) claim it is unreliable and unscientific because:  major bio-E cannot be observed;  E-theory does not predict;  E-theory cannot be falsified.  Are these criticisms justified?
      OBSERVATIONS:  Bio-E is a historical science, and the limitations of historical data provide reasons to be cautious about conclusions.  But scientists have developed methods for reducing the practical impact of data limitations, so (although we should critically examine these methods) a historical science can be scientific.
      PREDICTIONS:  In historical science, prediction is not necessary.  Instead, in bio-E the goal is to retroductively explain by constructing a historical scenario (connecting the initial and final situations of a historical episode) that is consistent with E-theory.
      FALSIFICATION:  If scientists retain methodological naturalism (MN) and assume that "everything in the history of nature happened by natural process," they cannot falsify a theory of "bio-E somewhere in the universe."  But even without MN, it would be difficult to falsify bio-E due to its explanatory flexibility:  bio-E has powerful explanatory resources (linked genes, changes of function, developmental genes, statistical bottlenecks, long periods of time,...) that — combined with the creative imagination of a scientist — provide lots of flexibility for explaining almost anything, and it is difficult to distinguish this hyperflexibility from genuine plausibility.  In addition, appeals to "future science" make bio-E even more difficult to falsify.  { Yes, bio-E could be falsified by evidence for out-of-order origins, such as rabbits before trilobites, but this is irrelevant if E is wrong in ways that are less obvious. }

      Two Kinds of Science: Operations and Historical
      The Scientific Method does not exist, because no single method is used in the same way by all scientists at all times.  But scientists use scientific methods that are variations on a few basic themes.
      Some variations are due to differences between operations science (to study the current operation of nature, what is happening now) and historical science (to study the previous operation of nature, what happened in the past).  Both types of science are similar in most important ways, especially in their use of scientific logic, but there are minor differences.
      Although repeatable controlled experiments (with the situation set up by humans) can be done in operations science, this is not possible for historical events.  But this limitation has inspired scientists to develop methods that reduce the practical impact of this limitation.  One way to reduce the impact is to use repeatable uncontrolled experiments (with the situation set up by nature) and look for consistencies.  These consistencies let scientists develop reliable theories that usually are related to (and are consistent with) theories in operations science.
      A common misconcpetion about historical science involves the timing of inference.  A theory-based inference about "what happens if this theory is true" can be logically valid even if it is made after an event has occurred, or after observations are known.  In historical science, the goal is to describe and explain what did happen, not predict what will happen.

      Is it bad to be a theory?
      Would it be accurate to summarize a claim (which is disputed above) that "evolution is unreliable because it is historical" by proclaiming that "evolution is just a theory"?  No, because in science, "theory" does not mean "unreliable."  In fact, I don't think the word "theory" should have any special significance, so in my model of Integrated Scientific Method a theory can have a high or low plausibility, a simple or complex structure, a narrow or broad domain, and it can be descriptive or explanatory.  Therefore, simply calling a proposal a "theory" says nothing about its quality or characteristics.

      Can we predict the designs of God?
      Do "imperfect adaptations" provide evidence for bio-E?  "God surely would not have used a collection of parts generally fashioned for other purposes. ... Odd arrangements and funny solutions are the proof of evolution."  Here, in The Panda's Thumb, Stephen Jay Gould asserted that God "surely would not," as if he knew what God would have done.  But in the Bible, it seems that God does not want a history that appears optimal, or theistic action that appears obvious.  { Why isn't God more obvious? Can we prove God? }
      The possibility of design-action that is not "optimal and obvious" is opposed by advocates of bio-E, who want a competitor with predictions that are different from E-theory and easy to falsify.  Young-earth science meets both qualifications, and with independent creation (either young-earth or old-earth) we might expect designs to appear independent and optimal.  By contrast, old-earth creation by genetic modification (oeCmodification) predicts that new species will appear to be modifications of old species, because this is what they are.

      Distinguishing between Similar Theories
      Can we distinguish between oeCmodification and evolution?  Maybe.  With detailed data — such as lab reports (for physiology, structure, genome-DNA,...) for all organisms during a period of change — it would be easy.  But with the data we actually have, it is more difficult.
      Because oeCmodification includes two mechanisms, continual natural-appearing evolution and occasional miraculous-appearing macromutations, oeCmodification is consistent with most evidence for evolution.  The major differences are that oeCmodification raises questions (re: irreducible complexity, rates of change, extrapolating from micro-E to macro-E,...) about important details of bio-E.  When we compare oeCmodification with bio-E that is totally natural, the plausibility of bio-E is usually over-estimated, as explained in Principles for a Logical Evaluation of Evolution.
      As a theory, does oeCmodification have a responsibility to be "more different"?  No.  A "high contrast" with other theories is not a requirement for a scientific theory.  For example, in most situations the predictions of Newton's classical mechanics and Einstein's special relativity are almost identical.  But we don't demand that, if we are to take Einstein seriously, his theory must be modified to make it differ from Newton's theory in other ways, for our convenience, so we can more easily distinguish between his theory and Newton's theory, using data that is easy to collect and analyze.  Instead, it's more important to use the criterion of perceived accuracy by asking, "Based on empirical evidence and logical evaluation, does a theory seem to match the way the world really is?"

Later, there is more about Evolutionary Predictions and Retroductions.

a transition from 7G to 7C, from evolution to design:
In a theory of old-earth design, a scientist accepts the current consensus of scientists about an old earth, but rejects the current consensus about naturalistic evolution.  Is this logically inconsistent?  This question inspired another page that, in the four paragraphs below, explains the difference between proponents of three theories (young-earth creation, old-earth nondesign, and old-earth design) when we ask:  What are the capabilities of historical science, for questions about age and design?

      When we ask questions about age, most young-earth flood geologists are super-skeptical about the ability of historical science (as in geology or astronomy) to reach any reliable scientific conclusions about history.  They ask, "Were you there?", and declare that a "no" means "therefore you can't know much about ancient history."  [the two old-earth views disagree]

      Similarly, when we ask questions about design, most proponents of non-design theories (proposing a totally natural origin of life and development of complexity) are super-skeptical about the ability of historical science to determine anything about historical design-directed action by an agent, at least if the agent and action might have been supernatural.
      By contrast, design theorists are confident that scientists have developed, and will continue improving, scientific methods (which are based on a logical evaluation of observable evidence) to cope with the challenges of scientifically distinguishing between design and nondesign.  As with all science, for design questions we cannot obtain proof, but we can develop a rationally justified confidence about "a good way to bet."

      Old-earth design theorists say that science can do more (when thinking about age-questions) than is claimed by skeptical young-earth flood geologists, and also that science can do more (when thinking about design-questions) than is claimed by opponents of design theories.  Thus, there are differences between the claims of old-earth design (about what is science, re: age and design) and the claims (about what is not science) that are often made by proponents of young-earth and non-design theories.

Can a theory of design be scientific?  Many arguments against design theories are just arguments against any theory in historical science (in areas like biology, paleontology, geology, or astronomy) about the history of nature.  Therefore, to establish a foundation for deciding "Can design be scientific?" (in Section 7C below), I have asked (in Section 7G above) the analogous question for evolution.

For an overview of design theory (what it is and isn't, what it does and doesn't claim, and why it can be scientific) that will help you understand the following section, you can read two pages — what design is and design in science — that explain ideas in 7A-7B and give you a different perspective on 7C.

      7C. Can intelligent design be scientific?

      The most common type of Closed Science is defined by methodological naturalism (MN), a proposal to restrict the freedom of scientists by requiring that they include only natural causes in their theories.  This section examines arguments — practical, methodological, metaphysical, and trivial — for Open Science (that rejects MN and allows theories of design) and Closed Science.  The basic question is simple:  In scientific thinking and education, do we want to give a higher priority to logic or naturalism?

      PRACTICAL Questions
      In a search for truth about nature, a design theory can be useful because it might be true.  If design really was involved in the origin of a feature, but we ignore this possibility (by refusing to consider evidence for design), a false conclusion is unavoidable.
      When design encourages critical thinking about non-design, this can improve the accuracy of our evaluations (for current non-design theories) and our speculations (about future theories).  A theory of design can stimulate creative thinking and productive action by proponents of non-design (who are motivated to defend and improve their theories, and find new experimental support) and design.  In this way, and others, theories of design can help us improve our understanding of nature.
      Design would have little impact on the overall productivity of science, since most areas are not affected.  And many scientists will continue their non-design research even if they hear a design claim that "maybe there is no non-design explanation."  Proponents of design want research about non-design to continue, because the goal is to find truth;  they want to supplement non-design research, not replace it.  { If your keys are not in the kitchen, can you find them by searching in the kitchen?  Perseverance and Flexibility }
      Current scientific customs (with most scientists accepting MN) are not necessarily optimal, because traditional customs are decided by people, and "what now is" does not determine "what should be in the future."  MN is a choice, and we can ask, "Is MN always wise?"  If we're being practical, should we view science as an artificial game with rules (which exclude design) or a real-life activity with goals?  { The important difference between games and reality is illustrated by a Strong-Man Contest. }
      What about past failures of claims for design?  A current design theory should be judged on its own merit, not the weakness of superficially similar theories in the past.

      METHODOLOGICAL Questions
      Typically, questions about methods are in four areas: history, agency, observability, and falsification.
      HISTORY:  The methods used in design are similar to methods in other historical sciences, such as astronomy, geology, archaeology, and evolutionary biology.  Because most arguments against the scientific status of design are also arguments against the scientific status of any historical science, I suggest that you read the first half of Section 7G — Can a theory of evolution be scientific? — and then read the rest of this section.
      AGENTS AND PREDICTION:  When "what happens" depends on the actions of an agent, this introduces an element of unpredictability.  But in a historical situation with agent action, a scientist (in psychology, sociology, anthropology, archaeology, or forensics, and maybe in origins) only has to determine what did occur, not predict what will occur.  In this situation, the best explanation for "what did occur" is an agency theory of design (proposing "agent action"), not a mechanistic theory of non-design (proposing "only undirected natural process").  In most ways, historical theories of design and evolution are analogous, but a major difference is that intelligent design is an agency theory, while naturalistic evolution is a mechanistic theory.  Which type of theory is better?  The answer depends on what actually happened in history, if we define a "better" theory as one that more closely corresponds with the truth.  If agent-action did occur, an agency theory is better.  If there was no agency-action, a non-agency (mechanistic) theory is better.
      OBSERVABILITY:  Modern scientists often infer the existence of an unobservable cause (an electron, idea,...) due to the observable effects it produces.  Similarly, if we observe "signs of design" we can infer design-directed action, even if the agent and action were not observed.
      FALSIFICATION:  Due to the mutually exclusive relationship between non-design and design (when they're carefully defined it's either one or the other), the status of design can be increased or decreased by empirical evidence (by observations), so a theory of design is empirically responsive and is testable.  Based on a logical evaluation of observations, we can develop a logically justified confidence in the falsity or truth of a design theory, concluding that a design theory is probably true (if all non-design theories seem highly implausible) or is probably false (if one non-design theory seems highly plausible).

      The Bottom Line:  Inferences to design can be scientifically justified.
      For example, in Section 7A the "prime number" design theory (re: a sequence of 2, 3, 5, 7, 11, 13, 17,...) is a result of scientific thinking:  You observe a signal, creatively construct and critically evaluate theories, and conclude that design-directed action is more plausible than undirected natural process.  Scientists propose design to explain a wide variety of features such as bird nests, ant hills, prey killed by a predator, a human murder victim, paintings on a cave wall, metal satellites in orbit, and faces on Mount Rushmore.  In many areas of science, a logical inference that "design-directed action did occur" can be scientifically justified.
      For most critics of design, the main concerns are metaphysical, not methodological.

      METAPHYSICAL Questions
      MIRACLES:  Does design violate methodological naturalism?  Maybe.  A basic design theory, claiming only that a feature was produced by design-action (which could be either natural design-and-construction or supernatural design-and-creation), does not require miracles, but it does allow miracles.  /  And science is compatible with occasional miracles, if the universe usually operates according to normal natural patterns.
      LIMITED CLAIMS:  In any area (radioastronomy, homicide, origins,...) an inquiry about design is a two-stage process:  first ask "Was there design-directed action?" and then investigate the details.  We should evaluate a design theory based on what it does claim (that design occurred) instead of what it does not claim (that it can explain the "how, when, why, and who" details of design-and-production).
      DESIGN AND CREATION are not the same.  A design theory can be supplemented with details (about the designer's identity and actions) to form a variety of theories about supernatural creation or natural non-creation.  A design theory does not claim that we can distinguish between "creation design" and "non-creation design" by scientific analysis, it just claims that "design did occur."  /  analogy:  A scientific conclusion that "it happened by a natural process which appeared to be undirected" does not require a metaphysical conclusion that "it happened without God."  Similarly, a scientific conclusion that "it was the result of design-directed action" does not require a metaphysical conclusion that "God did it."
      METAPHYSICAL MOTIVATIONS:  Even if a theory of design (or non-design) is motivated by a desire to show that "God did it" (or "God didn't do it"), this motivation should be irrelevant during theory evaluation.  A scientific evaluation should be based only on scientific evidence and logic;  our evaluations of a theory should not be influenced by our suspicions about the motives of scientists who are proposing and defending the theory.

      TRIVIAL Questions
      Some arguments against design seem impressive until you think about them, and then they seem trivial.
      For example, a claim that "in natural science, natural phenomena and natural history should be explained by natural causes" is trivial.  It is just faulty circular logic (assuming "science is natural" to conclude "science is natural") camouflaged with verbal ambiguity (using "natural" to mean "pertaining to nature" and also "normal-appearing").
      It is also trivial to view science as an artificial game with rules (which exclude design) rather than a real-life activity with goals.

      important questions:
      Should we define the main goal of science as a search for NATURAL explanations, or a search for LOGICAL explanations?  Of course, when we ask "Should science be logical?", everyone agrees that YES is the answer.  But disagreements occur when we ask: If there is a conflict between logical and natural, which criterion should have the higher priority?  Should we be forced (by methodological naturalism) to accept a "scientific conclusion" that is less logical, simply because it is natural?

      tough questions:
      In Section 7B the concepts of proof (which is impossible in science) and rationally justified confidence (which is the practical goal in science) are illustrated by thinking about how we can logically evaluate a design theory for the origin of life.  What would make this design theory unscientific?  a claim that a natural formation of life is extremely improbable?  a perception (by others) that this claim implies a non-natural cause?  proposing a non-natural cause?  Is there any limit to the severity of criticism before a design theory becomes unscientific?  If severe criticism is accompanied by a proposal for a new natural theory, does this make it scientific?  Can we admit that "we are far from finding the answer," but not "maybe there is no natural answer"?  Or, consistent with the restrictions of Closed Science, should we control the thinking of scientists by removing their freedom to think that "maybe..."?

      my conclusion:
      Some arguments against design are trivial, while others (especially those about practical effects and methodology) are more worthy, if only because they stimulate productive thinking and discussion.  But even though some arguments for Closed Science may seem strong initially, I think the counter-arguments are stronger and more logical, and the closer we examine Open Science, the better it looks.

      a note to the reader:  More than any other section, 7C has been diminished in "conceptual content" during its condensation into the medium-short version you've been reading.  To make it easier for you to explore the arguments and counter-arguments more thoroughly, I've collected three versions of Section 7C (medium-short [it's what you've been reading], medium-long, and long) into a 7C-page.


A. Illustrative Examples (with Keys and Games) for Section 7C are in another page that contains two sections:
Perseverance and Flexibility
Is science a game with rules?

The following excerpts — from the original "long version" of Section 7G (from 1998) — offer insights that I think you'll find interesting and useful:

      Evolutionary Predictions and Retroductions
      PREDICTIONS:  In a historical science it is difficult to make predictions that are both precise (with specific details) and accurate, especially over long periods of time, due to complexity, sensitivity, and randomness:  evolutionary contexts and causal factors are complex;  outcomes are sensitive to small variations in the initial conditions;  all bio-E processes involve some randomness, and some factors (like mutation and genetic drift) are extremely random.  These elements combine to produce historical contingency:  if 10 situations with similar initial conditions were "allowed to run" 10 times, there would be 10 different results.  But even though E-theory could not precisely predict these results, it claims the ability to retroductively explain each already-known result if scientists can construct a historical scenario (connecting the initial and final historical situations) that is consistent with the causal mechanisms of E-theory.  /  The goals of scientists, during their studies of bio-E, may be to reconstruct past contexts, to estimate the rates of evolutionary change or the importance of various causal factors (selection, drift, isolation,...), or to build models for large-scale and/or long-term changes.  { The rest of this page explores the logical process of constructing historical scenarios based on evolutionary theory. }

      Evolutionary theory can be used to propose a theory-based scenario for a particular historical period that spans time in a series of historical situations.  An initial situation, analogous to the initial conditions for an experimental system in a laboratory experiment, includes the initial environments and characteristics of one or more species.  Intermediate situations and a final situation are similarly defined.  The entire episode involves a sequence of situations: initial, intermediate(s), and final.
      RETRODUCTIONS:  In contrast with deductive logic that predicts by asking "If this is the initial situation and theory, then what will be the observations for the final situation?", retroductive logic asks a reversed question in the past tense, "These are the observations, so what could the scenario (composed of situations and theories) have been?"  Historical retroduction is a creative-and-critical thinking strategy, with imagination guided by deduction, whose goal is to generate a historical scenario, constructed by combining situations with theory, that would produce the known observations for a historical period.  During retroduction a scientist can adjust either of the sources, situational or theoretical, that are used to construct a scenario.
      SCENARIOS:  For a period in the history of nature, evolutionary theorists try to construct a sequential scenario -- for the environments and characteristics of species in initial, intermediate, and final situations -- that is consistent with the causal mechanisms of E-theory.  Part of the challenge is to decide how to use available observations to reconstruct, as completely and accurately as possible, the details of situations at various times.  And because the Modern Synthesis of E-theory contains a variety of mechanisms (for producing and expressing genetic variability, and for changing the gene frequencies in a population), scientists also must decide which causal mechanism(s) to propose for the "changes in situation" they have proposed.  Due to the flexibility in proposing situations and mechanisms, for one episode it may be possible to construct a number of competitive scenarios, all consistent with the core concepts of E-theory.  Each scenario can be compared with available data, to check for degree of agreement and, when necessary, to make adjustments in the components (the postulated situations, mechanisms,...) used in constructing the scenario.
      GOALS:  In analyzing an episode, the primary goal(s) of scientists might be descriptive and/or theoretical.  /  descriptive goals: Scientists can use observations and theory-based analysis in an attempt to reconstruct the characteristics of a past situation in a way that is more detailed and accurate.  /  theoretical goals: Scientists can analyze the observation-based situations they have proposed, in an effort to estimate the rates of change for genotypes and the resulting phenotypes, or the roles played by various factors (natural selection, random drift, geographic isolation,...) in a causal mechanism, or the importance of genes that control developmental pathways, or...
      LEVELS:  An evolutionary study can involve different levels of time, change, and space.  /  Levels of Time:  Scientists can try to construct scenarios for periods that are short or long.  /  Levels of Change:  The changes being studied can be small (within one species or a group of closely related species) or large (involving major changes between widely divergent species).  /  Levels of Space:  Scientists can focus on a small area, or they can do a large-scale biogeographical analysis that integrates the analyses already done for smaller areas.  /  A wide variety of evolutionary research projects are possible, involving analysis and synthesis for various combinations of time, change, and space.  /  { Critics of evolution question the logical validity of extrapolating from small-scale studies of micro-E and minor macro-E (which have solid empirical support) to large-scale theories of Total Macro-E (which have far less empirical support). }

      OBSERVATIONS:  Most information about evolution comes from field studies of past events, with data about species preserved in fossils.  But scientists can also gather information in other ways, including detailed examinations of different modern-day species (to compare their physiological, structural, and behavioral characteristics, and the composition of their proteins and genes) and current events (such as observing natural selection in bacteria or mutations in fruitflies, analyzing current biogeographical patterns, or exploring an ecosystem to learn more about the interactions of species with each other and with their environment).  Usually scientists are working with data they already know, but not always.  Sometimes predictions can be made about data that will be observed in the future, whether this data is from events in the distant past (like old fossils that are newly discovered) or in the future (such as lab data about gene sequences, to be used for comparative analysis).  The inherent limitations of historical data are that for the major events in macroevolutionary history there can be no direct observations, and there can be no rigorously controlled experiments with reproducible data.

      PREDICTIONS:  In most historical sciences, including evolution, it is difficult to make predictions that are both precise (with specific details) and accurate.  Generally, accuracy decreases when precision increases, and when time-duration increases the accuracy and precision both decrease.  In making predictions the main obstacles to achieving "precision with accuracy" are complexity, sensitivity, and randomness:  evolutionary situations and causal factors are complex;  outcomes are sensitive to small variations in initial conditions;  all processes involve some randomness, and some factors (like mutation and genetic drift) are extremely random.  These three elements combine to produce historical contingency:  if similar initial situations were "allowed to run" 10 times, there would be 10 different results.  But even though E-theory could not precisely predict the results, it claims the ability to retroductively explain each result if scientists can construct a scenario (connecting the initial and final situations) that is consistent with the causal mechanisms of E-theory.  And despite a typical divergence in results for long-term runs, sometimes there is convergence such as similar environmental niches being filled by species with similar characteristics.
      Some of the uncertainties in evolutionary prediction (re: complexity, sensitivity, randomness) are roughly analogous to those in weather prediction.  And relationships for time duration are similar;  as time increases, accuracy decreases.  The scientific principles of weather forecasting allow precise short-term predictions, such as what will happen in the next hour in a specific location.  Consistencies in seasonal weather patterns allow nonprecise long-term predictions (in Wisconsin it will be cold with occasional snow in January, and hot with occasional thunderstorms in July) and probabilistic predictions such as the total snowfall in January or the probability of snow on January 16.  But it would be impossible, a month in advance, to make accurate long-term predictions about the precise times when snowfalling or thunderstorming will occur.
      When these phenomena do occur, however, they will be consistent with theories of meteorological science (which are applications, for complex situations, of conventional theories in physics,...) and they can be retroductively explained using these theories.  Similarly, evolutionary theories can be used to make precise short-term predictions (sometimes), nonprecise long-term predictions, probabilistic predictions, and (most important for historical science) theory-based retroductions.

This website for Whole-Person Education has TWO KINDS OF LINKS:
an ITALICIZED LINK keeps you inside a page, moving you to another part of it, and
 a NON-ITALICIZED LINK opens another page.  Both keep everything inside this window, 
so your browser's BACK-button will always take you back to where you were.

Here are other related pages:

Part 1— Historical Science: Is it scientific?
Part 2 — Historical Science for Age-Questions
Part 3 is the page you've been reading.

Historical Science (pages by other authors):
Can it be scientific and can it help us reach
reliable conclusions about AGE and DESIGN?

Design for the Origin of Life?
 Closed Science and Open Science 
What is a theory of intelligent design?
Can a theory of design be scientific?
Can design be supported?  proved?
What is design?  Is design scientific?

Can a theory of evolution be scientific? (long version)

Can a theory of design be scientific? (long version)

other pages about Origins Questions by Craig Rusbult

This page is

Copyright © 2001 by Craig Rusbult, all rights reserved