Laws of Nature, Causation, and Information

"Laws of nature know no exceptions." — Werner Gitt, Journal of Creation, 2009¹

Summary

Every law of physics describes a regularity:  all electrons have the same charge, everywhere and at every time anyone has looked.  This article asks what such a law is, whether it needs a lawgiver, why it is reasonable to expect it to hold tomorrow, and what "information" means when creationists say that matter cannot produce it.

1. Philosophers give four accounts of what a law is. On the regularity account of David Hume and David Lewis, a law is only a summary of what happens: "a theorem (or axiom) in each of the true deductive systems that achieves a best combination of simplicity and strength."² On the necessitation account of David Armstrong, Fred Dretske and Michael Tooley, a law is a real relation between properties that makes the regularity hold. On the powers account, things have built-in dispositions and the laws describe them. On the theistic account of Robert Boyle, Isaac Newton and the Bible, the laws are God's "ordinances", his regular way of governing what he made.³
2. The lawgiver argument is not a pun on the word "law." At full strength, in Richard Swinburne and John Foster, it argues from the regularity itself: the uniformity of nature is a striking fact, theism explains it by a single agent's constant will, and every rival view must take it as a brute fact. The best naturalist replies are that a Humean law needs no explanation, because it is only the pattern, and that every view ends in some brute fact. Both replies are coherent. The theist's advantage is real but modest: theism expects an orderly world, and naturalism merely permits one.
3. Induction has no proof on any view. Hume showed that every argument from experience assumes "that the future will resemble the past."⁴ Theism gives a reason to expect uniformity, God's faithfulness, and so does a necessitation account of laws. A pure regularity account gives none, as Armstrong argued. No view proves that tomorrow will resemble today.
4. Emergence and reduction matter mainly for the mind. Weak emergence, where new behaviour follows from the parts but is hard to predict, is everywhere in physics and biology and is no problem for naturalism. Strong emergence, where a property cannot even in principle be deduced from the parts, has one serious candidate: consciousness. That question belongs to Mind, Reason, and Consciousness.
5. "Information" has at least six meanings, and the creation debate often runs them together. Shannon information and Kolmogorov complexity ignore meaning, and random noise maximizes both, so nature produces them easily. Functional information, as Jack Szostak and Robert Hazen define it, measures how rare it is to do a named job. Werner Gitt's "universal information" includes meaning and an intended purpose. Physicists' information is a physical quantity with a measurable energy cost.
6. Gitt's "laws of information" fail as laws of nature. Gitt defines information so that it already contains a sender's goal, which makes "information needs an intelligent sender" true by definition and unusable on DNA without assuming the answer. If purpose is read as biological function, as Gitt's own examples do, his claim that "There can be no increase in information" meets observed counterexamples.⁵ His argument that information cannot be physical because it lacks an SI base unit is wrong about the SI.
7. One version survives. Stephen Meyer argues that intelligence is "the only known cause of specified information (at least starting from a nonbiological source)."⁶ That is an inference about the origin of life, not a law, and the origin of the genetic code remains, in a mainstream review's words, "perhaps, the most formidable problem of all evolutionary biology."⁷

Verdict.  The existence of an orderly, law-governed world is better expected on theism than on naturalism, and the inductive confidence science runs on has a ground on theism that a pure regularity view lacks.  Neither point proves God.  The information argument should be stated as Meyer states it, about the origin of codes from non-living matter, and never as a law of nature that mutations break.

Part 1: What is a law of nature?

A law of nature states that something always happens:  any two masses attract, every electron carries the same charge.  Philosophers agree on that much.  They disagree about what makes such a statement true.  There are four main answers, and each has a different view of causation, of what a law can explain, and of whether a law needs a source.

The regularity account: laws describe what happens

David Hume, writing in 1748, held that experience never shows us one event making another happen.  "When we look about us towards external objects, and consider the operation of causes, we are never able, in a single instance, to discover any power or necessary connexion".⁴  What we see is "the constant conjunction" of two kinds of event, and the idea of necessity comes from the habit of expecting one after the other.⁴  On this view a cause is a regular pattern, and a law is a statement of the pattern.

David Lewis gave the modern form in 1973.  He took it from Frank Ramsey:  "a contingent generalization is a law of nature if and only if it appears as a theorem (or axiom) in each of the true deductive systems that achieves a best combination of simplicity and strength."²  Imagine writing down everything that ever happens, then finding the shortest set of axioms that captures as much of it as possible.  The axioms of that best system are the laws.  Lewis drew the consequence himself:  "I doubt that laws of nature have as much of a special status as has been thought."²

On this account, a law does not make anything happen.  It is a summary of what does happen.  It needs no lawgiver, because there is nothing for a lawgiver to do.

The necessitation account: laws make the pattern hold

In 1977 Fred Dretske and Michael Tooley, and in 1983 David Armstrong, argued that a summary cannot be a law.  Dretske put the objection sharply:  "To say that a law is a universal truth having explanatory power is like saying that a chair is a breath of air used to seat people."⁸  The fact that all men are mortal does not explain why any one man is mortal.  It only restates it.

Their alternative is that a law is a relation between properties.  Dretske:  "laws are expressed by singular statements of fact describing the relationship between universal properties and magnitudes."⁸  Armstrong's version is the best known:  "Suppose it to be a law that Fs are Gs.  F-ness and G-ness are taken to be universals.  A certain relation, a relation of non-logical or contingent necessitation, holds between F-ness and G-ness.  This state of affairs may be symbolized as 'N(F,G)'."⁹  The relation N makes every F a G.  The regularity is the result, not the law.

The word "contingent" matters.  Armstrong's necessitation could have been otherwise.  So the account immediately raises a question the regularity view does not:  what makes N(F,G) hold rather than some other relation, or none?  Armstrong treats it as a basic fact.  John Foster's answer, taken up in Part 2, is that a will imposes it.

The powers account: laws describe what things are disposed to do

A third group, including Stephen Mumford and Alexander Bird, puts the source of order in things themselves.  An electron has a built-in power to repel other electrons, as salt has a power to dissolve.  Bird puts it this way:  "the dispositional essentialist view will regard the motor and cement of the universe as residing ultimately not in the laws themselves but rather in the dispositional nature of properties.  The laws are, in a sense, epiphenomenal."¹⁰  Mumford goes further and drops laws altogether:  "In my account, there is no such thing to be found in the world that should be called a law of nature.  But I do not deny that there is order and regularity that permits reliable inference.  The explanation I offer for such phenomena is internal and immanent, as opposed to external and transcendent."¹¹

On the powers view, things behave alike because they are alike.  That moves the question rather than ending it:  why do all electrons share exactly the same powers?

The theistic account: laws are God's regular way of acting

The founders of modern physics held a theistic account.  The Hebrew Bible speaks of "the ordinances of heaven and earth" that God has "appointed", and it ties their constancy to his covenant:  "If my covenant of day and night fails, if I have not appointed the ordinances of heaven and earth," then God would also cast off Israel.³  The Psalms say the same:  "Your laws remain to this day, for all things serve you."³  Here a law is a promise kept.

Isaac Newton closed his Principia in this vein.  "This most beautiful System of the Sun, Planets and Comets, could only proceed from the counsel and dominion of an intelligent and powerful being."¹²  "This Being governs all things, not as the soul of the world, but as Lord over all".¹²

Robert Boyle, Newton's older contemporary, was more careful about the word.  In 1686 he argued that matter cannot literally obey a law:  "to speak properly, a Law being but a Notional Rule of Acting according to the declar'd Will of a Superior, 'tis plain, that nothing but an Intellectual Being can be properly-capable of receiving and acting by a Law."¹³  For Boyle, "law of nature" is a figure of speech.  What is real is God's ordinary action, keeping up the powers he gave to matter.  His view is close to the powers account, with God as the one who sustains the powers.

Alvin Plantinga gives the account its modern form.  On Christian belief, God acts in the world all the time, and a miracle is a case where "he sometimes deviates from the usual way in which he treats the stuff he has made."¹⁴  A law of nature, on this view, describes that usual way.  It does not bind God, and a miracle is not a broken law but the same agent acting differently for a reason.

How the accounts bear on causation

The accounts carry different views of causation.  On the regularity view, to cause is to be followed regularly, and Hume's verdict stands:  no "necessary connexion" is ever observed.  On the necessitation and powers views, causes really produce their effects.  The premise "everything that begins to exist has a cause", central to the kalam argument, reads naturally on a production view.  On a regularity view it is harder even to state, because a cause is only something regularly followed by its effect.  That argument is assessed in The Cosmological Argument.

Part 2: Do laws need a lawgiver?

The weak version is a pun

The weak version runs:  there are laws, every law has a lawgiver, so nature has a lawgiver.  It fails, because "law" in physics is a metaphor taken from human government.  A summary of how electrons behave is not a command.  Sean Carroll, then a physicist at Caltech, presses exactly this point:  "it is sometimes suggested that the regularities we label “laws of nature” are inexplicable in the absence of some entity that ensures those laws are obeyed, as if the reliability of such laws implies the existence of a legislative body or a law-enforcement agency.  It is precisely this kind of intuitive and metaphorical reasoning that we should be suspicious of in this context."¹⁵  Robert Boyle, a devout Christian, agreed that the word is a figure of speech when applied to matter.¹³  So the argument cannot rest on the word.

The strong version argues from the regularity

The serious argument does not mention the word "law."  It starts from the fact that nature is uniform, and asks why.

Richard Swinburne of Oxford states it in terms any physicist will recognize:  "In the near past, as in the more remote past, every positron has continued to attract every electron with exactly the same force inversely proportional to the square of their distances apart.  There are innumerable other ways in which objects could have behaved, almost all of them such that the simplest extrapolation from their past behaviour would not have correctly predicted their subsequent behaviour."¹⁶  He then runs the argument through each account of law.  On the powers account, the question is why so many things have identical powers, and he answers with an analogy:  "Just as we would seek to explain all the coins of the realm having an identical pattern in terms of their origin from a common mould, or all of many pictures having a common style in terms of their being painted by the same painter, so we should seek to explain all physical objects having the same powers in terms of their deriving them from a common source."¹⁶  His conclusion is a judgement of probability, not a proof:  on all three accounts "it is very improbable that there would be in a Godless universe laws of nature sufficiently simple for rational beings to extrapolate from past to future with normal success."¹⁶

John Foster, also of Oxford, gives the most careful version.  He first argues that past regularity calls for explanation.  To suppose it is "no more than a pure coincidence" is coherent, he grants, but "the situation postulated by H would surely be a very strange one—a situation whose obtaining, if we knew of it, would warrant huge surprise."¹⁷  The natural explanation is a law in the necessitation sense.  But a necessitation relation is itself something that needs a source:  "For how has it come about that things are constrained to behave in certain regular ways unless there is something which has imposed, or continuously imposes, these forms of constraint—something which is causally responsible for (as it were) laying out the nomological tramlines?"¹⁷  His conclusion covers both cases:  "If nomological explanations are excluded, there is a strong case for explaining the regularities by appeal to the agency of God.  And if nomological explanations are accepted, there is a strong case for concluding that it is God’s imposing of the regularities that creates the relevant laws.  Either way, there is a strong case for accepting the existence of God."¹⁷

The naturalist replies

The Humean reply:  a description needs no author.  Helen Beebee gives the cleanest statement.  "For the anti-Humean, laws (unlike accidentally true generalizations) do something—they govern what goes on in the universe—and they therefore require some sort of ontological basis (N, for instance) that gives them this ability.  Humeans, on the other hand, do not require laws to 'do' anything."¹⁸  On her view the direction of explanation runs from the facts to the laws:  "For the Humean, since the laws are descriptive, what the laws are depends on what the facts are—including future facts."¹⁸  If laws are only the best summary of the facts, there is no legislation, and so no legislator.  Barry Loewer defends the same picture:  "Humeans claim that there is no fundamental necessity in nature connecting spatio-temporally non-overlapping events in non-overlapping portions of space-time."¹⁹

The brute-fact reply:  every view stops somewhere.  Carroll grants more than many atheists do:  "the presence of regularities such as the laws of nature is itself something we might want to explain".  But he denies that it must be explained by anything beyond the world:  "there is no logical or empirical reason why such an entity must exist; the universe can just be."¹⁵  And he argues that a creator does not end the chain either:  "invoking a creator does not provide us any escape from the need to posit something that simply exists because it does, without further reasons to which we can appeal."¹⁵

The anthropic reply.  A naturalist can add that observers could only exist in an orderly world, so it is no surprise that observers find order.  This is true as far as it goes, but it explains only as much order as observers need.  Living things need stable chemistry here and now.  They do not need atoms in remote galaxies to obey the same laws, yet those atoms do.  In 2022 Michael Murphy and colleagues measured the fine-structure constant, which sets the strength of electromagnetism, in a gas cloud whose light is stretched by redshift to more than twice its original wavelength.  It matched the laboratory value to about one part per million.²⁰  The anthropic reply covers local order, not universal order.  How far a world of observers needs order is the subject of the Boltzmann-brain problem, treated in Entropy and the Arrow of Time.

Where each side is strongest

The theist is strongest against the pure regularity view.  On Lewis's account the uniformity of nature is not explained at all.  It is the pattern, and nothing makes it hold.  Foster's coin example shows why that is hard to accept:  a coin that lands heads a hundred times running is evidence of a bias, and a universe that has behaved uniformly for billions of years is evidence of something that keeps it so.¹⁷  The regularity view also leaves induction without a reason, as Part 3 shows.

The naturalist is strongest on the brute-fact point.  Every view ends somewhere.  The theist says the order comes from God's will.  Carroll asks why God, rather than the universe, should be the stopping point.  The theist's answer is that a personal explanation ends in an agent's purpose, which explains why the pattern is simple and orderly in a way a bare fact does not.  Whether that answer is good depends on whether God is a better stopping point than the laws, the question of The Cosmological Argument.

Neither side can put a number on it.  Swinburne's argument is a claim about probability:  an orderly universe is much more likely if God exists than if not.  That would be decisive if the probabilities could be computed.  They cannot.  There is no agreed way to count the "innumerable other ways in which objects could have behaved", so there is no agreed prior probability for an orderly universe without God.  The argument therefore rests on a judgement that most people, theist or not, share:  a uniform world is not what one would expect from nothing in particular.

Verdict on the lawgiver.  The argument from the existence of laws is not a pun when stated as Swinburne and Foster state it.  It shows that theism expects what we observe, a uniform and simple order, while naturalism must accept that order as a brute fact.  That is real evidence for theism, of modest strength.  It is not a proof, because a naturalist can coherently stop at the laws, as Beebee, Loewer and Carroll do.

Part 3: Induction and the uniformity of nature

Hume's problem

Every scientific prediction assumes that nature will go on behaving as it has.  David Hume saw that this assumption cannot be proved from experience:  "For all inferences from experience suppose, as their foundation, that the future will resemble the past".  Any argument from experience for that principle would assume it, so "It is impossible, therefore, that any arguments from experience can prove this resemblance of the past to the future".⁴

Bertrand Russell gave the problem its most famous picture in 1912.  "The man who has fed the chicken every day throughout its life at last wrings its neck instead, showing that more refined views as to the uniformity of nature would have been useful to the chicken."²¹  Russell agreed with Hume that the principle of induction "is equally incapable of being proved by an appeal to experience."²¹  Nelson Goodman added a second problem in 1955.  He defined "grue" as applying to things examined before a time t "just in case they are green but to other things just in case they are blue."  Every green emerald seen so far then supports the prediction that future emeralds will be grue exactly as well as the prediction that they will be green:  the two predictions "are alike confirmed by evidence statements describing the same observations."²²  So even "nature is uniform" is not enough.  One also needs to know which patterns are the real ones.

No one has solved either problem.  Science works, and everyone, theist or atheist, relies on induction in daily life.  The question here is narrower:  which view of the world gives a reason to expect induction to work?

The regularity account gives no reason

On the regularity account, a law is only the pattern of what happens, past and future together.  Armstrong drew the consequence.  "Suppose, however, that laws of nature are conceived of as mere Humean uniformities.  Then, I contend, this explanation of the necessity of the rationality of induction must fail."  The law "is nothing more than the conjunction of its observed manifestations with its unobserved manifestations.  Such a law is not an explanation of the observations."⁹  If nothing explains the pattern, nothing in the past gives a reason for the future.  Swinburne makes the same point about the Humean view:  "It is only if there is a common explanatory cause of the behaviour of objects that there is any reason to suppose that they will behave in the same way."¹⁶

Necessitation and theism give a reason, but not a proof

The necessitation account offers a way out.  If a real relation between properties makes every F a G, then the observed Fs are explained by that relation, and an inference to the best explanation can carry the relation forward to unobserved cases.  Tooley argued that his account gives "a solution to the epistemological problem for laws."²³

John Foster built his whole case on this point.  He argues that induction is justified only as an inference to laws:  "we can be justified in postulating laws as a way of explaining the regularities which have held good in our past experience, and can then appeal to the presence of these laws to justify the belief that the regularities will, or will on certain conditions, continue to hold for the unexamined cases."¹⁷  This only works if laws are "not mere factual regularities, but forms of natural necessity."¹⁷  And, as Part 2 showed, Foster then argues that natural necessity itself needs a source, which he identifies with God.

Theism gives the oldest reason.  The Bible grounds the order of nature in a promise:  "While the earth remains, seed time and harvest, and cold and heat, and summer and winter, and day and night will not cease."³  If a faithful God upholds "all things by the word of his power", the future will resemble the past because he keeps it so.³  The mathematician and philosopher Alfred North Whitehead, who was not writing as an apologist, argued in 1925 that this belief came first historically.  He held that "the faith in the possibility of science, generated antecedently to the development of modern scientific theory, is an unconscious derivative from medieval theology."²⁴  He was careful about the limits of the claim:  "I am not arguing that the European trust in the scrutability of nature was logically justified even by its own theology.  My only point is to understand how it arose."²⁴

The historian of Chinese science Joseph Needham made a related point.  Chinese thinkers saw order in nature, but "it was not an order ordained by a rational personal being, and hence there was no conviction that rational personal beings would be able to spell out in their lesser earthly languages the divine code of laws which he had decreed aforetime."²⁵  Needham and Whitehead show where the expectation of lawful order came from in Europe.  Neither shows that the expectation is true because of its source, and Whitehead says so.

What the theistic answer does and does not do

The theistic answer is a reason to expect uniformity, not a proof of it.  Two limits should be stated plainly.

First, it does not escape Hume.  A believer's confidence that God will keep his promise rests on God's character, known from revelation and experience, and any inference from past faithfulness to future faithfulness is itself inductive.  The theist has a ground for uniformity, but no non-circular proof that the ground holds.

Second, theism predicts regularity with exceptions.  On the theistic account a miracle is God acting differently from his usual way, so theism does not promise that the future will resemble the past in every case.  It promises enough regularity for creatures to live and reason, which is what Genesis 8:22 promises.

Even with those limits, the comparison is real.  On a pure regularity view, the success of induction so far is a fact with no explanation, and its future success is a matter of luck.  On the necessitation view it is explained by relations that are themselves unexplained.  On theism it is explained by an agent's purpose to make a world creatures can understand.  Each view ends somewhere.  Theism ends with a reason the order exists at all.

Part 4: Emergence and reduction

Reductionism says that everything that happens is fixed by the laws governing the smallest parts.  Emergence says that wholes have properties their parts lack.  The two can be combined, and the useful question is which kind of emergence is meant.

David Chalmers draws the line most clearly.  A phenomenon is weakly emergent when it "arises from the low-level domain, but truths concerning that phenomenon are unexpected given the principles governing the low-level domain."  It is strongly emergent when "truths concerning that phenomenon are not deducible even in principle from truths in the low-level domain."²⁶

Weak emergence is everywhere and threatens no one.  The physicist Philip Anderson made it famous in 1972.  He granted that nearly all scientists accept reduction, then argued that the reverse fails:  "The constructionist hypothesis breaks down when confronted with the twin difficulties of scale and complexity."  The behaviour of large aggregates cannot be understood by extrapolating from a few particles;  "Instead, at each level of complexity entirely new properties appear".²⁷  Superconductivity, the shape of a snowflake and the behaviour of an ant colony all follow from their parts, yet none could be predicted in practice from particle physics.  Chalmers notes that this kind of emergence actually supports physicalism, because it shows "how all sorts of phenomena that might seem novel and irreducible at first sight can nevertheless be grounded in underlying simple laws."²⁶

Strong emergence would be different.  "Strong emergence, if it exists, can be used to reject the physicalist picture of the world as fundamentally incomplete."²⁶  Chalmers answers that "there is exactly one clear case of a strongly emergent phenomenon, and that is the phenomenon of consciousness."²⁶  No amount of physical description seems to entail what it is like to be a conscious subject.  Whether he is right is the question of Mind, Reason, and Consciousness, and this article does not repeat it.

Two cautions follow for apologetics.  First, life is not a case of strong emergence.  The chemistry of a cell is complicated, but no property of a living cell is known to be undeducible in principle from its chemistry.  Arguments from the origin of life therefore have to be arguments about how the parts were assembled, which is the subject of The Origin of Life, not arguments that life is something over and above its parts.  Second, Gitt's claim that information is a "non-material" entity that matter cannot produce is in effect a claim of strong emergence for meaning.  Part 6 shows that his case for it rests on definitions rather than on any physical test.

Part 5: Six things called information

"Information" names at least six different quantities.  Most arguments about evolution and information go wrong at the same point:  a premise uses one meaning and the conclusion another.  The table sorts them.  The articles that own each measure are linked rather than repeated.

Shannon and Kolmogorov: information without meaning

Claude Shannon built his theory for telephone engineers, and he set meaning aside on its first page:  "These semantic aspects of communication are irrelevant to the engineering problem. The significant aspect is that the actual message is one selected from a set of possible messages."²⁸  His measure is greatest when every symbol is equally likely, which is also, in his words, "the most uncertain situation."²⁸  A string of coin flips therefore carries more Shannon information than a sentence of the same length.  Gitt agrees:  by Shannon's measure "a random sequence of letters possesses the maximum information content."²⁹

Andrei Kolmogorov proposed a second measure in 1965.  The complexity of a string is the length of the shortest computer program that produces it:  "the minimal length l(p) of the 'program' p for obtaining y".³⁰  Gregory Chaitin, who developed the same idea, explained what follows.  A pattern such as 0101 repeated a million times has a short program.  A patternless string does not, and "A series of numbers is random if the smallest algorithm capable of specifying it to a computer has about the same number of bits of information as the series itself."³¹  A random string of a given length is therefore about as complex as any string of that length can be.  Kolmogorov complexity, like Shannon information, rewards randomness and is blind to meaning.

So in both of these senses, unguided processes make information easily.  Every random mutation that lengthens a genome adds Shannon capacity.  This is why a critic who says "gene duplication plus mutation adds information" and a creationist who says "mutations never add information" usually talk past each other.  The critic is counting bits;  the creationist means function or meaning.  Robert Carter of Creation Ministries International heads his article on the question "It depends on what you mean by ‘new’ and ‘information’."³²  Szostak, from the other side, writes that classical information theory "does not consider the meaning of a message."³³

Functional and specified information: information about a job

Szostak proposed a measure that does track function.  Functional information is the negative logarithm of the fraction of all possible sequences that do a stated job at least as well as a stated level.³³  Hazen and colleagues generalized it and stressed its limit:  "Functional information is defined only in the context of a specific function x."³⁴  Szostak's worked example is an RNA of 70 letters that binds ATP:  about 37 bits of functional information, against 140 bits to specify one exact 70-letter sequence.³³  This is the only measure in the table that matches what both sides mean by biological information.  It cannot yet be computed for a whole gene or genome.  Measuring Biological Information sets out why, and how Durston's Fits relate to it.

Dembski's specified complexity adds a rule for inferring design:  "specified information of complexity greater than 500 bits cannot reasonably be attributed to chance."³⁵  He drew a law from it:  "natural causes are incapable of generating CSI."³⁵  Whether that law holds once selection is in the chance model is assessed in The Design Inference.

Gitt: information with a sender built in

Gitt's "universal information" is the only meaning in the table that includes a mind.  It requires all five of his levels, the last being "intended purpose."¹  It has no unit, and Gitt's own book says that measures for its higher levels are still to be defined.⁵  Part 6 tests it.

Physicists: information as a physical quantity

Physicists mean something close to Shannon's quantity, and they treat it as physical.  In 1961 Rolf Landauer of IBM showed that erasing information has an unavoidable energy cost.  Resetting a bit "requires a minimal heat generation, per machine cycle, typically of the order of kT for each irreversible function", and for a simple reset the minimum is "0.6931 kT per restored bit".^{36 36}  The number 0.6931 is the natural logarithm of 2;  k is Boltzmann's constant and T the temperature.  In 2012 Antoine Bérut and colleagues held a single colloidal particle in a double-well potential, used its position as a bit, and found that "the mean dissipated heat saturates at the Landauer bound in the limit of long erasure cycles."³⁷  Information in this sense is a property of physical states, and it has a price in joules.  Its link to thermodynamic entropy is treated in Entropy and the Arrow of Time.

John Wheeler went further.  He proposed that physics itself rests on information:  "every it — every particle, every field of force, even the spacetime continuum itself — derives its function, its meaning, its very existence entirely — even if in some contexts indirectly — from the apparatus-elicited answers to yes or no questions, binary choices [52], bits."³⁸  He even called its source immaterial:  "It from bit symbolizes the idea that every item of the physical world has at bottom — at a very deep bottom, in most instances — an immaterial source and explanation".³⁸  A creationist might welcome this.  It does not help Gitt.  Wheeler's bits are yes-or-no answers registered by measuring equipment.  They have no sender, purpose or meaning in Gitt's sense, and the same paper lists "No laws" among its working principles, which is the opposite of a lawgiver.³⁸  "It from bit" is a research proposal, not an established result.

Digital physics and the simulation argument

Some physicists propose that the universe is a computation.  Konrad Zuse, the German computer pioneer, suggested in 1969 that space might be a grid of tiny automata:  "The first result of viewing the cosmos as a cellular automaton is that the single cells represent a finite automaton."³⁹  Seth Lloyd computed the most the universe could have computed since the Big Bang:  "The universe can have performed no more than 10¹²⁰ ops on 10⁹⁰ bits."⁴⁰  Max Tegmark goes furthest:  "Our external physical reality is a mathematical structure."⁴¹

These are speculations, and they cut in two directions.  If the world is a mathematical structure, its regularities might be necessary rather than contingent, which would remove the need for a lawgiver.  But a computation needs rules and a substrate, and a mathematical structure has to be picked out from all the others that could have been real.  Nick Bostrom's simulation argument shows how close this line of thought comes to theism.  He argues that one of three claims is true, the third being that "we are almost certainly living in a computer simulation."⁴²  He then notes that "the posthumans running a simulation are like gods in relation to the people inhabiting the simulation", having "created the world we see", with power to "interfere in the workings of our world even in ways that violate its physical laws".⁴²  A simulation hypothesis does not remove the question of where laws come from.  It moves it to the simulators' world, whose laws would need the same explanation.

Part 6: Gitt's laws of information, tested against his own definitions

Werner Gitt was a director and professor at Germany's Federal Institute of Physics and Technology in Braunschweig.¹  His argument is the most ambitious form of the information argument, because it claims the rank of a law of nature.  He set out "to formulate laws of nature" about information, and he concluded that those laws "exclude the possibility that information, including biological information, can arise purely from matter and energy without reference to an intelligent agent."¹  He compares evolution to a perpetual motion machine:  "Those who believe that evolution is a plausible concept believe in a “perpetuum mobile of information”."⁴³

Gitt's position in his own words

Gitt calls what he defines "universal information."  It is present only when a system shows all five of his levels:  statistics, syntax, semantics, pragmatics and apobetics.¹  Apobetics is his word for purpose.  In one sentence, universal information is "an encoded, symbolically represented message conveying expected action and intended purpose."¹

He then states four laws:¹

1. SLI-1: "A material entity cannot generate a nonmaterial entity".
2. SLI-2: "Universal information is a non-material fundamental entity".
3. SLI-3: "Universal information cannot be created by statistical processes".
4. SLI-4: "Universal information can only be produced by an intelligent sender". The sender must be conscious, have a will of its own, be creative, think autonomously and act purposefully.

From SLI-4 he derives four narrower rules.  The two that matter here are SLI-4a, "Every code is based upon a mutual agreement between sender and receiver," and SLI-4b, "There is no new universal information without an intelligent sender."¹  He applies them to the cell directly:  DNA's "substrate is not the original sender; rather, it is only the intermediate sender."¹

His conclusions are large.  In a companion paper he draws eight, among them "There must be an intelligent sender!", "Big bang is impossible", and "All theories or concepts of chemical and biological evolution (macroevolution) are false."^{43 43 43}

Gitt also states the test his laws must pass.  "Laws of nature know no exceptions," and "A law of nature can, in principle, be refuted—a single contrary example would end its status as a natural law."¹  He says none of his four has been refuted:  "None of them has been falsified by way of an observable process or experiment."¹  That is a fair standard, and the tests below use it.

Test 1: the purpose level puts the sender into the definition

Gitt's definition of the highest level reads:  "We have already recognized that for any given information the sender is pursuing a goal."¹  His 1996 paper says the same of codes.  Theorem 5:  "The assignment of the symbol set is based on convention and constitutes a mental process."  Its closing summary:  "No code can exist without a free and deliberate convention."^{29 29}

So universal information is defined as the product of a sender with a goal.  SLI-4 then says that universal information can only come from an intelligent sender.  That is true, but it is true by definition, the way "every husband is married" is true.  No observation was needed to establish it, and none could overturn it.

The weight therefore falls on a different step:  showing that DNA meets the definition.  To count DNA as universal information, one must first show that its effects were an outcome "intended by the sender."¹  That is the very thing in dispute.  Gitt's first conclusion does not show it.  It asserts it:  "Because it can be established that all forms of life contain a code (DNA, RNA), as well as all of the other levels of information, we are within the domain of our definition of information."⁴³  Read this way, the argument assumes a sender in order to prove one.

Test 2: read purpose as function, and the law meets counterexamples

A defender can reply that Gitt means something observable by purpose:  the job a sequence does.  Gitt's own examples support that reading.  He lists "the development of biological cells, the functions of organs" among the actions information causes, and he gives bird song as an example of purpose:  "With its song, the male bird would like to gain the attention of the female or to lay claim to a particular territory."^{29 29}  Gitt would say that the bird is only an intermediate sender and that the purpose belongs to its Creator.  But then the purpose is recognized from the job the song does.  In practice Gitt identifies purpose in biology by function.

On that reading the definition is no longer circular, and SLI-4b becomes a real claim about the world.  It then meets Gitt's own standard of one contrary example.  His book states the claim about mutation flatly:  "There can be no increase in information, and in general the results are injurious."⁵  In 2012 Justin Meyer and colleagues in Richard Lenski's laboratory let phage lambda evolve with its host.  In 24 of 96 replicate populations the virus gained the ability to enter cells through a second protein, OmpF, that it could not use before.  All 24 isolates that could do it shared four changes in the same viral gene:  two identical mutations, a third in the same codon as one of them, and a fourth within one short stretch of the gene.  Phage with only three of the four could not use OmpF, and the authors concluded that "all four are required for λ to infect through OmpF."⁴⁴  A new receptor function arose by mutation and selection in a flask, with no one choosing the mutations.  The citrate-using E. coli of the same laboratory is a second case.  Both, and their limits, are set out in When Evolution Creates Information.

Gitt's best reply is that such changes modify existing information rather than create new information.  That reply needs a way to tell new information from modified information, and Gitt's book says the needed measure does not yet exist:  "it is necessary to define measures for semantic information which must be generally valid."⁵  His 1996 paper ends the same way:  "A quantitative survey is among the many tasks still to be performed."²⁹  Without a unit, "no increase" cannot be checked against any mutation, for or against.  Robert Carter of Creation Ministries International, who adopts Gitt's definition, has dropped the flat claim:  "The phrase, “Mutations cannot create new information” is almost a mantra among some creationists, yet I do not agree."³²

The dilemma is the result.  If purpose means an intending sender, SLI-4 is true by definition and cannot be applied to DNA without assuming the answer.  If purpose means function, SLI-4b is an empirical claim, and observed gains of function contradict it.

Test 3: is information non-material?

SLI-1 and SLI-2 say that information is non-material and that matter cannot produce it.  Gitt's test for a non-material thing has four conditions, and the third is "SC3: The entity does not originate in pure matter."¹  That condition is the conclusion SLI-1 is used to reach.  It cannot also be evidence for it.

His second argument is that information cannot be built from the seven SI base units, "and therefore information is not a physical magnitude!"¹  The SI does not support this.  Its official brochure says that "some quantities that cannot be described in terms of the seven base quantities of the SI" are counts of things, "a number of cellular or biomolecular entities," and that "These quantities are also quantities with the unit one."⁴⁵  The same brochure uses the bit:  "one kilobit represents 1000 bits and not 1024 bits."⁴⁵  A count of molecules is not built from base units either, and nobody calls it non-material.

Gitt's third argument is sound as far as it goes.  The same message can be written in chalk, smoke or magnetized domains, and wiping the board destroys the message while every grain of chalk remains.¹  This shows that a message is a pattern and not a substance.  It does not show that the pattern needs a mind to arise, because many patterns are carried by many media:  a sound wave by air or water, a cubic crystal shape by many different salts.  Meaning, as distinct from the pattern that carries it, does seem to need a mind.  But that is the claim about DNA in dispute, not evidence for it.

Test 4: the genetic code as a convention

SLI-4a is the strongest of Gitt's claims, because it points at a real fact.  As far as anyone knows, chemistry does not dictate which codon stands for which amino acid.  Eugene Koonin and Artem Novozhilov, in a mainstream review, conclude that "the code is not based on a straightforward stereochemical correspondence between amino acids and their cognate codons (or anticodons)."⁷  Gitt infers that the assignment "was at some prior time freely chosen."¹

The inference does not follow by itself.  An arbitrary assignment is also what Francis Crick's "frozen accident" proposal predicts.  In Koonin and Novozhilov's summary, Crick held that once the code covered all twenty amino acids, "any change in the code would result in multiple, simultaneous changes in protein sequences and, consequently, would be lethal, hence the universality of the code."⁷  On that view an early assignment fixed by chance would look just as arbitrary as one fixed by choice.  And assignments do change in living lineages.  Koonin and Novozhilov report more than twenty variant codes.  They explain that "The underlying mechanisms of codon reassignment typically include mutations in tRNA genes, where a single nucleotide substitution directly affects decoding."⁷  Gitt's own example is one of these, though he names the wrong codon.  He writes that "the triplet GAC" means leucine and is read as serine in some yeasts.¹  GAC codes for aspartate.  The codon in question is CUG, which "in the fungus Candida zeylanoides is decoded as either leucine (3–5%) or serine (95–97%)."⁷

What survives is the fact underneath.  Nobody knows how a coding system and the machinery that reads it first arose.  Koonin and Novozhilov call it "perhaps, the most formidable problem of all evolutionary biology" and describe "a dreary vicious circle" at its heart.⁷  That is a real argument, taken up in The Origin of Life.  It is an argument from an unsolved problem, not from a law.

Test 5: physicists say information is physical

Physicists use "information" in a sense close to Shannon's, and in that sense they say the opposite of SLI-2.  Landauer's result, described in Part 5, is that erasing one bit releases at least kT ln 2 of heat, and Bérut's group measured that bound.^{36 37}  Information in this sense is tied to physical states and interacts with matter through energy.

This is not a contradiction of Gitt's semantic claim, because the two use the word differently.  It does refute his condition SC1, that information has "no physical or chemical interaction with matter," for information at Shannon's level, which is the level Gitt himself calls "the bridge between the material and the non-material world."^{1 1}

What survives

The surviving form is Stephen Meyer's, and Meyer states it more carefully than Gitt.  He argues that "intelligent design stood as the only known cause of specified information-rich systems and, therefore, that ID provides the best, most causally adequate explanation for the origin of the information necessary to produce the first life."⁶  He limits the claim himself:  "Since intelligence is the only known cause of specified information (at least starting from a nonbiological source), the presence of specified information-rich sequences in even the simplest living systems points definitely to the past existence and activity of a designing intelligence."⁶  The words "at least starting from a nonbiological source" give up the ground Gitt claims for SLI-4b, all new information inside living things.  What remains is an argument about the origin of life.  It reasons from known causes to an unknown one, which is induction, and Part 3 explains both why that is legitimate and why it can never be closed.

Objections and responses

"Laws are descriptions.  Asking who wrote them is a category mistake."  Against the weak argument, this is right, and Part 2 grants it.  Against the strong argument it misses the target.  Swinburne and Foster do not ask who wrote the description.  They ask why the thing described is so uniform.  Beebee's reply, that Humean laws need not "do" anything, is coherent, but it leaves the uniformity with no explanation at all.¹⁸  That is a cost a naturalist may choose to pay, not a refutation.

"Who made God?"  Carroll's version is that "we are left to explain the existence of a creator."¹⁵  Every view ends somewhere, and theism ends in God.  The argument from order does not claim that theism has no stopping point.  It claims that theism's stopping point predicts an orderly world, while a bare universe does not.  Whether God can be a stopping point of a different kind from the universe is argued in The Cosmological Argument.

"Theism could explain any world, orderly or chaotic, so it explains nothing."  The Bible does not leave the matter open.  It promises regular seasons and day and night "while the earth remains."³  A world without stable regularities would count against the biblical God.  Swinburne gives the reason a good God would choose order:  embodied creatures can only learn, plan and act responsibly in a world whose behaviour they can predict.¹⁶

"Science works.  Induction needs no further justification."  That science works is the fact to be explained.  Using its past success to justify its future success assumes the very principle in question, which was Hume's point.⁴  Theists rely on induction as much as anyone.  The question is only which view makes its success something to expect.

"This is a God-of-the-gaps argument.  Science may yet explain the laws."  Science explains laws by deeper laws.  It cannot explain why there are laws at all, because every scientific explanation uses one.  Swinburne makes the point:  science could not explain "why the most fundamental natural laws of all hold.  Either these are brute facts about the world, or they have an explanation of a different" kind.¹⁶  If physics one day derives every law from a single principle, the question moves to that principle.  Tegmark's proposal that the world simply is a mathematical structure is the one serious attempt to close the gap, and it is a hypothesis without evidence.⁴¹

"Miracles contradict the theistic view of laws as God's faithfulness."  On the theistic account a law is God's usual way of acting, and a miracle is God acting differently for a reason.¹⁴  The promise of Genesis 8:22 is a promise of seasons, not a promise that God will never act in a new way.  A miracle is no more a broken law than a parent's unusual act is a broken habit.

"Gitt is a straw man.  Serious design theorists do not use him."  Gitt's papers are published and circulated by Creation Ministries International, and Robert Carter of that ministry adopts his definition of information.³²  His argument is the source of the common claim that information "cannot arise from matter," so it needs testing.  This article identifies Meyer's version as the one that survives.

"When creationists say information, they mean specified information.  Citing Shannon is an equivocation."  That is the article's point, made from the other side.  Once information means functional information, the claim that mutation never adds it becomes a claim about observed rates and origins.  Then it has to be tested on data, and the tests are in When Evolution Creates Information and Measuring Biological Information.

"The lambda and citrate experiments were designed by scientists, so intelligence was involved."  The experimenters chose the flasks, the medium and the host.  They did not choose the mutations.  Gitt's law concerns the selection of symbols:  "A sequence of characters are selected from an available, freely defined set of symbols".¹  In the lambda experiment no one selected the four changes;  they arose and were kept because the viruses carrying them multiplied.  The same thing happens in a pond.

"Information is immaterial.  You cannot weigh the meaning of a sentence."  Agreed:  meaning is not a substance with mass.  But a DNA sequence is a physical arrangement of molecules, and Shannon-level information has a measurable energy cost.³⁶  Whether DNA carries meaning in Gitt's sense, rather than function, is the point in dispute.  The immateriality of meaning cannot be used to settle it.

"Wheeler showed the universe is made of information, so it must come from a mind."  Wheeler's bits are answers to yes-or-no questions posed by measuring equipment.³⁸  They have no sender and no purpose, and "it from bit" remains a proposal.  It gives no support to the claim that a mind is the source of physical information.

"The order of nature is just what observers would see.  There is nothing to explain."  Observers need order where they live.  They do not need the same atomic physics in gas clouds across the visible universe, and that is what spectroscopy finds, to about a part per million.²⁰  The anthropic reply explains local order.  It does not explain universal order.

Issues and unanswered questions

1. The probabilities cannot be computed. Swinburne's argument says an orderly universe is far more probable on theism than on naturalism. No one has a measure over possible sets of laws, so neither probability can be given a number. The verdict of "modest" evidence rests on a shared intuition, not a calculation, and a reader is entitled to weigh it differently.
2. Foster's earlier papers were not read. His 2004 book cites his paper of 2000–1, "Regularities, Laws of Nature, and the Existence of God", as the first statement of the argument, and his 1982–83 paper on induction is the root of his inductive argument. Neither was read. The book states both arguments in full.
3. Some Humean sources were read in versions other than the published ones. Loewer's paper was read as a pre-publication draft, so its quotation is located by section, not by the journal's page. Lewis's paper of 1994, "Humean Supervenience Debugged", was not read. Mumford's and Bird's journal articles were not read; their books were.
4. The genetic code question is open in both directions. Chemistry does not dictate codon assignments as far as anyone has shown, and nobody knows how the coding system arose. If a future experiment showed a coding system arising from chemistry without guidance, the surviving form of the information argument would fall. If origin-of-life research continues to stall at the "dreary vicious circle", the argument gains weight.⁷
5. The theology of secondary causes is not covered. Theists disagree about whether God acts directly in every event, the view called occasionalism, or sustains created things that act by their own powers, which is Boyle's view. The difference matters for miracles and for how to read the laws, but not for the argument from order.
6. Topics left to other articles. The second law and thermodynamic information belong to Entropy and the Arrow of Time. Dembski's conservation of information belongs to The Design Inference. The black-hole information paradox, the mathematical effectiveness of physics and quantum indeterminism are left to the other articles of Foundations of Physics.
7. Errors in sources. Gitt's note on codon reassignment names GAC where the codon is CUG.¹ The Stanford Encyclopedia of Philosophy's entry on laws gives Armstrong's N(F,G) passage as page 85 of What Is a Law of Nature?; it is on page 79.⁹ Neither error changes the argument.

Verdict

A law of nature is either a summary of what happens or something that makes it happen, and the evidence does not force a choice.  On either reading, the uniformity of nature is a fact that theism expects and naturalism has to accept without explanation.  Swinburne and Foster state that argument without any pun on the word "law," and the best naturalist replies, from Beebee, Loewer and Carroll, show that it is not a proof without showing that it is weak.  Induction has no non-circular proof on any view.  Theism and the necessitation account at least give a reason to expect it to work, and a pure regularity account gives none.

The information argument has to be sorted before it can be judged.  In Shannon's and Kolmogorov's senses, nature makes information freely.  In Gitt's sense, information is defined to need a mind, so the claim that it needs one tells us nothing about DNA until DNA is shown to fit the definition, which is the question at issue.  Read as function, Gitt's claim that information never increases is contradicted by observed gains of function.  What survives is Meyer's narrower claim:  that intelligence is the only known cause of coded, functional information "at least starting from a nonbiological source."⁶  That is an inference to the best explanation about the origin of life.  It is as strong as the failure of the alternatives, and it should be stated that way, never as a law of nature.

Sources