Argument

The Record-Keeping Argument

Intro

Everything you know about the past, you know from something in the present. A memory in your brain. A photograph. A fossil in a rock. Light from a star that left it long ago. A letter in an archive. History, science and your own life story all depend on records.

Here is the surprise physics has found. The basic laws of physics run the same forwards and backwards. Taken alone, they do not tell you that a record really comes from the event it seems to show. The only reason a footprint means someone walked here, or a memory means something happened, is that the universe began in an extraordinarily orderly state and has been running down ever since. Without that special beginning, the physics says that your memories and every record around you would most likely be random flukes.

The Record-Keeping Argument says that a universe whose past can be known is a gift. It required a starting state so special that one physicist put the odds of hitting it by chance at a number too large to write out. That gift is exactly what you would expect from a God who wants His creatures to remember, a God who tells His people again and again, "Remember the days of old."

◎ See this page's graph and discover linked topics →

In full

The Record-Keeping Argument is a likelihood argument from the physical preconditions of historical knowledge. Its explanandum is not the low-entropy beginning as such, which the Fine-Tuning Argument and the Argument from Thermodynamics treat, but the fact that the beginning makes the past knowable. Because the fundamental dynamical laws are, for this purpose, time-reversal invariant, statistical mechanics applied to the present macrostate alone retrodicts a past of higher entropy, in which present records are fluctuations rather than traces. Albert's Time and Chance (2000) argues that the asymmetry of our knowledge, our having records of the past and not of the future, requires adding the Past Hypothesis, that the universe began in a state of very low entropy; Penrose (1989) estimated the specialness of that state at about one part in 10^(10^123); and Carroll (2017) argues that cosmologies dominated by fluctuated observers are cognitively unstable, since they undercut the evidence used to believe them. The spine (P1 to P5) argues that a knowable past, which requires such a state, is far more expected on a God who means creatures to know, remember and be accountable to history than on brute initial conditions or chance. The strongest rival, Callender's argument that special initial conditions need no explanation, is answered in the defeater gate. The conclusion is confirmatory, and it overlaps with the fine-tuning of the initial entropy, which the ledger counts once.

Format note

This page is a braided case: a valid spine, each premise supported by an inductive strand, an abductive comparison and a convergence note, followed by a knot, a likelihood ledger and a defeater gate. Four fields are braided. Physics shows that records are reliable only given a special low-entropy beginning. History is the discipline built on records, and so rests on that beginning without knowing it. Psychology shows that memory is itself a physical record, fallible and reconstructive, yet trusted by every sane person. Philosophy asks whether such a beginning needs an explanation at all, and what follows if it does. Other arguments in this format are listed at Ris3n Arguments.

The spine

Validity. P1 and P2 show that a knowable past requires a low-entropy beginning. P3 shows how special that beginning is. P4 says such a beginning, as the precondition of knowable history, is far more expected on theism. P5 converts that into confirmation. The form is a likelihood comparison closed by modus ponens.

Strength of the conclusion. Confirmatory. The physics of P2 and P3 is widely accepted, though the numbers in P3 are estimates; whether the beginning needs any explanation is contested, which is the subject of the defeater gate.

P1, Records

Inductive strand

1. History is knowledge by traces. The French historian Marc Bloch, in the work published as The Historian's Craft, described historical knowledge as indirect: we know the past only through the witnesses and traces it has left.
2. Science reads the past from present records. Geology reads strata, paleontology fossils, cosmology the Cosmic Microwave Background, which is light from about 380,000 years after the big bang.
3. Personal memory is a record. A memory is a physical trace in the brain, laid down by an event and read later.
4. Reid made memory a first principle. Thomas Reid, Essays on the Intellectual Powers of Man (1785), held that we must trust memory as a basic source of knowledge, since every attempt to prove its reliability would use it.

Abductive strand

Convergence

The methodology of history (Bloch), the practice of the natural sciences and the philosophy of memory (Reid) agree on P1. It is not disputed.

Objections and rebuttals

1. "We also know the past by inference from laws." Rebuttal. Inference from laws starts from present data, which are records, and P2 shows the laws alone point the wrong way without the Past Hypothesis.

P2, Records need the Past Hypothesis

Inductive strand

1. The laws are time-symmetric for this purpose. The fundamental dynamical laws relevant to thermodynamics run the same in both directions of time. The small violation of time symmetry in certain particle decays plays no role in the thermodynamic arrow.
2. Statistics alone retrodicts the wrong past. Given a present state of middling entropy, almost all past histories compatible with it are of higher entropy, not lower. So, using the laws and statistics alone, a broken egg is more likely to have assembled itself from scrambled matter than to have been laid by a hen. Richard Feynman, in his 1964 Messenger Lectures, said it is necessary to add to the physical laws the hypothesis that in the past the universe was more ordered, in the technical sense, than it is today (The Character of Physical Law, 1965, chapter "The Distinction of Past and Future").
3. Albert's analysis. David Z. Albert, Time and Chance (Harvard, 2000), argued that the asymmetry of our knowledge, that we have records of the past but not of the future, follows only when the dynamical laws and the statistical postulate are supplemented by the Past Hypothesis, the claim that the universe began in a very low-entropy state. His chapter 6 treats the asymmetries of knowledge and intervention.
4. Boltzmann saw the danger first. Ludwig Boltzmann, "On Certain Questions of the Theory of Gases" (Nature 51, 1895), considered that our ordered region might be a rare fluctuation in a universe mostly at equilibrium.

Abductive strand

Convergence

Statistical mechanics (Boltzmann, Feynman), philosophy of physics (Albert, Price) and cosmology (Penrose, Carroll) reach P2. Physicists and philosophers who disagree sharply on other issues accept that the Past Hypothesis is needed.

Objections and rebuttals

1. "Entropy increases; that is just the second law." Rebuttal. The second law is itself explained by the low-entropy past. Without the Past Hypothesis, statistical mechanics predicts entropy increasing toward the past as well as the future. See Second Law of Thermodynamics.
2. "Records are reliable because we have checked them." Rebuttal. Every check uses other records. The question is what makes the whole system of records trustworthy, and the answer physics gives is the special beginning.

P3, The beginning is extraordinarily special

Inductive strand

1. Penrose's estimate. Roger Penrose, The Emperor's New Mind (Oxford, 1989), estimated, using the entropy of black holes as a measure of the available states, that the initial state of the universe occupied about one part in 10^(10^123) of the possible states, a precision he pictured as the Creator's aim. The number is an estimate, and the measure behind it is debated, but no one disputes that the initial state was extremely special.
2. Fluctuated observers. Andreas Albrecht and Lorenzo Sorbo ("Can the universe afford inflation?" Physical Review D 70, 2004) gave the name "Boltzmann brains" to observers produced by random fluctuation, and showed that some cosmological models predict them in abundance. See Boltzmann Brain.
3. Cognitive instability. Sean M. Carroll, "Why Boltzmann Brains Are Bad" (2017, in Current Controversies in Philosophy of Science), argued that theories which predict that most observers like us are fluctuations are "cognitively unstable": they cannot be both true and justifiably believed, because they imply that the records on which our belief in them rests are probably spurious.

Abductive strand

Convergence

Mathematical physics (Penrose), cosmology (Albrecht and Sorbo, Carroll) and philosophy of science (Albert, Price, Callender) agree on the facts of P3. Penrose and Carroll are not theists.

Objections and rebuttals

1. "Penrose's number is made up." Rebuttal. It is a calculation under stated assumptions, and other measures give different numbers. The argument uses it as an illustration of a point no one disputes: the beginning was special far beyond what chance makes expected.
2. "Inflation explains the low entropy." Rebuttal. Penrose and others have argued that inflation itself needs a special, low-entropy starting patch, so it moves the question back rather than answering it. This remains debated.

P4, The hypotheses

Inductive strand

1. Scripture's God commands remembrance. "Remember the days of old, Consider the years of many generations: Ask thy father, and he will show thee; Thine elders, and they will tell thee" (Deuteronomy 32:7, ASV). "I will remember thy wonders of old" (Psalm 77:11, ASV).
2. The faith rests on recorded history. Luke writes that he had "traced the course of all things accurately from the first," so that his reader "mightest know the certainty concerning the things wherein thou wast instructed" (Luke 1:3-4, ASV). Christianity stakes its truth on events in the knowable past.
3. God keeps records and judges by them. "A book of remembrance was written before him" (Malachi 3:16, ASV). A God who will judge history by its true record has reason to make that record knowable.
4. Creation has a beginning set by God. "In the beginning God created the heavens and the earth" (Genesis 1:1, ASV).

Abductive strand

On theism, a knowable past is not an accident; it is part of making creatures who can remember, repent, give thanks, keep covenant and be judged justly. On naturalism, the initial state has no purpose, and its fitness for knowers is a coincidence or an observer-selection effect. Observer selection explains why we exist in a low-entropy region; it does not explain why the beginning was low-entropy on a cosmic scale rather than just locally, which is what the Boltzmann-brain problem shows.

Convergence

Biblical theology (remembrance and covenant), historical theology (Christianity as a religion of recorded events) and the physics of P2 and P3 meet at P4: a universe whose past can be read.

Objections and rebuttals

1. "Theism could explain any initial state." Rebuttal. Not equally. Theism specifically predicts a world in which persons can know their past and be accountable to it. A Boltzmann-brain world, in which records are spurious, is the opposite of what a truthful God would make.

P5, The likelihood principle

Inductive strand

Evidence favours the hypothesis on which it is more expected. Carroll's own argument against Boltzmann-brain cosmologies uses a related principle: a theory that makes our evidence improbable or self-undermining is disfavoured.

Abductive strand

Its rival, that boundary conditions are never evidence for anything, would also forbid the cosmologist's argument against fluctuation-dominated models.

Convergence

Bayesian confirmation and the physicists' own reasoning about cosmologies agree. See Bayesian Probability.

Objections and rebuttals

1. "There is only one universe; probabilities do not apply to it." Rebuttal. Then physicists may not reject Boltzmann-brain cosmologies on probabilistic grounds either, yet they do, and rightly.

The knot

The strands come from separate traditions: historiography (Bloch), philosophy of memory (Reid), statistical mechanics (Boltzmann, Feynman), philosophy of physics (Albert, Price, Callender), cosmology (Penrose, Albrecht and Sorbo, Carroll) and Scripture. The main overlap is with the fine-tuning of the initial entropy: Penrose's number is the same datum used in the Fine-Tuning Argument and in the Argument from Thermodynamics. This argument does not count it again as evidence of design in general. Its distinct datum is the conjunction: the special beginning is precisely what makes records, memory and history possible. A second overlap, with Argument from the Information-Conservation Convergence, is avoided: that page concerns whether information is preserved in physics, while this one concerns whether information about the past is readable by creatures.

The ledger

Prior-agnostic: the ledger asks only how expected each datum is under each hypothesis. The ratings are judgments argued above, not measurements.

Net. Most of the argument's force rides on the same low-entropy beginning that fine-tuning arguments use, and the ledger does not count that twice. What this argument adds is the specific role of the beginning: it is the precondition of knowable history, and a world of spurious records would be the opposite of what a truthful, covenant-keeping God makes. That is a modest but distinct confirmation.

The defeater gate

Special initial conditions need no explanation. The strongest objection is Craig Callender's, in "Measures, Explanations and the Past: Should 'Special' Initial Conditions Be Explained?" (British Journal for the Philosophy of Science 55, 2004). Callender argues, against Huw Price (Time's Arrow and Archimedes' Point, 1996), that the low-entropy initial state is a boundary condition, and boundary conditions are where explanation starts, not something that itself calls out for explanation. Drawing on Humean arguments, he contends that being improbable on some measure does not make a fact require explanation, and that the explanations on offer are no more satisfying than accepting the Past Hypothesis as brute.

The reply has three parts.

1. Physicists do not treat it as idle. Cosmologists reject Boltzmann-brain cosmologies because they make our evidence spurious (Carroll, 2017). That is a judgment that the initial state matters to whether a theory is acceptable. Once that is granted, it is fair to ask which hypothesis about the world makes a knowable past expected.
2. Brute is not the same as explained. Callender may be right that science can accept the Past Hypothesis as a starting point. But that is a decision about where science stops, not a demonstration that nothing explains it. The theist offers an explanation; the Callender view declines to look for one.
3. The datum is specific. The argument does not say "improbable, therefore designed." It says the initial state has a specific feature, making the past readable to knowers, which a truthful Creator has reason to secure. Specificity plus improbability is the standard mark of a fact that calls for explanation.

What the gate concedes: a consistent Humean can accept the Past Hypothesis as a brute law and stop there, and the argument cannot compel the further question. Its claim is that a world whose history can be known is far more at home on theism than as an unexplained boundary condition.

Concessions

• Grant: the Past Hypothesis may be part of the laws. Now collect: then the laws themselves are fitted to make history knowable, which is a fact about the laws worth explaining.
• Grant: Penrose's figure depends on assumptions. Now collect: every estimate agrees the beginning was special far beyond chance.
• Grant: memory is fallible and reconstructive. Now collect: we can only know that by comparing memories with other records, which presupposes the trustworthiness of records in general.
• Grant: observer selection explains why we find ourselves in an ordered region. Now collect: it predicts the smallest ordered region compatible with observers, not a whole universe of reliable records.

Cheatsheet

The 30-second version (read aloud). Everything you know about the past comes from records: memories, photos, fossils, starlight. But physics runs the same forwards and backwards. The only reason a record points to a real past event is that the universe started in an incredibly orderly state, so special that Roger Penrose put the odds at one in ten to the ten to the one hundred twenty-three. Without that beginning, your memories would most likely be random flukes, and so would history. A knowable past is a gift. It is exactly what you would expect from a God who keeps saying, "Remember."

Fast facts:

1. Albert (Time and Chance, 2000): our knowledge of the past requires the Past Hypothesis, a low-entropy beginning.
2. Penrose (1989): the initial state was special to about one part in 10^(10^123).
3. Carroll (2017): theories dominated by fluctuated observers are "cognitively unstable."
4. Feynman (The Character of Physical Law, 1965): physics must add the hypothesis that the past was more ordered.
5. Bloch: history is knowledge through witnesses and traces.

The closing line. If the past can be known, someone made it knowable.

Cross-examination

Round 1, How do you know the past?

• Q: Can you name any knowledge of the past that does not come from a present record?
• If they say no: then everything rests on records being reliable. What makes them reliable?

Round 2, What makes records reliable?

• Q: Do the laws of physics alone tell you that a footprint came from a foot rather than from a random fluctuation?
• If they say yes: point to Albert and Feynman: the laws are time-symmetric, and the statistics alone point the other way.
• If they say no, you need the low-entropy past: then why was the beginning like that?

Round 3, Brute or explained?

• Q: Would you accept a cosmology in which most observers are random fluctuations?
• If they say no: then you agree the initial state matters to our knowledge. Which hypothesis makes a knowable past expected?
• If they say it is brute: then you have stopped explaining at the precondition of all history.

Dodge table

Live-cite kit

Scripture (ASV):

• Deuteronomy 32:7: Remember the days of old, Consider the years of many generations: Ask thy father, and he will show thee; Thine elders, and they will tell thee. (Deuteronomy 32:7, ASV)
• Psalm 77:11: I will make mention of the deeds of Jehovah; For I will remember thy wonders of old. (Psalm 77:11, ASV)
• Luke 1:3-4: it seemed good to me also, having traced the course of all things accurately from the first, to write unto thee in order, most excellent Theophilus; that thou mightest know the certainty concerning the things wherein thou wast instructed. (Luke 1:3-4, ASV)
• Malachi 3:16: and a book of remembrance was written before him, for them that feared Jehovah, and that thought upon his name. (Malachi 3:16, ASV)
• Genesis 1:1: In the beginning God created the heavens and the earth. (Genesis 1:1, ASV)

Scholarly:

• David Z. Albert, Time and Chance (Harvard, 2000): the Past Hypothesis and the asymmetry of knowledge.
• Roger Penrose, The Emperor's New Mind (1989): one part in 10^(10^123).
• Sean M. Carroll, "Why Boltzmann Brains Are Bad" (2017): cognitively unstable theories.
• Craig Callender, BJPS 55 (2004): the strongest rival, special initial conditions need no explanation.

Aphorism: "Every record is a receipt for a gift given at the beginning."

Connection to Scripture

• Genesis 1:1, the beginning is God's act.
• Deuteronomy 32:7 and Psalm 77:11, the command and practice of remembrance assume a knowable past.
• Luke 1:1-4, the gospel rests on traced, recorded history and eyewitness testimony.
• Malachi 3:16, God's own book of remembrance.
• Hebrews 11:3, the worlds framed by the word of God.

Scholarship

Published academic works this page draws on, each matched to a publisher, index or library catalog record.

• David Z. Albert, Time and Chance (Harvard University Press, 2000). doi:10.4159/9780674020139
• Andreas Albrecht and Lorenzo Sorbo, "Can the universe afford inflation?" Physical Review D 70 (2004). doi:10.1103/physrevd.70.063528 open access
• Ludwig Boltzmann, "On Certain Questions of the Theory of Gases," Nature 51 (1895). doi:10.1038/051413b0
• Craig Callender, "Measures, Explanations and the Past: Should ‘Special’ Initial Conditions be Explained?" The British Journal for the Philosophy of Science 55 (2004). doi:10.1093/bjps/55.2.195
• Sean M. Carroll and Jennifer Chen, "Spontaneous Inflation and the Origin of the Arrow of Time" (2004). arXiv:hep-th/0410270
• Sean M. Carroll, "Why Boltzmann Brains Are Bad" (2017). arXiv:1702.00850
• Richard Phillips Feynman, Frank Wilczek and Sean Runnette, The Character of Physical Law (MIT Press, 1965). Open Library
• Roger Penrose, The Emperor's New Mind: Concerning Computers, Minds, and The Laws of Physics (Oxford University Press, 1989). doi:10.1093/oso/9780198519737.001.0001
• Huw Price, Time's Arrow & Archimedes' Point (Oxford University Press, 1997). doi:10.1093/acprof:oso/9780195117981.001.0001

See also

• Arguments, the master index
• Ris3n Arguments, the hub of ris3n's original arguments
• Argument from Thermodynamics, the low-entropy beginning as evidence that the universe began; this page asks what that beginning makes knowable
• Fine-Tuning Argument, where the same initial-entropy datum is counted
• Argument from Twin Asymmetries, the arrow of time paired with forgiveness
• Argument from the Information-Conservation Convergence, whether information is preserved, not whether it is readable
• Argument from the Memory-Continuity Convergence, memory and personal identity
• Boltzmann Brain, the fluctuation problem
• Second Law of Thermodynamics, the arrow that records ride on
• Memory, the biology of the record inside us
• Cumulative Case for Christian Theism, where this argument contributes