祝融说。

The Clockwork of a Gram of Radium

The letter with which Chapter One opened — Einstein to Born, December 1926: He does not throw dice — was disputing the standing of determinacy in the world. This chapter redeems the second rewriting the introduction promised: how the necessity of the individual itself was put back into process. First the crack must be stretched to its widest: a gram of radium in the laboratory. Radium atoms are restless; sooner or later every atom decays, emits radiation, and becomes another element. What can physics say about this gram? An astonishing amount: the number of atoms still alive at any moment decays along a precise exponential curve; the half-life of radium is about sixteen hundred years — after sixteen hundred years, half remains; after another sixteen hundred, a quarter — a curve smooth, stable, and dependable, graduated like the face of a clock. But of any one specified atom, physics cannot say in which second it will decay. And what it gives is not the modest phrase "we do not know": quantum mechanics can in principle say of a single event only its probability, and this is a clause in the body of the theory's grammar, not an unpaid debt of the measuring instruments. Set the two faces of one and the same gram of radium side by side — clockwork in the aggregate, untimed in the individual — and the crack of "random individuals, precise totals" is the deepest door of twentieth-century physics. The thesis this chapter argues can accordingly be put first as one sentence: probability is not noise.

When the crack opened must be told from three deposits.

Two Steps of the Quantum

The first deposit was in Berlin. In October 1900 Planck put together a formula for the spectrum of black-body radiation that agreed with Rubens's experimental data at every point; but to explain the formula, on December 14 he had to propose to the German Physical Society a premise he could not himself accept: energy, in its exchanges with matter, is not continuous but proceeds in one "quantum of action" after another, each equal to the frequency multiplied by a constant. Planck later recounted that this was an act of desperation — for years afterward he still tried to digest the quantum within the classical framework, and did not succeed. The second deposit was in Bern. In 1905 Einstein studied the photoelectric effect and went a step beyond Planck: it is not only the exchanges of energy that are discrete; light itself comes in quanta, one portion at a time; the energy of the electrons knocked out of a metal surface depends only on the frequency of the light, not on its intensity — a judgment doubted at the time by nearly everyone, confirmed a decade and more later by Millikan's experiments, and become the first experimental cornerstone of the quantum hypothesis.

The third deposit bears directly on statistics. Around 1898 the Curies isolated radium; between 1902 and 1903 Rutherford and Soddy established the theory of radioactive decay: decay is a spontaneous transmutation of the atom, and the number of atoms decaying per unit time is proportional to the number present — here is the origin of the exponential curve, and with it the concept of half-life entered physics. What matters is the character of this law: it is statistical by birth. It has nothing to say about "this one atom"; it gives a strict account only to "this population of atoms". For the first time in history, the body of a fundamental physical law carried the words "the individual is unpredictable" — not "not yet worked out", but such is the grammar of the law. The three deposits together: Planck gave discontinuous energy, Einstein gave the quantum of light, Rutherford–Soddy gave the statistical individual. The door stood open — but no one had yet pushed through it, until in the twenties a group of people in their twenties built a new mechanics.

Probability Becomes Grammar

In the summer of 1925 Heisenberg was on Helgoland curing his hay fever, but what he wrestled with was another kind of allergy: the classical concept of the orbit ran into walls everywhere inside the atom. He decided simply to restate mechanics using only observable quantities — the frequencies and intensities of spectral lines — and the multiplication that resulted was non-commutative (a times b is not b times a). Born and Jordan recognized that the strange calculus was matrices, and matrix mechanics took its name from them. The next year Schrödinger, following out de Broglie's matter waves, gave another form — the wave equation, later named for him — and the two apparently dissimilar mechanics were soon proved equivalent. In June 1926, in a paper on scattering, Born set the tone of the new mechanics: the wave function describes no real wave; its squared modulus gives the probability of the particle's appearing — the probability interpretation made its formal debut in a supplementary note, and for it Born received the Nobel Prize of 1954. In the spring of 1927 Heisenberg wrote the limits of the new grammar into a theorem: for the conjugate pair position and momentum, the product of their determinacies has a lower bound — the uncertainty principle. The grammar of quantum mechanics was thereby complete: of single events it says only probabilities; of statistical averages its predictions are of unmatched precision — the hydrogen spectrum, the chemical bond, the energy bands of solids: experimental buildings rose by its drawings, with errors counted in many decimal places. The "unreasonable effectiveness of mathematics" of the introduction here reached its peak.

But over that one word "probability" in the grammar there hung, from the beginning, a debate that never dispersed. October 1927, Brussels, the fifth Solvay Conference, its subject "Electrons and Photons". Einstein came at Bohr again and again with thought experiments: how can the complete formulation of a theory give nothing but probabilities? Bohr took up each challenge in turn, reconstructing the apparatus overnight and next day turning the objection back by the theory's symmetries. The debate then passed into letters and papers. In 1935 Einstein, Podolsky, and Rosen published the famous EPR paper: quantum mechanics predicts instantaneous correlations between a pair of particles that once interacted and then separated; unless "hidden variables" are added — unless one holds that behind the probabilities there is still a deterministic mechanism we have not seen — one must either admit action at a distance or admit that the theory is incomplete; in the same year Schrödinger coined the word "entanglement" for these correlations. Bohr replied under a nearly mirror-image title: quantum mechanics is complete. To many at the time the quarrel looked like a dispute of philosophical taste — until 1964, when Bell turned it into physics.

Bell's question was clean and sharp: if EPR-type local hidden variables exist, the correlations of measurements on two separated particles must satisfy a certain inequality; and quantum mechanics predicts that the correlations can exceed it. This could be adjudicated: not a contest of interpretations but a contest of numbers. In 1972 Clauser and Friedman performed the first test, and quantum mechanics won; in 1982 Aspect's group at Orsay outside Paris improved the apparatus, closing the "prior arrangement" loophole with rapidly switched polarization analyzers, and the result stood — the Bell inequality was violated, and the quantum correlations stood on the side of the experimental data. The experiments that followed tightened every kind of loophole generation by generation: the "loophole-free" tests of 2015, the Big Bell Experiment that gathered a hundred thousand people worldwide to choose measurement settings freely (published in 2018), the Nobel Prize of 2022 — the conclusion has not turned back. The road of local hidden variables was vetoed by the world. Probability is not scaffolding that quantum mechanics borrowed for the time being; at least in a local world, there is no deeper deterministic script to be mined beneath it.

Redeeming Section 1.2 of the General Outline

Now the deepest layer of Part Four can be laid open. RC's General Outline writes in Section 1.2, "The Emergence of Determinacy":

"The stability of matter at the macroscopic level comes from the iterative convergence of multi-level observation; when observation across subjects forms a stable consensus, classical physical entities are expressed as stable determinate structures. The probabilistic character at the microscopic level, meanwhile, exposes the observational divergence that may exist between different observing subjects."

The RC paper, discussing the emergence of determinacy in Section 2.2, pushed the latter half of the sentence to its limit: "the probabilistic character at the microscopic level exposes the observational divergence between different observing subjects that has not yet been smoothed away — probability is not an intrinsic randomness of the world, but the expression of a convergence that has not yet achieved agreement across subjects." The wording of the General Outline is the more cautious ("the observational divergence that may exist"), the wording of the paper one step further ("not yet smoothed away"); this chapter proceeds by the cautious version, setting the history of quantum mechanics against it word for word.

First the macroscopic half. The "clockwork" face of the gram of radium with which this chapter opened — the exponential curve, the constant half-life — is precisely the iterative convergence of multi-level, large-number observation: the decay events of hundreds of trillions of atoms superpose upon one another, every statistical fluctuation averaged away by the sheer mass of the total, re-verified across laboratories, across continents, across decades, an account grown thick enough to agree with measurement through many decimal places. Macroscopic stability is not assumed; it is saved up — Chapter Three told of Newton, this chapter tells of radium: one and the same mechanism, one and the same shape, across two centuries. Then the microscopic half. The single atom is untimed: every observation of it receives a different answer, and what the theory gives is exactly the distribution of the answers. On RC's reading, this is not a dice-throwing hand hidden inside the world, but the faithful record of convergence at the microscopic level — observation after observation has not converged into one and the same answer; and the theory honestly writes "the distribution of the divergence" into its body. Section 1.2 of the General Outline has a further half-sentence that serves this chapter exactly: "nor does observation demand strict progression level by level — high-precision measuring instruments can intervene directly in the observation of microscopic particles: humans observe the instrument, and the instrument observes the particles." In the human observation of microscopic particles, every link of the chain has changed eyes; what the Bell experiments adjudicated is precisely whether, at the end of this chain, a layer of deterministic script visible to all can be dug out — and the world's answer is no.

This chapter's core transcription can now be stood up: the probability amplitude is not the world's "intrinsic randomness", but the formal description of a convergence that has not (or in principle cannot) achieved agreement across subjects. Chapter One planted a line here: the "undetermined" of everyday speech comes in two kinds — "not yet known" (the number is in fact fixed, and we do not know it) and "not yet existing" (the determinacy has not yet been locked); RC takes the latter. The achievement of a hundred years of quantum experiments lies precisely in sealing the first exit section by section: the hidden-variable, local "in fact already fixed" was excluded by the violation of Bell's inequality; what remains is one honest narrow path — probability lives not in our ignorance but in the incompleteness of the process. Einstein demanded to find "the fixed" at the deepest level; Chapter One has already given RC's answer, and it can be quoted here in full: the deepest level is the Ground, and the determined has never lived at the deepest level. Born's probability amplitude, on this reading, is the most successful mathematical writing of the fact that "locking is under way". As for the measurement problem — how exactly the locking occurs, how the wave function "collapses" — this chapter uses it only at the philosophical level: it is the physical development of the observational locking process; the several interpretive traditions contest the mechanism of the development, while RC uses only the fact of the "locking" and does not adjudicate its technical details.

The boundaries must be nailed fast, lest this section be read past its edge. First, RC does not compete with the mathematical content of quantum mechanics: within its own snapshot quantum mechanics is unrivaled in precision, the most exact physical theory ever devised; this chapter offers it nothing but respect, and no amendments. Second, what RC offers is not a new result of physics but a philosophical reading of the standing of probability — the world presents probability at the microscopic level, physics answers "how to calculate", and the philosophical reading answers "what is the status of what has been calculated". Third, "has not yet" and "cannot in principle" are two different assertions: the former is a statement of fact (the history of science may still rewrite the account of microscopic probability), the latter is a reading from within the program (within the RC framework, the margin of the Ground, A7, guarantees that locking never exhausts the Ground — that microscopic convergence stops at the statistical level is precisely the development of inexhaustible margin). Probability is not noise; it is the way available margin keeps its books in physics.

Kuhn as the Control Group

With the history of quantum mechanics told, the control group should take the stage. In 1962 Thomas Kuhn published The Structure of Scientific Revolutions, giving the most-cited picture of the history of science in the twentieth century: normal science — the community solving puzzles within an established "paradigm"; anomaly — observations the paradigm cannot digest; crisis — anomalies accumulating, discipline loosening; scientific revolution — paradigm shift, the new paradigm taking over, old questions voided, and the new and old paradigms "incommensurable" with each other. Kuhn's exemplars were precisely the two revolutions this chapter and the last have told: the old case from geocentric to heliocentric, the recent ones from Newton to relativity and from the classical to the quantum.

RC's uptake of Kuhn is conspicuous, and the items can be matched one by one. Kuhn's "paradigm" and the "rule framework" of General Outline 1.4 are all but isomorphic in name: observational consensus is summarized into causal laws, and causal laws into the rule framework — the paradigm is the institutionalized form of this accumulation in the scientific community. Normal science is puzzle-solving within the snapshot: the controlled observation of Chapter Twelve and the prediction routine of Chapter Thirteen are the daily assignments of normal science — not doubting the framework, but balancing the books within it. Crisis is the accumulation of observational divergence: the three acts of the last chapter — the null result on suspense, the 43 arcseconds of Mercury, patches multiplying layer on layer — are the literal expansion of the word "crisis". Scientific revolution is re-convergence: the snapshot re-locked, the accounts publicly changed. One may say that Kuhn, with a historian's craft, operationalized the dynamics of General Outline 1.4 within the history of science; the three-act play of the last chapter is precisely two developments — Kuhn's narrative and RC's mechanism — of one and the same material.

But RC and Kuhn part ways at two crucial points, and the parting must be set down exactly. First, Kuhn says that the old and new paradigms are "incommensurable" — after a revolution the meanings of terms are renewed wholesale, there is no neutral public language for adjudicating merit, and theory-choice accordingly approaches a gestalt switch, with even a sociological component. RC's reading is the opposite: re-convergence is local surgery, and the account is not zeroed. Newtonian mechanics continues in office within relativity as the limiting case of low velocities and weak fields (Chapter Three, the last chapter), and classical probability goes on working within quantum statistics; the ledger is changed, the stubs are not burned. The plates of 1919 could adjudicate in public precisely because a cross-paradigm public measure exists — error bars, plates, arcseconds are the courtroom language the two theories share. Kuhn weighted incommensurability too heavily; the historical fact is the continuity of the accounts. Second, Kuhn denies a direction of convergence: paradigms merely replace one another and approximate no "truth"; science tends toward nothing, it only changes tools. RC's position stands in the sentence of General Outline 2.1 on processual completeness: the completeness of cognition should ultimately and only inhere in the capacity of the process to iterate, not in the degree of completion of its conclusions — there is no terminus, but there is an account: the snapshots grow thicker and more general generation by generation, converging ever wider bodies of observation into ever fewer ledgers; from Ptolemy's epicycles to general relativity is not a changing of shoes in place but a growth of the account. Kuhn described the form of revolution; RC gives the mechanism and the account of revolution — and the value of the control group lies precisely in bringing out these two differences. There is one further division of labor worth remarking: in Kuhn's picture, the line between normal science and revolution is often read as a relativist slippery slope (if paradigms merely replace one another, why speak of progress?); RC's ledger supplies exactly the missing segment — the phase law of General Outline 3.4 quoted in the last chapter (formal stability at the surface, adjustment of the deep rules, divergence surfacing, becoming explicit, re-convergence) gives "revolution" a recognizable structure of precursors: it is not that any framework can be overturned at any moment, but only when divergence has accumulated sufficient magnitude and the old framework can no longer issue cash that the changing of the account becomes the collective act of the community. Revolutions have a form, and they have an account age.

One boundary of modesty must be kept beyond this: this chapter's criticism of Kuhn is aimed only at the most influential reading of The Structure of Scientific Revolutions; Kuhn himself in his later years revised "incommensurability" again and again, admitting that a translatable public part remains between paradigms. What the control group sets out is stance against stance, not person against person in victory or defeat.

The Close of Part Four

The four chapters can now be closed. Chapter Twelve, controlled observation: isolating a fragment of nature from the mixed stream of observation so as to minimize the noise of cross-subject re-verification — experiment is a manufactured question, and the norms are consensus reinforcement institutionalized. Chapter Thirteen, the prediction machine: locking a convergence path publicly before observation and handing it to the world for verdict — theory is a cognitive relay station, and predictive failure is the entry by which divergence comes onto the books. Chapter Fourteen, May 29, 1919: divergence on suspense, patches maintaining the surface, re-convergence re-locking the space-time snapshot, and ritualized re-observation closing the consensus — the first rewriting of "necessity". This chapter, probability is not noise: the macroscopic clockwork and the microscopic dice are two account ages of one and the same mechanism — accounts already leveled, and accounts not (or not in principle) to be leveled; the second and deeper rewriting of "necessity" rewrites the individual itself.

The introduction's two promises of rewriting are hereby paid, and together they give the final reading of Part Four: science is fallible convergence in dialogue with the world (Chapter Three) — its fallibility is not a defect but its way of keeping the dialogue open; its probability is not noise but its honest accounting of the places not yet locked. The navigator's chart keeps its blanks forever, and precisely because it keeps its blanks, the chart goes on being drawn. The epilogue will return to the night of 1610: Galileo's telescope pushed observation one step farther, the inclined plane made it one measure cleaner, the photographic plate handed it to the whole world for re-verification, and the probability amplitude wrote its incompleteness into grammar — across four hundred years this dialogue-machine with the world has been built ever finer, and its deepest lesson is a single line: determinacy is the achievement of convergence, not the background color of the world. The three forms of reason — the chess player, the architect, the navigator — have each handed in their papers; Part Five should set them back at one and the same table, to compare their speed of convergence, their cross-subject scope, and their resistance to re-convergence, and to answer the question with which the whole book opened: how a reason that is not a miracle can be understood, maintained, and kept going.

Back to the die of Chapter One, the final account can be rendered. Einstein said the Old One does not throw dice; RC's reading is: there is no Old One, and no dice — only observation locking, convergence saving up its accounts, and margin waiting for the next round. Each time the die comes to rest is one locking; and because the Ground has never been locked through, there is always a next throw on the table. Probability is not the disgrace of reason but its honesty: it books "not yet locked" as it stands, and keeps the interface ready for the next re-convergence.