祝融说。

One Table, Three Ledgers

The night of Chapter Three deserves to be remembered. The chess player shuts his book of games — tomorrow the same set of rules will be waiting for him unchanged; the architect draws a new building on her sheets — the conventions of the foundation can be renegotiated; the navigator puts into harbor — one line from today's measurements is added to the chart. That chapter said that the three labors go each to its rest, and that the secret they share is one and the same: everything solid that they rely upon — the rules, the foundation, the chart — was once made, is being maintained, and still keeps its interface for being remade. Across the three parts that followed, this book paid its visits to these three sites of work in turn: from the walking path of Athens to the round table of Königsberg, to watch how the chess player's rules were forged and how, under the strictest examination, they showed their own boundary; from the boundaries of Alexandria to the constructions of forcing, to watch how the architect's foundations were published, how they were exchanged, and how freedom grew out of the exchanging; from the plank with the groove cut in it to the grammar of the probability amplitude, to watch how the navigator learned to put questions to wind and sea, how he was talked back to, and how, in the talking back, he changed his charts.

Now the people from the three sites have come back. The chess player brings his axioms and transformation rules; the architect brings her isomorphisms and categories; the navigator brings his plates and error bars. The task of Part Five is to seat them at one and the same table, spread the ledgers out side by side, and answer the question Chapter Three laid down and Part Four's close reaffirmed: in speed of convergence, in cross-subject scope, and in resistance to re-convergence, what, after all, is the difference among the three forms of reason? And once those differences have been weighed out, one question more fundamental still remains — since all three are products of convergence, by what title do they deserve to be trusted?

This chapter does the unifying; Chapter Seventeen does the examination; Chapter Eighteen looks ahead.

One Mechanism, Reaffirmed in One Sentence

The foundation of the unification was laid in Part One; here it need only be reaffirmed in a single sentence.

Axiom A6 of the RC paper (consensus reinforcement): observations across subjects and levels verify one another, forming a positive reinforcement loop, converging into a stable objective reality. Chapter One fitted this axiom with all its parts: the Ground, observation, locking, snapshot, account, available margin. Chapter Two gave it its form in time — observational consensus is summarized into causal laws, causal laws into a rule framework, the framework presents itself in reverse as a priori, and the whole is maintained by the continuous feeding of dissipation and the closed loop of consensus. Chapter Three gave the classification: there are not three rationalities; there are three employments of one mechanism of convergence, and the entire difference condenses into one sentence — the object of convergence and the interlocutor differ.

The twelve chapters that followed sent this sentence into history along three routes, sorted by interlocutor. The accounts are now called in; the three routes can be audited one by one.

Logic: Explicitation, Locking, and the Inner Limit

The four graduations of the logic route were held one apiece by the chapters of Part Two.

Chapter Four gave the starting point: on the walking path of Athens, inference became for the first time an object of observation. Aristotle drew the forms repeatedly vindicated in the practice of disputation out of their particular content, and — with variables, with the classification of figures and moods, with the counterexample strike-down method, an entire technical apparatus — made "validity" checkable independently of "what is being talked about". The point of that revolution lies not in the conclusions but in the object — the explicitation of pure structures of convergence had its first historical specimen. Chapter Four's verdict deserves to be brought back as it stands: form was not "discovered"; it was made by a set of techniques.

Chapter Five gave the locking: Frege and Russell compressed inference into mechanically executable transformation rules. That chapter left with this book's logic file its standing formulation — a formal system is the symbolization of an already-locked structure of convergence: axioms and rules are the explicitly locked achievements of convergence, and theorems are future deposits mechanically replayed. The symbolic language is an artificial accelerator of consensus reinforcement: it does not change the mechanism of A6; it adjusts the gear clearances of the mechanism to industrial precision — disagreement is not forbidden; it is left nowhere to attach. The same chapter left a piece of testimony about the boundary: even a symbolic language born for the elimination of disagreement could not exempt its own history of adoption from path dependence — it was not the best notation that won, but the notation most absorbable by the account of the time, and what the Peano lineage held over Frege's two-dimensional notation was not expressive power but the feasibility of absorption by the thickest account.

Chapter Six gave the genesis of the crack: once a system has gained a sufficient capacity for self-observation, the observer falls into his own range of observation. "The set of all sets" was an illegal total observation — the total snapshot cannot catch the hand that presses the shutter. The two patches, the theory of types and Zermelo's axiomatization, did one and the same thing: declare total observation an illegal operation. That chapter's fine qualification deserves to be reaffirmed here: this was not a re-convergence — no old consensus was overturned; it was a preventive planning, a preserving of margin for the system.

Chapter Seven gave the limit, and with it the heaviest verdict of the logic route: Gödel is not the failure of reason but the strictest proof, inside the world of symbols, of the thesis that determinacy equals what is locked. The second incompleteness theorem was filed away by Part Two in one sentence — Gödel's result is not a counterexample to A6 but A6's inner limit: convergence cannot complete itself from inside a single point; verification must cross levels — a system's determinacy was never a certificate it issued to itself. And what is locked necessarily carries the margin that "it could have been locked otherwise" — Axiom A7 — and the undecidable propositions and the undefinable truth predicate are the registered residence of this axiom in the world of symbols. Chapter Seven's closing sentence can be carried straight into this chapter's ledger: of the three forms of reason, it is the first to know, in the form of a theorem, where its own boundary lies.

Put the four graduations together and the portrait of logic is complete: it makes the structures of convergence of the practice of inference explicit, locks the explicit structures fast into symbol systems, sees itself at the place of the locking, and measures, with the rigor of mathematics, the distance between locking and Ground — a distance that never closes. The appearance of the absolute is exactly what convergence looks like at its most complete (Chapter One); and logic pushed this law to its maximum — even the form of negating it has been formatted by it.

Mathematics: Closure, Expansion, and the Bookkeeping of the Cracks

The four graduations of the mathematics route were likewise held one apiece by the chapters of Part Three.

Chapter Eight gave the site: Euclid's axiomatic method is the technique of explicitly publishing the starting point of a convergence path, the oldest specimen of the artificialization of consensus reinforcement — once the starting points are published, verification no longer needs to re-walk the road of discovery; it needs only to check the road that is given. Two thousand years of cross-linguistic, cross-civilizational re-verification without failure saved up the thickest single account of the ancient world; and zero failures is not the merit of the building but the merit of its site — a book that no longer converses with the world cannot be revised by the world. The appearance of "eternal truth" was thereby taken apart: postulates are only starting points not yet questioned, not self-evident truths; the fifth postulate is the one brick in the building of which one can still see that "it was laid on".

Chapter Nine gave the hinge: the birth of non-Euclidean geometry walked the whole road of "observational divergence, system rebuilt elsewhere, re-convergence" — the textbook case of the dynamics of General Outline 1.4. The two-thousand-year failure to prove the fifth postulate sprang from the will to argue that the starting point of one convergence path is the only possible starting point — impossible in principle, because locking determinacy does not exhaust the Ground (A7). That chapter filed away two formulations for this book, and this chapter's table will quote them directly as the mathematics row: first, the choice of axioms = the choice of convergence paths — the adopting and discarding of postulates is the selecting of a path, not the pronouncing of a truth; second, mathematical necessity = self-consistency inside the snapshot — "necessary" says exactly the freedom from contradiction of what is derived, by explicit paths, from a chosen starting point, and not one grain more. Necessity halts at the boundary of the snapshot; and sixty years later, the world picked one frame from among Riemann's several geometries.

Chapter Ten gave the profile: the common form of the three crises — incommensurability, the infinitesimal, the paradoxes of set theory — is the "foundation" at some level of convergence being pierced by observation into the next level down, and the old "it has always been so" standing out as a locking never re-audited. The three foundational stances are three different snapshots, three strategies of re-convergence: formalism demotes "truth" to "self-consistency"; intuitionism makes the presence of locking the criterion of citizenship in mathematics; Platonism reads the thick account of consensus reinforcement backward into a proof that the object of the account preceded the account. No one of the three ever stepped outside the boundary of his own snapshot — and the building grew all the while.

Chapter Eleven gave the destination: structuralism is mathematics' self-understanding returning to "convergence paths". Isomorphism is different realizations of one and the same structure of convergence, a category is the law of transformation of structures of convergence — mathematics at last recognized the object of its study as its own method. The choice of axioms thereupon came to the front of the stage and became the daily work: the continuum hypothesis independent of the foundations in force, the community's grammar of practice become: choose a path, publish it explicitly, sound out its boundary, keep the interface for changing roads. And the introduction's first puzzle was here paid: the effectiveness of mathematics for the world owes no debt to miracle — it owes the debt of a common history of formation and the wages of one re-convergence after another.

Put the four graduations together and the portrait of mathematics is also complete: it walls its convergence paths in by axiomatization and closes itself off, gains within the closed territory the freedom of expansion, and every crisis is not the price of that freedom but its bookkeeping — margin is not mathematics' disgrace but the proof that mathematics can still re-converge. Chapter Eleven's closing sentence is brought back as it stands: every building she raises is inwardly necessary, and her freedom lies wholly in the exchangeability of the foundations.

Science: Control, Notarization, and the Changing of Accounts

The four graduations of the science route were likewise held one apiece by the chapters of Part Four.

Chapter Twelve gave the keel: experiment is controlled observation — isolating a fragment of nature from the mixed stream of observation, shutting the irrelevant variables out one by one, cleansing the relation under interrogation into a weighable series of numbers, so that anyone, anywhere, can walk the same path by the published steps. Controlled observation solves the minimization of noise; publicity solves the maximization of verification; taken together, they turn A6 from an accident into an institution: results public, method reproducible — these two meta-norms of the scientific community are Axiom A6 institutionalized. Bacon supplied the program, Galileo the craft, Boyle the charter, Newton the synthesis — and heaven and earth thenceforward kept one shared account.

Chapter Thirteen gave the notarization: prediction is the public locking of a convergence path before the observation occurs, and the handing of it over to the world for verdict — the most stringent cross-subject test. Halley had been dead for sixteen years when a German farmer he had never met redeemed it in a yard on Christmas night; Neptune was born on paper first and claimed by the telescope after. A successful prediction is a publicly locked path receiving the world's stamp; a failed prediction is the formal entry of observational divergence into the snapshot. That chapter also stood up one of the most-used concepts in this whole book: theory is a cognitive relay station — a theory's value lies not in being forever true, but in converging vast numbers of observations into workable determinacy at the least cost; its accounts are depreciated from the first by the triple boundary, and when it changes hands it does not burn the instruments.

Chapter Fourteen gave the whole drama of the changing of accounts: the zero result of Michelson–Morley was a suspended account — an observational divergence that cross-subject re-verification could not smooth away yet that was, for a time, not booked; the Lorentz contraction was the physics edition of "formal stability at the surface, adjustment of the deep rules" (General Outline 3.4); general relativity was the re-convergence that re-locked the space-time snapshot itself; and the two expeditions of 1919 together with the joint session of the Royal Society were the re-observation ritual of re-convergence — without a public, cross-subject re-observation, the new snapshot cannot be locked. The newspaper headlines read the direction wrong: Newton was not overturned; the old path continues in office as the limiting case — local surgery, the ledger reopened, the world not falling apart.

Chapter Fifteen gave the deepest layer: 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 claim of General Outline 1.2 was redeemed word for word, and the probability amplitude is the most successful mathematical writing of the fact that "locking is under way". Kuhn was set up as the control group: paradigm and rule framework all but isomorphic, revolution and re-convergence the same in form — but RC parts from him at two points: the account is continuous rather than incommensurable, and the account grows rather than changing shoes in place.

Put the four graduations together and the portrait of science is likewise complete: it is the least mysterious and the most honest of the three forms — it has made fallibility into an institution, and pays the tuition of learning with the world's talking back. Part Four's closing sentence is brought back as it stands: 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 Table of Comparison

Now the three ledgers are spread side by side. The table below is the book's unified formulation; every cell is quoted from the verdicts filed away in the earlier chapters, none of it invented here.

One place in the table needs a separate remark, lest the column of "typical events of re-convergence" be misread in the logic row. Gödel's incompleteness is not a re-convergence: Chapter Six already ruled that the two great changes in the history of logic — the theory of types and axiomatic set theory — were preventive planning, not innovation of consensus, and Gödel's result did no more than prove the boundary. It is entered in the logic row precisely because that form stands in the most peculiar situation of the three — the fewest events of re-convergence in its history, corresponding exactly to its greatest resistance to re-convergence; what it handed in to its own history is not the ritual of a changing of accounts but the certificate of a boundary. A system in dialogue with itself meets, at its limit, itself; the deepest self-renewal it can perform is to turn "no deeper" into a theorem. The table is thus also the tabular form of Chapter One's law: the appearance of the absolute is exactly what convergence looks like at its most complete — and the form whose convergence is most complete must use even itself to rewrite itself.

Three Scales

The table sets out the structure; the scales weigh the weight. The three comparisons that Chapter Three announced and Chapter Fifteen reaffirmed are here redeemed one by one.

First scale: speed of convergence. Once logic is made explicit, its convergence is all but complete — Barbara was fully verified on the day it was written down, and the two thousand years since have been a thickening of deposits, not an extension of the path; the formal system pushed this trait to its extreme, the output speed of theorems being limited only by paper and patience. Mathematics' convergence is counted by proofs — a proposition once proved is locked forever within the chosen axioms, but the boundary of the whole discipline (which axioms, which structures) stands forever open. Science's convergence is forever in the present continuous — not one law is "proved"; each is still waiting for the next re-verification and the next counterexample; its account has thickness only, no ceiling. The spectrum is then clear: logic is fastest, because it owes the world nothing; science is slowest, because every entry of its account must wait for the world's reply.

Second scale: cross-subject scope. Logic's community is all who reason; mathematics' community is all who can read a proof; both take their community as unlimited by default — Chapter Three said that this default is their privilege. Science alone fits its community with sluice gates and ledgers: review, assessment, repetition, membership granted selectively, every verification registered conscientiously. Superficially science's community looks smaller; in fact its joint account is opened larger — because what it registers is not only "who agrees" but "who saw what under what conditions"; the finer the ledger, the stronger the transferability. Chapter One said that accounts can be transferred: a modern person's certainty that the Earth goes round the Sun contains not one cent from personal observation. The cross-subject scope of the three forms should be measured by the radius of transfer of the ledger, not by the verbal professions of the members: mathematics' proofs can be transferred to any patience; science's reports can be transferred to any apparatus; logic's rules travel the farthest — so far that they have been taken for the anatomy of thought itself.

Third scale: resistance to re-convergence. Logic is most resistant: the observational divergence required to rewrite it does not exist — it does not converse with the world, and the world cannot refute a rule of inference; nothing can shake it but its own moment of limit, and that limit has been proved to be speakable only from one level higher. Mathematics stands in the middle: a single theorem is in no danger whatever of being overturned within its snapshot, but the foundations can be exchanged — the parallel postulate was exchanged, the continuum hypothesis hangs suspended, and the choice of axioms is itself routine work; mathematics' resistance is concentrated at the layer of theorems and open at the layer of axioms. Science is least resistant — no, that states it backwards; by Chapter Twelve's reading it should be said: science has made fallibility into an institution, and its resistance to re-convergence has been deliberately tuned low — low enough that every talking-back of the world can be booked. On this axis the three forms are not a ranking of better and worse but a division of labor: logic holds the part that does not change; mathematics holds the knob between the exchangeable and the unchangeable; science specializes in translating the world's non-accommodation into updates of the account.

Put the three scales together, and Chapter Three's sentence can be weighed out in full: there are not three rationalities; there are three employments of one mechanism of convergence. The chess player's speed, the architect's stability, the navigator's diligence are three speed curves run out by one and the same A6 in three directions of dialogue.

Applying the Reduction to the Three Themselves

With the comparison done, one step remains — the step the words "theoretical dimensional reduction" in this chapter's title point to: to apply the theoretical dimensional reduction of General Outline 2.3 to the three forms of reason themselves.

RC's General Outline, discussing theoretical dimensional reduction in Section 2.3, says: every theory is an interpretive system under a finite horizon, a dimension-reducing projection of the real world, its validity constrained by the triple boundary — the level of observation determines the theory's limits of representation; the co-construction of the cognizing subject and its environment means that a theory necessarily bears the imprint of subjectivity; and explanatory power decays along the dimension of time; every theory accordingly has a boundary of application. But the dimensional reduction of a theory does not constitute a negation of its value; on the contrary, it establishes its standing as a cognitive relay station — a theory achieves the continuous renewal of cognition through being continuously re-observed.

The three forms of reason, on RC's reading, are all theories — all interpretive systems under a finite horizon. They are not exceptions to the principle of dimensional reduction; they are its three largest positive specimens. Take the boundaries one by one.

Logic's boundary is the meta-system boundary. Chapters Seven and Nine each left a boundary stone: Gödel and Tarski proved that consistency and truth are predicates that can be spoken only from one level higher, that guarantee is always a conferment across levels (Chapter Nine's Beltrami demonstrated the actual operation of this conferment from the other side), and that Tarski's ladder has no top. All of logic's necessity is locked inside the system — and the inside of the system is precisely where its own certificate of safety cannot be issued. This is not a defect of logic; it is a definitional component of "being locked": a system that truly locked away every margin would be unable to lock out even its own freedom from contradiction; it could not so much as exist as a consistent system. Logic is therefore the purest form of the cognitive relay station: what it relays is inference itself, and above it there hangs forever the letter it cannot sign for itself.

Mathematics' boundary is the axiom-choice boundary. Necessity does not cross the boundary of the snapshot — that is Chapter Nine's verdict, and it is the measured circumference of all mathematics' claims to "eternal truth". Within the chosen axioms, mathematics' certainty comes the closest of the three forms to the absolute; but the act of choosing itself lies outside necessity: the parallel postulate was once held self-evident, the continuum hypothesis hangs suspended to this day, and the adopting and discarding of axioms is decided by the community's working. Mathematics' imprint of subjectivity shows itself here as well: Chapter Ten told of the working mathematician — a Platonist on weekdays, a formalist on weekends — and the rotation of three snapshots in use is the most honest shape the subject's imprint takes. Mathematics is the grandest of the relay stations: it relays the most structure, but on every one of its foundations is engraved "exchangeable here".

Science's boundary is the observation-level boundary. The precision of instruments and experiments fixes the precision of the questions that can be asked — this is the first boundary Chapter Twelve nailed fast; the subject's imprint is written into the design of the apparatus, and the decay of time is written into the theory's depreciation schedule. Science's triple boundary is not its disgrace but its self-awareness: it is the only one of the three forms to have written the triple boundary into methodology — error bars, scope of validity, conditions of failure: the relay station's depreciation schedule is posted in public. Science is the busiest of the relay stations: its station sign is changed most often, and for that very reason it is the easiest to see that it is only a station.

Set the three boundaries together, and the reversal this chapter most wants to stand up comes into view. The boundaries of the three forms — the meta-system's, the axiom choice's, the observation level's — are not three different boundaries but three showings of one and the same boundary: determinacy is the locked achievement, locking carries margin, and margin, invisible from inside, can be recognized only in the shape of a boundary. Logic proves it with theorems; mathematics registers it with the choice of axioms; science meters it daily with error bars. Here the three forms are no longer three professions but three testimonies to one and the same thing.

And the second half of General Outline 2.3 thereby receives its largest footnote: the dimensional reduction does not negate value; it establishes precisely the standing of the cognitive relay station. To say that logic, mathematics, and science are all limited is not to pull them down from the altar; it is, for the first time, to say clearly where each of them stands — the value of a relay station lies not in being forever true, but in converging vast numbers of observations into workable determinacy at the least cost, and in not burning the instruments when the transfer is made. The cognitive history of the human species is the history of this network of relay stations woven ever denser: logic holds the seams of the track, mathematics supplies the countless parallel tracks, science is responsible for broadcasting every arrival of the world. The network has no terminus; Chapter Seventeen will show that this is not a regret but a condition of the design.

Clarifications as Usual, and the Handing On

The customary clarifications, two of them.

First, this chapter is not saying that the three forms are indistinguishable. The differences weighed out by the three scales are real: logic cannot exchange the law of non-contradiction; mathematics can exchange the parallel postulate; science has exchanged even its framework of space and time, twice over. A unified mechanism does not level structural differences — any more than one and the same physical law levels the differences among ice, water, and steam: different histories of temperature, different phases. The whole point of the comparison lies precisely in reducing the differences to computable combinations of variables, not in proclaiming them "essentially the same".

Second, this chapter is not saying that the relay stations can be dismantled at will. Quite the opposite: the accounts of the three forms are real deposits saved up from hundreds of millions of re-verifications, and Chapter One's boundary is here reaffirmed for the last time — to admit that determinacy is locked does not in the least hinder you from picking up the cup steadily, from walking onto the bridge with an easy mind, from designing the antenna by Maxwell's equations. What Chapter Sixteen has always wanted to prevent is not trust but misreading: taking deposits for endowment, taking the ledger for a miracle, taking the relay station for a terminus.

The audit of the ledgers is complete. But one question still hangs, and it has followed this book from the introduction to this chapter: since all three forms are limited, fallible, and dependent on cross-subject re-verification, how much weight, after all, do the words "reason" still carry? By what title does it yet deserve to be trusted — and what should count as its health? This question belongs not to comparison but to examination — Chapter Seventeen hands it to two examiners of the twentieth century: one drew a single criterion for demarcating science; the other pointed out that the criterion itself stood in need of correction. And after them there stands on the horizon a stranger figure still: a machine that is learning to observe. Chapter Eighteen will bring it on — but only to let it speak one opening line.