Chapter 1: Inherited Forms

The idea at the center of this essay — that an organization’s meaning should live in a durable form while its applications, models, and even its people come and go — was not invented here. It is among the oldest ideas in philosophy and among the most quietly consequential in twentieth-century science. What is new is not the idea but the physics we propose to give it. This chapter traces the inheritance as a relay of corrections, each thinker repairing a flaw in the bequest of the one before: Plato supplies the architecture; Aristotle moves form into function; Langer makes the form an instrument; Deleuze dissolves its essence; Shannon makes distinction countable by declining to discuss meaning at all; and Deutsch and Marletto return meaning to physics as a counterfactual property, handing this book its opening question at the door.

I. The First Projection Architecture

There are three beds in the tenth book of the Republic. The first is the Form of the bed — made by no craftsman, unique, changeless, graspable by intellect alone. The second is the carpenter’s bed, made of wood by a man who works, Plato says, with his eye on the Form. The third is the painter’s bed: an image of the carpenter’s, an imitation of an imitation, standing at third remove from what is real. The carpenter does not invent the bed. He consults something prior and independent of him, and produces from it a transient instance — one bed among many, subject to woodworm and fashion, while the thing consulted stays exactly what it was.

Read with modern eyes, this is a projection architecture, and it is the first one on record: a durable template; instantiations derived from it; and a strict direction of dependency, in which instances answer to the template and never the reverse. The theory generalizes far beyond furniture. The Forms are eternal, unchanging, and outside space and time; particular things participate in them and are intelligible only through them; the famous images of the Cave and the Divided Line rank the whole of experience by its distance from them. Strip the metaphysics and an insight remains that every schema designer has independently rediscovered: meaning does not live in instances. Any particular bed is a poor guide to what a bed is, because the instance mixes essence with accident — this bed is also heavy, also oak, also broken. Only the template holds the meaning unmixed.

Give Plato his due in the vocabulary of this book, because it is considerable. He got the direction of dependency right: the template governs the instance. He got the cardinality right: many beds, one Form — many projections, one assembly. He even got degradation-by-distance right. The painter’s bed, a projection of a projection, is where fidelity goes to die, and every organization knows this figure intimately: the slide summarizing the report generated from the application built on the schema is the painter’s bed of enterprise life, at third remove and proud of its colors.

Ask what maintains the Forms, and the theory gives a serene answer: nothing maintains them, because nothing needs to. They are, in Plato’s phrase, themselves by themselves — unmade, untended, outside time. Durability is their essence, and so persistence is free. But because persistence is free, there is no feedback anywhere in the architecture. Nothing in the world of instances can correct a Form, and nothing ever checks one; the dependency is not merely directional but absolute. In the vocabulary this book will develop, Plato’s realm is a non-load-bearing representation: perfect, and connected to nothing that would fail if it were wrong. It is here that this book parts company with the theory, and the parting can be made with respect, because the theory of Forms has itself proved remarkably durable — twenty-four centuries of objection, beginning with Plato’s own in the Parmenides, and it is still with us. But notice how it survived: not in the manner of a Form, by standing outside time, but in the manner this book will defend — by being taught, contested, and re-instantiated in every generation that argued with it. The theory endures by the very mechanism it has no room for.

The Platonic assumption — that a sufficiently true form preserves itself — has enjoyed a long afterlife in software. Every enterprise ontology scoped by coverage (model the domain, completely, correctly, once) is Republic metaphysics in XML, and its builders inherit the assumption along with the architecture; whereas nothing persists by nature. Things persist by being tended, and they are tended only when their divergence from reality costs somebody something. The Semantic Web, Chapter 4 will argue, faltered exactly there: not for lack of truth but for lack of consequence. This book keeps Plato’s architecture — the durable template above transient projections — and amends his physics. Durability is not an essence. It is an achievement, purchased continuously, and the currency is enforcement.

A generation later, Aristotle relocated form from a separate realm into the working thing. The form of an axe, he argues in effect in the De Anima, is not an ideal axe elsewhere; it is the capacity to chop — take away the chopping and what remains is an axe in name only, the way a sculpted eye that cannot see is an eye only by courtesy. Form became function; essence became what a thing does when it is working. For our purposes this is the hinge of the whole inheritance, because once form lives in functioning, form can fail — an axe goes blunt — and failure can be observed, attributed, and repaired. The possibility of maintenance enters philosophy. What Aristotle lacked was any account of representations — forms declared, recorded, and consulted, rather than embodied — as themselves functional things that work or fail. That account would wait a very long time.

II. Langer: The Form as Instrument

In 1942, in Philosophy in a New Key, Susanne Langer proposed that the fundamental and continuous activity of mind is symbolization — the transformation of experience into symbols — and drew a distinction this book will lean on twice. Signs (she later preferred signals) announce their objects: the bell announces dinner, and the dog responds to the announcement. Symbols do something categorically different: they are, in her formula, “vehicles for the conception of objects.” A creature that merely registers dinner can only await it. A creature that conceives dinner can plan one, postpone one, promise one, mourn one. The symbol does not stand in front of the thing as a proxy; it makes the thing thinkable.

For a book about organizational meaning, the force of this is that it breaks the naive reference picture — the assumption that a representation is a passive stand-in whose entire job is to match what it represents. If symbols are instruments of conception, then a representation is a tool that does work: it renders a domain thinkable, composable, plannable. And like any tool it can be sharp or dull, in daily use or in a drawer. Meaning stops being a static correspondence between marks and things and becomes a practice that is either happening or not. Langer had come to philosophy through logic — her early books were on symbolic form and symbolic logic — and she held that what a symbol shares with what it symbolizes is precisely logical form: pattern, structure, the shape of relations, abstracted from any particular material. A LogicalAssembly is, in her exact sense, an organization’s declared logical form — the pattern by which it renders itself thinkable to itself, and now to its agents.

Her second distinction bounds this book as usefully as the first licenses it. Language, Langer argued, is discursive symbolism: linear, built from a vocabulary by a syntax, translatable, paraphrasable — meanings strung one after another like beads. But much of what we know arrives otherwise: in the presentational symbol — the picture, the melody, the rite, the face of someone you have known for years — which delivers its meaning all at once, as a total structure whose elements mean nothing in isolation and which no paraphrase can replace. Langer refused to let the discursive claim the whole of meaning, and the refusal does double work here. It licenses the project: an organization’s concepts, relations, constraints, events, and policies are discursive through and through — articulable, composable, versionable, checkable — and nothing in their nature resists formalization. And it bounds the project: some organizational knowledge — timing, trust, which constraints are actually negotiable and for whom — is presentational, held in acquaintance rather than articulation. Chapter 10 patrols that boundary. We note here only that it was drawn, with great care, in 1942.

Langer’s correction of Plato can now be put in one sentence: forms are not the furniture of a separate realm but the instruments of a living activity, and a symbol that no one is using to conceive anything is not a dormant meaning but a dead one. She moves the weight of the theory from being to use. What she does not supply is the mechanism by which use disciplines the form — how employment corrects a symbol that has drifted from what it symbolizes. Her symbols live in use; they do not yet fail in use. For that, one more dissolution and one more physics are required.

III. Deleuze: Assemblies Without Essences

If Langer loosened Plato’s grip, Gilles Deleuze set out to break it, and said so: he announced the task of modern philosophy as the overturning of Platonism — the abolition of the hierarchy of model and copy, original and imitation. The constructive proposal, developed with Félix Guattari in A Thousand Plateaus, is the agencement — rendered in English, with less music, as assemblage: a whole composed of heterogeneous parts whose unity comes not from a shared essence but from co-functioning. The canonical miniature is the knight: man, horse, stirrup — flesh, animal, iron, three things with nothing essential in common — composing a unit of medieval warfare because the stirrup lets a charging man deliver a lance-blow without being thrown. Remove the stirrup and the assemblage decomposes. No essence is harmed, because none was ever involved.

Two features of assemblages matter to this book. The first is what Manuel DeLanda, the great systematizer of the concept, calls relations of exteriority: a part is not defined by its membership in the whole. It can be detached and plugged into a different assemblage — the horse into agriculture, the man into a phalanx — and the whole’s identity therefore lives in the pattern of co-functioning, not in the natures of the parts. The second is that assemblages are processes, not states. They territorialize — stabilize, sharpen their boundaries, harden their identity — and they deterritorialize — leak, drift, decompose. Stability is an ongoing achievement won against dissolution, or it is nothing. Readers will recognize the picture this book draws of organizations, and of representations, arriving here from an unexpected direction and twenty years early.

A third Deleuzian gift is planted here and harvested in Chapter 9. In Difference and Repetition, Deleuze distinguishes the virtual — real but not actual: a structured field of potential that genuinely exists — from its actualizations, which do not copy it but differentiate it, creatively, along diverging lines. He opposes this pair to the classical couple of the possible and the real, in which the possible is merely the real minus existence, waiting to be rubber-stamped into being. The relation between a LogicalAssembly and its projections — the applications, agent configurations, and reports derived from it — is nearer to virtual and actual than to template and copy. Projections are not clones of the assembly. They are divergent actualizations of it, each developing the domain’s potential in a direction the assembly does not itself contain, which is truer both to practice and to what makes the projection worth generating.

We do not follow Deleuze all the way down. This essay retains a hierarchy — the assembly is prior to its projections; a domain change is made once, in the assembly, and propagates outward — and an unreconstructed Deleuzian will smell Platonism in that. They are half right, and the half matters. The difference is the ground of the priority. The assembly is not prior because it is more real than its projections, closer to essence, or ontologically upstream. It is prior because it is the enforced narrow waist: the point through which the most consequences flow, and therefore the point at which divergence is made to fail first and most visibly. Its authority is positional and earned, not essential and given — let the enforcement lapse, and the assembly becomes an ex-monarch, all lineage and no army. Deleuze is thus the discipline that keeps our Platonism honest. A Lasm is an assemblage that has been deliberately and continuously territorialized — heterogeneous parts, schemas and predicates and event definitions and evaluation harnesses, co-functioning and held stable by enforcement.

IV. Shannon: The Productive Deferral

In 1948, in “A Mathematical Theory of Communication,” Claude Shannon opened by setting meaning aside: messages frequently have meaning, he allowed, but the semantic aspects of communication are irrelevant to the engineering problem, which is to reproduce at one point a message selected at another. From that single act of renunciation came nearly everything: the bit, the atom of distinction; entropy as a measure of a source’s freedom of choice; channel capacity, coding, redundancy; and the substrate independence that made all subsequent digital civilization interoperable. Warren Weaver’s gloss in the companion volume remains the best one-sentence account of the theory’s spirit: information relates “not so much to what you do say, as to what you could say.” The theory is about the space of possible messages; the actual message is almost incidental.

Consider what the renunciation deferred, and to whom. By bracketing meaning, Shannon made quantity tractable — and the bracket held for seventy years because at every endpoint of every channel there eventually stood a human being in whom the meaning could reside. The division of labor was so complete that it was never stated: the network moves distinctions; the human heads at the ends hold what the distinctions mean. The arrangement worked. Its cost was real but diffuse, itemized under headings nobody totaled — onboarding time, documentation, integration projects, the tribal knowledge that walks out the door in every resignation. The debt was invisible because the interest was paid in small installments everywhere. This book’s premise is that agentic systems present the whole note for payment at once: when the endpoint of the channel acts instead of reads, the meaning must be at the endpoint — represented, consultable, at runtime — or the action proceeds on a defective model of the world.

When information theory became a fashion in the mid-1950s — pressed into service in psychology, biology, linguistics, art criticism, management — Shannon published a one-page editorial, “The Bandwagon” (1956), gently suggesting that the theory’s importance had ballooned beyond its accomplishments, that its core was and remained an engineering matter, and that workers in other fields should adopt it only where it genuinely fit. He had cut meaning away knowingly, and he declined every subsequent invitation to pretend the cut had never happened. The discipline this book attempts — claims sized to their tests, a framework offered with its failure modes attached — takes inspiration from Shannon and his modesty.

Information as selection presupposes the set selected from: a message informs only against the background of the messages that were possible, and both parties must hold that background in common before a single symbol moves. The possibility space — shared, structured, prior — is where all the meaning Shannon declined to discuss silently resides. Pull on that thread and the theory of information starts to become a theory of possibility. Two physicists pulled it.

V. The Counterfactual Turn

David Deutsch’s constructor theory, proposed in 2012, reformulates the ambition of fundamental physics: instead of predicting what will happen from initial conditions and laws of motion, express all laws as statements about which transformations are possible, which are impossible, and why. The constructor — anything that causes a transformation while retaining the ability to cause it again: a catalyst, a robot, a factory — is the theory’s namesake but not its subject; it is abstracted away, leaving the tasks themselves as the basic entities. In 2014, with Chiara Marletto, Deutsch brought information into the frame, beginning from a diagnosis: the existing theory of information is circular at its foundation. Information is defined in terms of distinguishable states, and distinguishability is defined by what a measurement lets you learn — each concept leaning on the other, with Shannon’s theory never specifying what the act of distinguishing physically consists of. Their repair grounds the whole structure in tasks. Computation is defined first, as possible permutations over a substrate’s attributes; information variables are then defined as cloneable computation variables; distinguishability arrives afterward, downstream and non-circular. The result is that information becomes a physical property constituted by which transformations are possible on its carrier — counterfactual through and through. An object carries information, on this account, only if it could have been otherwise; what it is matters less than what could have been done with it.

Deutsch’s 2012 paper contains a further definition, easy to pass over and central to this book. Among constructors, one abstract kind stands out: knowledge, defined — after Popper, and pointedly without a knowing subject — as information which, once physically instantiated in a suitable environment, tends to cause itself to remain so instantiated: it survives noise, error, criticism, and the replacement of its carriers. Genes are the canonical case, causing the environment to build fresh instances of the very organisms that carry them. And Deutsch’s reading of the Ship of Theseus completes the thought: the planks are replaced, the generations of Athenians are replaced, the scroll bearing the specification is recopied — and the only invariant across the whole process is the abstract program that causes Athens to keep re-instantiating the ship, the scroll, and the shipwrights. The information is the constructor. Durability, in this picture, is not resistance to change; it is a causal loop that reproduces a pattern through change. Plato’s eternity has been replaced by a fixed point of maintenance.

Notice what has happened to the inheritance. The durable form has returned — Deutsch and Marletto are unembarrassed realists about abstractions that make a physical difference — but every Platonic attribute has been replaced by its earned counterpart. Not eternal, but continuously re-instantiated. Not self-subsistent, but loop-dependent. Not beyond all testing, but constituted by tests — by the tasks that are and are not possible on it. The counterfactual turn completes Langer’s functional turn by giving it a physics: a representation’s meaning is what can be done with it, and its durability is whether what diverges from it fails.

One gap remains, and the book lives in it. Physics gets its loops for free. Genes are error-corrected by differential survival whether or not anyone budgets for the process; tested scientific knowledge is re-instantiated by a community whose incentives happen to run that way. Organizations get nothing for free. Their representational loops — the checks, the gates, the alarms that make divergence fail visibly — must be designed, funded, staffed, and defended against every quarter’s pressure to spend elsewhere. The law is the same in both cases: what persists is what is re-instantiated by a loop. The difference is that nature’s loops are found and ours must be built. Deutsch and Marletto describe where self-maintaining information exists; this book is about constructing it where it does not. It is, to put the genre exactly, an engineering essay written under a physical law.

VI. What Remains

Set side by side, the corrections compose a single figure. The durable form survives every revision of the inheritance — no thinker on this line abandons it — but each revision strips it of another unearned property. It loses its separate realm and moves into the working thing. It loses its self-evidence and becomes an instrument that must be used in order to mean. It loses its essence and becomes a co-functioning that must be maintained to stay whole. It loses its innocence about meaning and becomes a possibility space that must be shared before a single message moves. And it loses, at the last, its immortality: whatever persistence it has, it has because some loop keeps re-instantiating it.

What remains is the figure the rest of this book will state less grandly and test more carefully. A durable form is not a truth that persists. It is a structure that is kept: a functional assemblage of distinctions, consulted by things that act, whose divergence from reality is made to fail — visibly, attributably, and early — before it can spread. The next chapter returns to the smallest item in the inheritance, the bit, and weighs it properly. It is the piece everything else was purchased with.