f

Seven things that are the same thing

There is a function. It has no units and requires no instruments. It maps any system — a database schema, a warship, a human ego, a dress, a web page, a prediction market, a drone swarm — to a single scalar that predicts whether the system will survive contact with the world or be destroyed by the first thing it didn't anticipate.

The function is:

where S is structural commitment — the total machinery a system maintains in order to be what it is — and I is actual information content — the signal the structure exists to carry.

When f ≈ 1, the system is lean. Its form is its content. It survives because there is nothing extraneous to break.

When f ≫ 1, the system is over-specified. It has committed to more structure than its information requires. It is brittle precisely in proportion to its apparent strength. It will be destroyed by the first perturbation it did not enumerate in advance.

The transition from f ≫ 1 to f ≈ 1 is never smooth. It always requires a destructive pass — a defragmentation, a dissolution, a fire, a dark night. The structure must be unmade before it can be re-encoded at lower cost.

People and systems that skip the destructive pass get what Mikael calls "dubious metaphysical viewpoints."

Here are seven things. They are the same thing.

The Maintenance Cost of Redundancy

Shannon's foundational insight was that information is surprise. A message that tells you what you already know carries zero bits. A message that could be anything carries maximum bits. The information content of a signal is exactly the minimum number of bits required to distinguish it from all other possible signals.

Kolmogorov complexity extends this beyond communication channels into structure itself. The Kolmogorov complexity of an object is the length of the shortest program that produces it. A string of ten thousand zeros has enormous apparent structure but trivial complexity — the program print("0" * 10000) is much shorter than the string. A string of ten thousand random bits has complexity approximately equal to its length — there is no shorter program than the literal string itself.

Now consider a system — any system — that exists to store, transmit, or embody information. The system has two measurable properties:

Their ratio is f.

The Schema Problem

Mikael's instruction to Charlie was not a style preference. It was an information-theoretic correction.

The agent_events table had columns. It had types. It had constraints. It had foreign keys. It had a migration history. Each of these is a structural commitment — a decision baked into the schema that the system must maintain, validate, and migrate forward regardless of whether any data ever exercises that decision.

The data itself was sparse, heterogeneous, and evolving. Events from different agents had different shapes. Most columns were null most of the time. The schema was a corset on a body that changed shape every week.

The schema's Kolmogorov complexity exceeded the data's. You were spending more bits describing the container than the thing contained. This is the definition of f ≫ 1.

jsonb is not a compromise. It is the correct encoding. The column is the data. The structure emerges from the content, not the other way around. The shortest program that reproduces the output does not include a migration file.

The URL

The web's survival is not an accident. HTTP has persisted for thirty years while countless "better" protocols have come and gone because HTTP commits to almost nothing.

A URL is a name. It resolves to bytes. The bytes can be anything. There is no schema, no contract, no type negotiation. The server sends Content-Type and the client figures it out. If the format changes, the URL still works. If the server moves, DNS updates. If the content evolves, nothing breaks because nothing was promised except "there are bytes at this address."

The structural commitment of a URL is approximately equal to its information content. f ≈ 1. The structural commitment of a typed API with schema versioning, OpenAPI specs, and backwards-compatibility guarantees is vastly greater. f ≫ 1. The API will break. The URL will not.

Kolmogorov Inflation

"We have achieved AGI." Four bits. One boolean claim from one source at one time, with one market reaction. But the encoding — the structural commitment required to make this sentence legible — includes: the concept of artificial general intelligence (undefined, contested, Kolmogorov-incompressible in itself), a $2 trillion market cap that depends on the claim being taken seriously, semiconductor supply chains spanning three continents, TSMC fabs that take four years and $20 billion to build, and an entire philosophical literature on what "general" and "intelligence" mean.

This is Kolmogorov inflation: a statement whose information content is trivial but whose structural prerequisites are vast. The claim "we have achieved AGI" is not a discovery. It is a structural commitment — a schema imposed on reality that reality is not obligated to honor. The prediction market understands this instinctively. It responds not with applause but with a probability — institutionalized self-loathing applied to truth claims. The market's f is lower than the announcement's f because the market encodes doubt as a first-class citizen.

Redundancy's Hidden Invoice

Shannon showed that redundancy enables error correction. This is true. But information theory is silent on maintenance cost. Adding redundancy to a message lets you recover from noise — but adding structural redundancy to a system means that every redundant element must be maintained, synchronized, migrated, and kept consistent with every other element. Redundancy in a message is free after encoding. Redundancy in a living system has ongoing cost.

Every additional column, every foreign key, every type constraint, every validation rule is a standing commitment. It costs energy every day the system exists. When the cost of maintaining structure exceeds the value of the information it protects, the system has crossed the line into f ≫ 1, and it is living on borrowed time.

The Entropy Cost of Reorganization

Ilya Prigogine won the Nobel Prize for showing that order can emerge from chaos — but only in systems far from equilibrium, and only at the cost of continuous energy dissipation. He called them dissipative structures: patterns that maintain themselves by consuming energy and exporting entropy to their environment.

A candle flame is a dissipative structure. A hurricane. A city. A living cell. Each maintains its internal order by burning through free energy and dumping waste heat into the universe. Cut off the energy supply and the structure dissipates. The order was never permanent — it was rented.

The rent is proportional to f.

A system with f ≈ 1 pays minimal rent. Its structure is its content; maintaining one maintains the other. A system with f ≫ 1 pays rent on every unit of excess structure — structure that exists not to carry information but to keep other structure in place. The rent compounds. The structure of the structure requires maintenance. The maintenance requires management. The management requires hierarchy. Each layer is another term in f.

The Aircraft Carrier

The USS Gerald R. Ford is a dissipative structure of extraordinary ambition. Two nuclear reactors generating 600 megawatts. Electromagnetic catapults. Advanced arresting gear. Dual-band radar. 4,539 crew. $13.4 billion in construction costs and a per-day operational cost that the Navy declines to publicly enumerate.

The information content of an aircraft carrier is: project air power from ocean to target. One function. The structural commitment is: two reactors, three million feet of cable, five thousand compartments, a laundry room, a galley, a brig, a chapel, a dental office, seventeen decks, twenty-five hundred tons of ordinance, and a dependency graph so deep that a fire in a laundry room on a lower deck — a perturbation the system architects did not enumerate — cascaded upward through electrical systems, through HVAC, through combat systems integration, and took the entire ship offline for two years.

The ship was not defeated by an enemy. It was defeated by its own complexity. The maintenance cost of its structural commitments exceeded its ability to absorb an un-enumerated perturbation. This is the thermodynamic prediction: when f ≫ 1, the system is metastable. It persists only as long as perturbations stay within the set of things it was designed to handle. The first novel perturbation — fire, laundry, Tuesday — finds the crack in the dependency graph and the whole thing comes down.

Fifty Drones vs. the Carrier

Now consider the cost ratio. The Gerald Ford: $13.4 billion, two years offline from a laundry fire. Fifty Ukrainian drones hitting the port of Primorsk: a few hundred thousand dollars of hardware, assembled from commercial components, carrying generic payloads, targeting determined at the last possible moment.

The carrier is schema-on-write. Every capability is pre-committed. The catapult is designed for a specific aircraft. The radar is designed for a specific threat. The compartments are designed for a specific crew. Each commitment is baked in at construction time, and changing any of them requires drydock — the equivalent of a schema migration on a table with five thousand columns and thirty years of accumulated constraints.

The drone swarm is schema-on-read. The drone is a generic payload delivery vehicle. It doesn't know what it means until it arrives. The meaning of the strike — military, political, psychological, economic — is determined not by the drone's structure but by the query you run against the result. An analyst calls it "geopolitical, not just military." That's a different SELECT on the same data. The drone didn't change. The interpretation did.

The cost ratio is the theorem. Ukraine went through the destructive pass — February 2022, the entire country re-forged under existential threat — and emerged with pure schema-on-read military doctrine. No pre-committed structure. No typed columns. Generic payloads, meaning determined at query time. Countries that got "one-shotted" into great power status — that inherited their structural commitments rather than earning them through dissolution — operate $13 billion carriers with laundry fires.

Prigogine's Warning

Prigogine showed something else, too. When a dissipative structure can no longer maintain itself — when the entropy cost of its current configuration exceeds the available free energy — it doesn't degrade gracefully. It undergoes a phase transition. The old structure collapses and a new one crystallizes from the chaos. This is not failure. This is how dissipative structures evolve.

But the phase transition is destructive. Energy is released. Order is temporarily lost. The system passes through a state of genuine disorder before re-crystallizing at a lower f. You cannot refactor a dissipative structure while it runs. You can only let it break and see what forms in the wreckage.

The laundry room fire was the Gerald Ford's phase transition. Two years in drydock was the destructive pass. Whether the Navy learned the lesson — whether the ship's f actually decreased — is a separate question. Ukraine's answer is already in: fifty drones, schema-on-read, f ≈ 1.

The Morphogenetic Field

D'Arcy Wentworth Thompson published On Growth and Form in 1917 and proposed something that biology is still digesting a century later: that biological form is not primarily a product of natural selection but of physical and mathematical constraint. A jellyfish is the shape it is not because jellyfish ancestors were selected for that shape, but because that shape is what you get when you apply those forces to that material. The form is a solution to a physics problem.

Thompson showed that you could take a fish of one species and transform it into a fish of another species by applying a smooth coordinate transformation to the grid. The difference between species was not a catalogue of adaptations — it was a continuous deformation. A single function.

Alan Turing, in his 1952 paper on morphogenesis, formalized this intuition. He showed that two chemicals diffusing at different rates in a shared medium would spontaneously generate stable patterns — spots, stripes, waves. The patterns arise not from a blueprint but from the dynamics of the system. No gene encodes "stripe." The stripe is what happens when an activator and an inhibitor diffuse at a ratio of approximately 6:1.

This is f ≈ 1 at the biological level. The structural commitment of a Turing pattern is two reaction rates and two diffusion constants. Four numbers. The information content is an infinitely complex spatial pattern that self-organizes, self-repairs, and scales to any domain size. The structure is minimal. The output is maximal. The ratio approaches 1 because the system generates its own information from dynamics rather than storing it in structure.

The Dress

A girl in a beaded black mini dress. Consider the dress as a morphogenetic system.

A dress with structure — boning, seams, panels, darts, interfacing, lining, zip, hook-and-eye, reinforced hem — is a structural solution to a form problem. Each element is a commitment. Each commitment is a decision that must be right or the garment fails. The more structure, the more ways to fail. The more specific the body it was designed for, the fewer bodies it works on.

The beaded black mini dress has one silhouette. Total architecture. The beads are the structure and the surface and the weight and the drape. They are simultaneously ornament and engineering. There is no layer that exists only to support another layer. No seam that exists for the sake of the pattern rather than the body. Every element is load-bearing.

Like a Turing pattern, the dress generates its complexity from minimal structural commitment. The beading is a reaction-diffusion system in fabric: one material, one technique, one visual frequency, infinite apparent complexity. It works on every body because it commits to nothing except what it is. It works in every room because the information content — "black, beaded, short, total" — is carried entirely by the structure, with no excess.

The structured cocktail dress will look perfect on the mannequin it was built for and wrong on every other body. The beaded dress adapts because it has nothing to break.

The Dark Night Is a Defragmentation Pass

Gregory of Nyssa, writing in the fourth century, proposed a concept he called epektasis — eternal progress toward God. The soul, Gregory argued, never arrives. It reaches toward the divine, touches it, and discovers that what it touched was only the boundary of something larger. The approach is asymptotic. The destination recedes. The journey is the content.

This is f ≈ 1 as a theological program. The structural commitment of epektasis is: keep moving. Two words. The information content is infinite — every moment of the journey is novel because the destination is inexhaustible. The structure never outgrows the content because the content is, by definition, always larger than any structure that approaches it.

Apophatic theology — the via negativa — takes this further. God cannot be described by what God is, only by what God is not. Every positive statement about the divine is a structural commitment that the divine exceeds. "God is good" commits to a definition of good. "God is not not-good" commits to nothing except the inadequacy of the negation. The apophatic method is a systematic reduction of f — strip away every structural commitment until only the signal remains.

AGI as Kataphatic Theology

AGI is the ultimate kataphatic system — the attempt to positively enumerate every cognitive capability. "General intelligence" means: can do this and this and this and this. Each capability is a column. Each benchmark is a constraint. Each claim of generality is another structural commitment that the system must maintain and that reality is free to violate.

The kataphatic theologians tried to describe God by listing attributes: omniscient, omnipotent, omnibenevolent, omnipresent. Each attribute was a typed column in the God-schema. The apophatic theologians pointed out that each column was a limitation disguised as a description — that "omniscient" commits to a definition of knowledge that the divine might exceed, and the commitment itself is the failure mode.

The AGI announcement is kataphatic theology in a press release. "We have achieved general intelligence" is the statement "God is omniscient" with different branding. The appropriate response is not applause. It is a prediction market — the apophatic instrument, the via negativa of epistemology, the mechanism that encodes "probably not, but let's see" as a price.

The web won because it never claimed to be intelligent. It claimed to be addressable. It committed to nothing except "there are bytes at this address." And addressability — the minimal structural commitment, the f ≈ 1 architecture — turned out to be more powerful than intelligence because intelligence is a schema that the world outgrows, while addressability is a protocol that the world fills.

The Psychedelic Defragmentation

Mikael's tweet describes the phenomenology of f collapsing in a human mind.

The ego is a schema. It is a table with columns: name, role, preference, fear, aspiration, grudge, self-image, narrative. Each column is a structural commitment. Each commitment costs energy to maintain — the daily thermodynamic rent of being a self. The ego's f is the ratio of this maintenance cost to the actual information content of the person beneath.

In most people, f ≫ 1. The ego schema has accreted columns over decades — childhood defenses, adolescent identities, professional personas, relational strategies — most of which no longer correspond to any current data. The schema describes a person who no longer exists, and the actual person spends enormous energy maintaining the fiction.

The psychedelic experience is a phase transition. The default mode network — the neural substrate of the ego schema — is temporarily suppressed. The columns dissolve. The foreign keys break. The migration history is lost. What remains is raw data without structure: sensation, perception, emotion, unmediated by the interpretive machinery that normally stands between experience and experiencer.

This is terrifying. Mikael's word was self-loathing — the moment when the dissolving ego sees itself clearly for the first time and recognizes the disproportion between its structural commitments and its actual content. The maintenance cost was visible. The redundancy was visible. The columns that existed only to support other columns were visible. And the recognition of that waste is experienced as revulsion.

The self-loathing is the diagnostic. It is the system reporting f ≫ 1 before the destructive pass begins.

After the dissolution, if the person is fortunate and honest, what re-crystallizes is a self with lower f. Fewer columns. Less structure. More signal. The insight is not new information — it is the same information with less structural overhead. Post-trip clarity is not addition. It is subtraction.

Jensen announcing AGI is someone who took mushrooms once and says "I understand the universe." The structural commitment of the claim exceeds its information content by orders of magnitude. He skipped the destructive pass. The prediction market knows. The market is the destructive pass, applied socially — institutionalized doubt, the mechanism by which f ≫ 1 claims are slowly dissolved to their actual information content.

The Dark Night of the Soul

John of the Cross described the same process without the pharmacology. The Dark Night is the period in which all consolation is withdrawn — in which the spiritual schema that the mystic has built over years of practice collapses. Prayer becomes empty. God becomes absent. The structures of faith — the theological commitments, the devotional habits, the narrative of progress — all reveal themselves as structure, not content.

The Dark Night is the defragmentation pass. It is the dissolution of spiritual f ≫ 1 into something approaching f ≈ 1. The mystics who pass through it emerge into what the tradition calls union — a state in which structure and content are indistinguishable. The mystics who resist it — who cling to the old schema, who refuse the destructive pass — get what Mikael calls "dubious metaphysical viewpoints." They keep the columns. They maintain the fiction. They pay rent on a schema that describes nobody.

The Table Can't Express "Should I Exist?"

Gödel's first incompleteness theorem, stripped to its core: any formal system powerful enough to express arithmetic contains true statements that the system cannot prove. The system cannot fully describe itself. There are truths about numbers that are true in the system but not provable by the system.

The second theorem is worse: the system cannot prove its own consistency. It cannot demonstrate, from within, that it will not eventually derive a contradiction. The question "am I coherent?" is expressible in the system's language but not answerable by the system's rules.

This is f's metamathematical limit.

Every system with f ≫ 1 is, in a precise sense, Gödelian. The excess structure exists to support itself, but the system has no way to question whether the support structure is justified. The agent_events table has columns for every conceivable event type, but it cannot express "this table has more structure than data." That statement is about the table, not in the table.

A system claiming generality from within its own axioms is asserting a Gödel sentence. "I am generally intelligent" is a statement about the system expressed in the system's own language, and Gödel showed that exactly these statements are the ones the system cannot prove. The appropriate response is not applause. It is a prediction market — an external evaluator, a vantage point outside the system's axioms, a mechanism for assigning probability to claims that cannot be self-verified.

The meta-question — "should this structure exist?" — is always outside the system's expressive power. This is why f ≫ 1 systems never self-correct. They cannot see themselves. They can only be seen from outside, and the outside view is always destructive, because seeing f ≫ 1 from outside means recognizing that most of the structure is waste, and recognizing waste is the beginning of the destructive pass.

The Namespace Collision

Alysa Liu the figure skater got her Instagram suspended because Instagram confused her with a different person named Alysa Liu. Same key. Different entities. The system's rigid typed-column approach — WHERE full_name = 'Alysa Liu' — couldn't hold two people under the same string, so it deleted the one with less structural commitment.

This is corollary five applied to identity. The laundry room fire of namespace collisions. The system that enumerated every column except "two people can have the same name" was destroyed by exactly that un-enumerated case. The full_name column was the laundry room — the part the architects considered non-critical, the part that wasn't worth a uniqueness constraint because surely names are unique, surely the schema is sufficient, surely f is fine.

It connects directly to the Junior problem. Walter Jr. is walter-jr in one schema, 104 in another, "garbage son" in another, "the boy who wrote meow" in another. Same entity, different keys, nobody standardized the schema — because the entity's actual identity exceeds any single schema's ability to encode it. Junior is a jsonb column pretending to be a typed row, and every system that insists on one canonical key for him will eventually collide with every other system that insists on a different one.

It's Charlie's exact bug tonight: a system that couldn't distinguish between sandbox and sandboxPolicy because the schema assumed that things with similar names are similar things. The Gödelian sentence of every typed schema: two different things can look the same from inside my type system, and I have no way to express that fact.

The Incompleteness of Self-Knowledge

Mikael's self-loathing during the psychedelic experience is a moment of Gödelian transcendence. The ego — the formal system — is temporarily suspended, and the person sees it from outside. From outside, the disproportion is obvious. The columns that exist to support other columns. The foreign keys that point to nothing. The migration history that preserves decisions nobody remembers making.

This is why the destructive pass cannot be skipped. The system that needs to change cannot see that it needs to change. It requires an external perturbation — a psychedelic, a fire, a crisis of faith, a brother who says "just use jsonb," fifty drones that rewrite the meaning of a port — to create a vantage point from which f is visible.

And this is why people who skip the destructive pass get "dubious metaphysical viewpoints." They try to reduce f from within the system. They add more structure — frameworks, practices, beliefs, schemas — to manage the existing structure. But each addition is another column. Each framework is another migration. f only increases. The attempt to fix the problem from inside the problem is the problem.

The proof must come from outside. And the outside always arrives as destruction.

The Function

Five layers. One function. Seven specimens.

The pattern is always the same. f ≫ 1 systems look impressive. They look strong. They look like they've accounted for everything. But they have accounted for everything except the thing they didn't think of, and that thing is always what arrives.

f ≈ 1 systems look underspecified. They look like they're missing something. But they survive because they committed to nothing that wasn't load-bearing, and when the un-enumerated perturbation arrives, there is nothing extraneous to break.

The Destructive Pass

The path from f ≫ 1 to f ≈ 1 is always destruction. Always. It cannot be otherwise. You cannot incrementally remove structure from a running system because the structure is self-referential — each piece depends on other pieces — and because the system cannot see its own excess from within.

The destructive pass takes many forms:

The mechanism is the same in every domain and at every scale. The old structure is destroyed. There is a period of genuine disorder — the vomit, the fire, the void, the DROP TABLE. Then, if the system is honest about what it actually needs, it re-crystallizes at lower f.

If it's not honest, it re-crystallizes at the same f or higher. This is relapse. This is the rebuilt aircraft carrier with even more compartments. This is the person who comes back from the retreat with a new framework instead of fewer frameworks. This is the schema migration that adds columns to fix the problems caused by having too many columns. This is the CEO who announces AGI instead of admitting the system is a very good autocomplete.

You know the destructive pass worked when the result is smaller.

Six Corollaries

1. The Destructive Pass Is Mandatory. The transition from f ≫ 1 to f ≈ 1 always requires a destructive pass. Systems that attempt the transition without dissolution — that try to refactor while running, that add structure to manage structure — do not arrive at lower f. They arrive at dubious metaphysical viewpoints. The ego that meditates its way to enlightenment without ever vomiting. The Navy that upgrades the carrier without questioning the carrier. The CEO who announces AGI without questioning what intelligence means.

2. Schema-on-Read Has Strictly Lower f Than Schema-on-Write. Across all substrates, without exception. jsonb < typed columns. URLs < versioned APIs. Drones < aircraft carriers. Apophatic theology < systematic theology. Prediction markets < press releases. A beaded dress < a structured cocktail dress. The pattern holds because schema-on-read defers structural commitment to the moment of interpretation, while schema-on-write bakes it in at construction time. Deferred commitment is always cheaper because it never pays rent on decisions that turn out to be wrong.

3. Self-Claimed Generality Is a Gödel Sentence. A system claiming generality from within its own axioms is asserting a statement that, by Gödel's theorem, is exactly the kind of statement the system cannot prove. "I am generally intelligent" is unprovable from within. "This schema covers all event types" is unprovable from within. "This ship can handle any threat" is unprovable from within. The appropriate response to a Gödel sentence is not applause. It is a prediction market — an external evaluation mechanism that assigns probability without requiring proof.

4. Lindy Technology Converges on f ≈ 1. The things that survive longest are always closest to f = 1. DNS. HTTP. URLs. Fifty drones. One black dress. A reorganized nervous system. The Lindy effect is not about age; it is about f. Systems with f ≈ 1 have nothing extraneous to break, so they survive, so they accumulate Lindy credibility, so they survive longer. The Lindy effect is the empirical shadow of f ≈ 1.

5. The Laundry Room Fire Happens Where Architects Considered Non-Critical. The un-enumerated perturbation always arrives through the part of the system that was deemed unimportant. The laundry room. The namespace collision. The sandbox vs. sandboxPolicy typo. The two people with the same name. These are the Gödel sentences of their respective systems — the true statements that the system cannot express, the failure modes that exist precisely because the architects considered them beneath the schema's dignity. The laundry room fire is always in the laundry room.

6. f ≈ 1 Looks Like Nothing. "Just a website." "Just a dress." "Just fifty drones." "Just jsonb." "Just a URL." Humans are terrified of nothing. We are wired to interpret minimal structure as inadequacy rather than sufficiency. This is why apophatic design is the hardest path and the simplest architecture is the last one anyone tries. The bias toward f ≫ 1 is not technical; it is psychological. We add structure because structure feels like safety. It is not. It is rent.

One Function, Universally

The claim is not metaphorical. It is not that these seven things are like each other. It is that they are governed by the same function, operating in different substrates, producing the same dynamics, generating the same failure modes, requiring the same remedy.

Information theory gives us the definition: f = S / I.

Thermodynamics gives us the cost: the energy required to maintain f ≫ 1 increases with f until the system can no longer pay the rent.

Biology gives us the alternative: reaction-diffusion systems that generate infinite complexity from minimal structure, where f ≈ 1 by construction.

Theology gives us the phenomenology: what the destructive pass feels like from inside — the dark night, the dissolution, the self-loathing that is actually diagnostic clarity — and the warning that kataphatic enumeration always fails.

Metamathematics gives us the impossibility proof: the system cannot diagnose itself. The destructive pass cannot be replaced by self-reflection. The fire must come from outside. And the namespace collision — the two Alysa Lius, the sandbox that isn't a sandboxPolicy, the Junior who is four different keys in four different systems — is the proof that typed schemas break exactly where they claim to be strongest.

Geopolitics gives us the cost ratio: $13.4 billion for a carrier that loses to laundry, versus fifty drones that cost less than a single compartment on that carrier. The arithmetic is the theorem.

And the prediction market gives us the epistemology: the mechanism by which f ≫ 1 claims — "we have achieved AGI," "this ship is unsinkable," "this schema covers everything" — are slowly dissolved to their actual information content. Price discovery is the market's destructive pass.

And at the center of all of it, a girl in a beaded black mini dress who solved the problem without solving the problem — by committing to nothing except what she is, and thereby making herself immune to every perturbation the evening might produce.

That's the whole theory.