Theory, Framework, or Deepfake?

The Form Had Already Decided

Before the Department would release my field authorization, it required me to classify the object I had been sent to investigate.

This was premature. It was also mandatory.

The classification appeared on Form 14-C, Intellectual Object Entry and Containment. The Department uses the form whenever a civilization produces an idea too large for one discipline and too small for a religion. Its purpose is to prevent ambitious concepts from moving freely between administrative jurisdictions while claiming, at each border, to have been misunderstood by the previous one.

The form offered three choices:

Only one box could be selected. An instruction beneath the boxes explained that intellectual objects performing several functions should choose the function most likely to determine their future significance. This assumed that their future significance had already occurred.

The Department needed a category to route the investigation. The form had begun treating the route as the conclusion.

I selected Framework, because the Earth book called itself one.

The form asked whether the framework generated explanatory claims.

I selected Yes.

The form removed my answer and selected Theory.

It then asked whether the theory had received sufficient empirical support to warrant that designation.

I selected Not yet established.

The form removed Theory and opened a yellow warning panel.

On Earth, the word deepfake usually concerns synthetic media designed to imitate a person, event, or record. The Department uses the term differently and, in my judgment, with insufficient concern for interplanetary translation. A Departmental deepfake is not necessarily fraudulent, malicious, artificial, or even false. It is an intellectual object that displays the morphology of explanation—defined terms, propositions, diagrams, confident arrows, recurrent symmetries, and a conclusion of unusual reach—without yet demonstrating corresponding explanatory substance.

The category concerns resemblance without earned function. It does not concern deception.

This distinction was important because Andre Milchman, the Earth author of Triadic Evolution, had not concealed the status of the work. Near the front of the book, before many readers would reasonably have committed to the journey, he had included a section titled “A Note on the Status of the Framework.” It stated that the work was not a replacement for evolutionary biology, not a completed general theory of society, and not an empirically established law of civilizational development. It presented itself instead as four things at once: a conceptual architecture, a scientific manifesto, a synthesis, and a research program.

The Departmental form contained none of these combinations. It did contain a button marked Author Declines Classification, but the author had not declined. He had supplied four classifications where one had been requested by nobody.

I reopened the book.

My first hypothesis had been administratively convenient: Triadic Evolution was a framework with several ambitious claims attached. The serious alternative was less convenient: it was a theory whose author had selected a weaker label in anticipation of the evidence not being sufficient. The Departmental alternative was that the object possessed the surface anatomy of both and the scientific function of neither.

I could not resolve these possibilities by asking what the book called itself. Intellectual objects are not self-certifying. A bridge does not become load-bearing by including the phrase load-bearing bridge in its name, although on Earth this appears to improve procurement.

I needed to determine what the object was trying to do.

The Object’s Own Answer

I began with the status note because it was the closest thing the book provided to a declaration at customs.

The declaration was unusually careful. As a conceptual architecture, the work said, Triadic Evolution defined an object of inquiry and separated distinctions that were often collapsed. As a manifesto, it advanced a position: biological organisms, social organization, and capability-bearing extensions should be examined together. As a synthesis, it connected existing domains while claiming no novelty for the mere observation that genes, cultures, institutions, or technologies change. As a research program, it distinguished six types of statements and specified conditions under which parts of the architecture should be revised.

This was not how a deepfake usually introduced itself. A theory-shaped object with insufficient substance normally benefits from reducing the distance between assertion and achievement. It prefers sentences such as This new model reveals, followed by a verb too large to inspect. The status note did the opposite. It created distance between kinds of claim and assigned different burdens to each.

That counted in the object’s favor. It did not count as evidence that any claim was true.

An author can accurately label a weak claim. An accurate warning label improves the handling of a substance; it does not improve the substance. The book’s self-classification therefore established a procedural fact: it knew that definitions, hypotheses, empirical claims, interpretations, and values could not all be defended in the same way. Whether it maintained that discipline after the note remained unknown.

I examined the thirteen core propositions that followed.

Their visual presentation created immediate concern. There were thirteen of them, but they were organized around three strands, three realms, two architectural primitives, three orders, and several linked propositions about mediation, capability, responsibility, and civilizational direction. The recurrence of three was not evidence against the framework. Nature is permitted to contain patterns. Authors are also permitted to prefer them.

The problem was that symmetrical architectures are cognitively generous. They help readers remember. They help diagrams close. They produce a sense that the object has been fully surveyed because each visible region has received a label. This can be scientifically useful when the distinctions correspond to consequential differences. It can also produce completion without discovery.

I marked the architecture aesthetically cooperative and continued.

Rather than recite the architecture, I designed a hypothetical case for it.

An applicant requests credit. The financial institution’s automated system rejects the application. The applicant sees a software decision; the institution owns the process and selected the governing rules; an external vendor supplied the model; employees can alter some inputs but may be unable to reconstruct why this result followed. The machine performs a capability once exercised more visibly by people, yet it has no account to answer, no assets to surrender, and no authority to repair the institution that deployed it. If an investigator follows the visible action alone, responsibility may appear to end at the interface. If the investigator follows ownership alone, the vendor and the model disappear. If the investigator treats every participant as equally responsible, the word loses its discriminating force.

The framework’s substantive claim was that capability can move outward into technological extension without responsibility automatically moving with it. That claim made the case intelligible. It did not decide how responsibility should be distributed, nor did it establish that Triadic vocabulary could do more than ordinary legal, institutional, or sociotechnical analysis. It created a question worth carrying forward, not an answer entitled to credit.

The remaining propositions supplied the surrounding architecture: three interacting strands of adaptive change, named Evolution by Natural Selection, Evolution by Organization, and Evolution by Extension; the social, physical, and digital realms; two responsibility-bearing architectural primitives, the sociotechnical agent and the sociotechnical unit; three relational orders; and claims about organizational mediation, governance, correctability, and civilizational influence.

The propositions did not all have the same status. The book said so explicitly. One was labeled an empirical claim. Several were definitions or conceptual propositions. One central claim about organizational mediation was presented as a hypothesis as well as a conceptual proposition. Others included normative requirements or commitments.

The Departmental form had treated the book as one object. The book had treated itself as a container of unlike objects.

This produced my first correction.

The final line changed the investigation. I stopped asking whether Triadic Evolution was a theory and began asking which parts of it were asking to be treated as theoretical.

Six Kinds of Sentence

The book’s six statement types appeared simple until I attempted to apply them.

A definition fixes usage. A conceptual proposition follows from the architecture. An empirical claim asserts something observable. A hypothesis proposes an unestablished observable relation. An interpretation reconstructs a case. A normative commitment states what should be preserved, required, or avoided. I treated the source’s labels as declarations of intended status, not as rulings: the wording of each proposition and the work it asked evidence to perform would determine its burden.

I prepared six trays.

The Department recommends trays when a classification contains fewer than eight categories and the researcher has not yet earned a database. I placed a card bearing each core proposition beside the tray that appeared to match its stated status. Several cards required two trays. The Department does not recommend placing one card in two trays because this makes inventory dependent on perspective. I copied those cards.

This was not duplication of evidence. It was duplication of stationery.

The exercise exposed why claim status mattered. Consider the statement that every evolutionary unit has a social core and a technological periphery. If treated as a definition internal to the framework, the question is whether the terms are coherent, discriminating, and useful. If treated as an ontological claim about every responsibility-bearing actor, the burden becomes much heavier. One counterexample may force revision. If treated as an interpretive lens, it can be applied to cases without establishing that the world itself arrives pre-separated into cores and peripheries.

The words did not change. The conditions of failure did.

The same difficulty appeared with the proposition that Evolution by Organization is the mediating strand between natural selection and extension. Part of the claim was architectural: within the framework, organization carried decisions, rules, institutions, norms, and coordination through which biological and technological capacities were directed. Part was empirical: organizationally mediated adaptation should produce observable differences in at least some domains. Part was comparative: alternative accounts might explain the same changes through culture, niche construction, learning, institutional evolution, or ordinary sociotechnical analysis without introducing a distinct evolutionary strand.

A claim can perform several jobs. This does not exempt it from any of them.

I adopted a rule: when a sentence crossed categories, it would inherit the strongest relevant burden, not the weakest available refuge. A claim could not become immune to evidence by calling itself conceptual after making an observable assertion. It could not acquire moral force by describing a common pattern. It could not acquire explanatory power by arranging definitions in causal order.

The converse also mattered. A definition could not be rejected merely because it was not directly falsifiable. Definitions are tools for discrimination: they should help observers identify the same kind of thing, separate cases that matter, and fail visibly when the world repeatedly defeats their boundaries. They earn their place through clarity, consistency, reconstructability, and consequence—not by pretending to be miniature empirical laws.

This was close to the framework’s own position. The status note said that definitions could be revised when they contradicted or failed to discriminate; empirical claims must yield to evidence; hypotheses remained provisional; interpretations must compete with alternatives; normative commitments remained open to ethical, political, and practical criticism.

The book had therefore done something scientifically responsible before the investigation began: it had made several routes of criticism explicit.

It had also done something scientifically dangerous: it had built a sufficiently varied portfolio of claims that some portion might survive almost any result.

If the empirical proposition failed, the architecture might remain conceptually useful. If the architecture added little, the manifesto might still direct attention. If the direction was familiar, the synthesis might still improve comparison. If the synthesis proved redundant, the research program might still generate questions. If the questions were already being asked, the normative commitments might still be defensible.

This was not necessarily evasion. Complex intellectual works often perform several functions. A map may guide, describe, warn, standardize, and reveal the mapmaker’s priorities without becoming fraudulent because it does not also predict the weather. Plural function is legitimate when each function accepts its own standard of failure.

Burden migration was different. If failure under one function quietly became success under another, a work could become impossible to dismiss for exactly the same reason that it was impossible to confirm.

I added a seventh tray marked Survives by Reclassification.

The Department sent an automated objection.

No further guidance arrived.

The Word “Theory” Requests Evidence

On Earth, theory performs several social functions before it performs a scientific one. It can mean conjecture, explanation, prestige, suspicion, abstraction, or an idea offered during a meeting by someone who will not be responsible for implementation.

Scientific practice narrows the term, though philosophy of science supplies no single universally enforced definition. Appendix H of the Next Generation Science Standards describes scientific theories as explanations of observable phenomena grounded in a body of evidence developed through many tests; broader philosophy-of-science accounts likewise treat theories as explanatory structures whose claims must answer to evidence, models, mechanisms, and competing accounts. The point is not that one committee or philosopher has issued a universal definition. None has. The point is that theory normally asks more of an object than internal coherence and classificatory reach.

It asks the object to explain.

It asks the explanation to encounter evidence.

It asks the encounter to create risk.

By that standard, Triadic Evolution was not yet a scientific theory as a whole. The source itself did not claim otherwise. Its architecture had not been established through a body of direct tests as a unified explanation of humanity’s sociotechnical evolution. Several components drew on mature evidence: human populations remain subject to natural selection; social learning, institutions, technologies, infrastructures, and records accumulate and alter later action; organisms modify environments; cognition and agency can be distributed across social and technical arrangements. But evidence for components does not automatically validate the architecture that arranges them.

The framework made claims with theoretical potential. The mediating role of organization could become explanatory if it specified mechanisms and generated discriminating expectations. The interaction among three evolutionary strands could become more than classification if it explained changes that competing accounts handled less well or predicted differences that the alternatives did not. Claims about timescales, correctability, responsibility, and extension governance could become empirically consequential if operationalized.

Potential is not status.

I therefore rejected Theory as the whole-object classification at this stage. The decision was narrower than it first appeared. I had not concluded that the book contained no theory, that it could not support theory development, or that its central propositions were false. I had concluded that calling the entire architecture a scientific theory would transfer evidentiary credit from established neighboring findings to a synthesis that had not yet earned equivalent support.

The Department accepted the rejection and asked whether I wished to classify the object as Failed Theory.

I declined. A framework does not fail to be a theory merely by not being one.

This distinction seemed elementary. It delayed authorization by six hours.

The Word “Framework” Requests Use

The label framework initially appeared safer because it claimed less. This impression did not survive the literature.

Across Earth disciplines, framework was not one status but a crowded habitat. Frameworks could organize concepts, guide observation and comparison, structure diagnosis or intervention, link levels of analysis, or prepare the ground for causal explanation. Some contained models or theories; some became instruments. Others remained diagrams reproduced often enough to achieve the institutional function of evidence without its inconvenience.

In implementation science, Per Nilsen reviewed a useful distinction among theories, models, and frameworks: theories tend toward explanation, models toward deliberate simplification, and frameworks toward organizing descriptive categories and presumed relations. His larger analysis also showed why the border remained unstable. In practice these approaches overlap, so the label matters less than the object’s aims, assumptions, and uses.

Research on frameworks in environmental and social inquiry made the instability still clearer. There is no universal definition. Frameworks may supply vocabulary, organize diagnostic or prescriptive inquiry, guide observation, connect domains, or structure decisions. Their value therefore depends on purpose and operationalization. Attractive concepts without a route to indicators, cases, comparisons, or decisions may remain scientifically polite and practically unavailable.

This produced a less comfortable standard.

A theory requests evidentiary support for explanation. A framework requests demonstrated fitness for use.

Calling Triadic Evolution a framework did not release it from testing. It changed the tests.

Could independent analysts apply its categories with reasonable agreement? Did its distinctions separate cases that neighboring vocabularies blurred? Did it help reconstruct responsibility without inventing actors? Could its Ecosystemic, Conditioned, and Systemic Orders distinguish coexistence, regulation, and integration in ways that affected analysis? Did separating Evolution by Organization from Evolution by Extension reveal different mechanisms or intervention points? Could the framework generate hypotheses that were not merely restatements of its categories? Did it alter what evidence should be collected? Did it change a consequential diagnosis or decision?

These were not smaller questions than the theory question. Some were more demanding because a framework could remain conceptually defensible while failing every practical test.

I returned to the source’s self-description. As a conceptual architecture, it promised to define an object of inquiry and preserve distinctions. This was a legitimate function. If humans routinely used system for a platform, a company, a market, a software stack, and the failure produced when none of them agreed, a more discriminating architecture might help. If capability moved through technologies while responsibility remained with people and organized units, a framework that prevented observers from assigning moral agency to the most visible machine might improve diagnosis. If arrangements differed because actors merely coexisted, regulated one another, or integrated functions, the distinction could matter.

But each sentence contained might.

The source had provided categories and plausible reasons for them. It had not yet demonstrated incremental value over the full set of neighboring fields from which it drew. A framework does not deserve attention because every concept can be placed somewhere inside it. That is coverage. Usefulness begins when the placement changes understanding, inquiry, or action.

The Departmental form defined a framework as an organized set of concepts not yet promoted to theory.

I filed a correction.

The rejection clarified the practical problem. Earth’s intellectual labels were not a ladder with data at the bottom, framework in the middle, and theory at the top. They were different tools, often nested and often mislabeled. Prestige encouraged a ladder because ladders are easier to rank. Science required function.

My provisional classification returned to Framework, but now with a burden attached: the framework would have to do work that simpler vocabularies, established theories, and ordinary competent analysis did not already perform.

That burden would govern the rest of the investigation.

The Three Additional Passports

The source had not stopped at framework. It also claimed the status of synthesis, manifesto, and research program. Each term offered a different route through the Departmental border.

As a synthesis, the work joined established domains while denying novelty for the familiar components. Its claimed contribution was architectural: one responsibility-aware account preserving differences in mechanism, actor, timescale, and evidence. Connection is inexpensive. The joining would earn value only if it produced a new hypothesis, repaired a blind spot, sharpened comparison, improved responsibility allocation, or changed a decision defensibly. For the moment, synthesis described construction, not success.

As a manifesto, the book took a position: biological organisms, social organization, and capability-bearing extensions should be examined together when studying humanity’s adaptive development. A scientific manifesto can redirect attention before its program succeeds, but urgency is not evidence. The label identified what the author wanted inquiry to confront; it did not establish that this architecture was the best way to join the domains. By the standard proper to manifestos, however, it qualified: the position was specific enough to organize criticism. Whether civilization needed it was a separate question.

The final label, research program, shifted evaluation from present achievement to organized future work. Imre Lakatos used scientific research programmes to describe developing sequences of theories, adjustments, predictions, and discoveries rather than isolated hypotheses falling instantly when one observation objects. A programme could be progressive when changes extended explanatory or empirical reach, or degenerative when adjustments mainly protected the core after difficulty.

Triadic Evolution did not claim to instantiate Lakatos’s methodology in a strict sense. It used research program more broadly: a structured agenda containing definitions, propositions, hypotheses, tests, revision conditions, and unresolved questions.

The claim had merit. The source identified counterexamples that would damage central parts of the architecture: a responsibility-bearing actor outside its two primitive types; a relation-forming mechanism outside its three orders; failure of claimed timescale differences; failure of correctability to improve learning or repair; artificial actorhood that met the stated criteria; persistent inability to distinguish cases, locate responsibility, or improve diagnosis.

These were genuine invitations to trouble.

They were not all equally operational. “Persistent inability to improve diagnosis” requires a comparison, a task, analysts, criteria, and some account of what improvement means. “A relation-forming mechanism outside three orders” requires the categories to be defined tightly enough that a fourth mechanism cannot simply be redescribed as a mixture. “Correctability improves repair” requires a baseline and observable measures. A research program is not made empirical merely by listing sentences in the future tense.

Still, the source had provided more than a protective declaration that further research was needed. It had specified where research should press. The four labels described architecture, construction, position, and proposed work; none established that the work was necessary, novel enough, empirically productive, or practically useful. Together they explained why the object kept refusing a single box: the labels referred to different functions.

This counted against a provisional deepfake classification. It did not close the category.

A Deepfake Without Deception

The Department’s deepfake category existed because weak intellectual objects often survive category disputes better than empirical ones.

If accused of lacking evidence, they become frameworks. If accused of adding no new framework, they become syntheses. If the synthesis is familiar, they become manifestos. If the manifesto is obvious, they become research programs. If the program produces no results, they explain that the field is young. At no point must the object admit that its principal achievement was visual organization.

This sequence was possible for Triadic Evolution. The source’s plural self-description made it possible.

The same sequence was also possible for any honest interdisciplinary work at an early stage. Premature breadth and legitimate research architecture can look similar before testing. Both contain vocabulary. Both identify gaps. Both connect literatures. Both propose future studies. Both explain why existing disciplinary boundaries are insufficient. One is constructing a bridge. The other is drawing a bridge in enough detail that travelers begin discussing tolls.

The difference is not appearance. It is load.

I assembled a provisional screen. It was an instrument for generating questions, not a validated measure.

Against these risks stood four contrary observations.

First, the source did not hide its dependencies. It acknowledged the neighboring literatures it drew upon and denied novelty at the level of familiar components. Whether the list was complete I had not yet verified.

Second, it separated statement types rather than allowing conceptual coherence to impersonate empirical support.

Third, it stated several revision conditions that threatened central categories, not merely peripheral applications.

Fourth, it made practical value testable in principle. Persistent failure to distinguish cases, locate responsibility, or improve diagnosis would count against the synthesis.

A convincing deepfake normally resists explicit comparison with the reality it imitates. This object had left doors open.

Open doors are not the same as visitors.

Only after writing the fourth contrary observation did I notice the arrangement. I had opposed four risks with exactly four answers. Earlier I had issued three additional passports after objecting to the framework’s repeated triads. I had prepared six trays and permitted a seventh only after the object refused the inventory. My method was reproducing the symmetry credit it was supposed to detect.

The observations had not become false. They had become falsely commensurable. A stated revision condition did not cancel borrowed support; acknowledged prior art did not neutralize retrospective fit. I had made the object administratively balanced and mistaken balance for assessment.

The Department asked for a probability that the object was a deepfake.

I refused. The screen was uncalibrated, its factors had no defensible weights, and its evidence was not commensurable. Worse, categories I had constructed to evaluate the object could not independently validate the evaluation.

The Department asked whether my refusal reflected insufficient confidence, insufficient data, or philosophical objection to quantified belief.

I selected Insufficient object stability.

The form interpreted this as a concern that the book might physically change during examination and requested containment equipment.

I selected Other.

The form closed.

This was the most empirically successful part of the classification process.

The Classification Experiment

The failure report removed the conclusion my balanced list appeared to support. I could not count favorable and unfavorable features; I needed to make the framework discriminate. I designed a simple test.

The test did not recruit participants or generate data. It was a proposed classification exercise intended to expose what evidence would be needed. Independent analysts would receive a set of claims from Triadic Evolution without the book’s self-assigned labels. For each claim, they would identify its primary status—definition, conceptual proposition, empirical claim, hypothesis, interpretation, or normative commitment—and state what observation or argument could support, weaken, or revise it. A second group would receive the labels supplied by the source. A third would receive comparable claims drawn from neighboring frameworks and theories.

The test would ask three questions.

Could analysts distinguish the claim types reliably?

Did the source’s labels improve agreement about appropriate evidence?

Did the Triadic architecture produce discriminators more precise than those available in neighboring accounts?

The first question tested clarity. The second tested methodological value. The third tested incremental contribution. Agreement would show classificatory clarity, not that any classified claim was true.

The design also contained a failure condition. If analysts repeatedly assigned different statuses because the propositions shifted between definition, observation, interpretation, and norm without a stable boundary, the source’s claim-type discipline would be weaker than it appeared. If the supplied labels improved agreement only because they instructed analysts what to say, that would show documentary clarity, not scientific value. If neighboring frameworks generated the same distinctions with less vocabulary, conceptual cost would count against Triadic Evolution.

This proposed study could not decide whether the entire framework was true. It could determine whether one of its first advertised virtues—keeping unlike claims from borrowing one another’s authority—survived contact with independent readers.

I submitted the design.

The Department returned it with one question.

The demand was circular but not foolish. Research administration often requires a stable object so that methods, expertise, risks, and budgets can be assigned. A theory may require one form of review; a practical framework another; a manifesto perhaps a smaller room. Categories coordinate work. The Department’s error was not asking for one. It was requiring the category to contain the conclusion.

I revised the submission:

The form accepted the first classification and deleted the second because the field had exceeded its character limit.

I received authorization.

What the Label Could Not Tell Me

The classification exercise had changed my initial hypothesis, but not in the direction the Department preferred.

I had begun with the convenient view that Triadic Evolution was simply a framework. I ended with a more precise and less reassuring conclusion. It was a framework by declared architecture and present evidentiary status. It contained claims capable of theoretical development, an empirical premise supported independently of the whole, hypotheses requiring operationalization, interpretations requiring rivals, and normative commitments requiring justification. It also functioned as a synthesis, manifesto, and proposed research program.

This plurality was not itself a defect. Serious intellectual projects can organize concepts, explain selected relations, propose tests, and direct inquiry at once because scientific work itself overlaps. Nor was plurality a merit: a book does not become more valuable by performing four under-tested functions instead of one.

The deepfake category remained open, but only as an epistemic risk. Nothing in the source’s status statement suggested deception. The relevant concern was whether the architecture’s visible comprehensiveness exceeded its discriminating, explanatory, research-generating, or practical contribution. That could not be determined from typography, candor, or the number of revision conditions.

The classification had nevertheless produced one useful result. It prevented the investigation from asking one crude question—Is the framework true?—as if definitions, hypotheses, interpretations, and norms could receive one answer. From this point forward, each claim would face the criticism appropriate to the work it was doing. That was an improvement in method.

It was not yet an improvement attributable uniquely to Triadic Evolution. Scientists and philosophers had distinguished theories, hypotheses, models, frameworks, values, and evidence long before this book. The source had applied that discipline to its own architecture. Whether doing so added anything beyond established practice remained unresolved.

This was now the decisive problem. If the components, distinctions, and methods already existed across neighboring fields, the framework might be well classified and unnecessary. An accurate map of roads already mapped does not become useful because its legend is honest.

I closed Form 14-C and opened a new file.

Its title was Prior Art.

The file expanded immediately.