Generativity Exploitation - A Baseline-Relative Capability Framework

Abstract

Knowledge, language, practical judgment, and creative organization can be used
without being depleted, while the conditions for developing and exercising them
remain scarce, controlled, or revocable. A participant can therefore retain a
method, a language, or a latent organization in full while losing effective
opportunity to initiate, continue, redirect, or expand it. This asymmetry
between retention and foreclosure is invisible to appropriation models built on
depletion, and it is the phenomenon this paper addresses. The paper proceeds in
four registers. It first analyzes why non-rivalry in generative forms coexists
with rival, congestible, excludable, and positional scarcity in their
realization conditions. It then locates the proposal among production-centered,
transactional, degradation-centered, domination-centered, and structural
accounts of exploitation, treating them as rival constraints rather than as a
consensus. It develops the capability object philosophically through real
opportunity, relationally constituted autonomy, epistemic standing, and the
externalization of knowledge into technical systems. It describes the
political-economic mechanisms through which control of realization conditions
operates, including rentiership, platform bottlenecks, and contributed input
without corresponding return, and it separates the distinct actor positions
that an artificial-intelligence configuration collapses. On this basis the
paper proposes a limited diagnostic for relations in which one party’s
advantage is produced or sustained through a material reconfiguration of
another party’s effective generative capability. The primary capability object
is a reachable viable repertoire over a declared domain, history, horizon,
resolution, action set, and arrangement. An enriched profile adds autonomy,
direction, attribution, continuity, or exit only when those dimensions are
independently defined. The diagnostic then requires an independently admissible
comparison arrangement, participant standing, an actor-linked
capability-changing relation, a material capability disadvantage, and a typed
benefit-through witness. Causal and constitutive witnesses are separated, and
the complete result is a three-valued profile that retains admitted, excluded,
and unresolved baselines rather than an unqualified binary verdict. One
conditional proposition proves repertoire inclusion under policy expansion,
preserved transition probabilities, and expanded viability. Countermodels show
that latent generativity can remain fixed while capability contracts, that
equal output can coexist with reduced autonomy, that coincident loss and gain
can lack a benefit-through path, and that classification can vary across
admissible baselines. The framework establishes no prevalence, wrongfulness,
liability, remedy, or universal metric, and it makes no empirical claim about
any named firm, platform, or industry. It leaves propagation, cycles,
recovery, accumulation, legal translation, and identification from finite
traces to companion work. Its contribution is a revisable dependency structure
that exposes where a proposed diagnosis is supported, relatively rejected,
baseline-sensitive, or unresolved.

Keywords: generativity exploitation; capability; feasible

Document Status, Preparation, and Reuse

This unnumbered section records the manuscript’s working status, preparation
process, and reuse conditions. These statements govern the document and carry
no evidential weight for its substantive position.

Status and correspondence.
This paper is a revisable discussion document. Its definitions are proposals,
its mathematical results are conditional on declared assumptions, and its
empirical and normative bridges remain open. Objections, counterexamples,
corrections, alternative formulations, and relevant literature are welcome at
mailto:huangwanhong@serendip.ngohuangwanhong@serendip.ngo.

Generative AI use.
Generative AI systems were used in the preparation of this work. Anthropic’s
Claude and OpenAI systems including ChatGPT and Codex supported exploratory
discussion, formal reconstruction, source discovery followed by verification,
and drafting in . The author determined the research questions,
theoretical commitments, and epistemic status of the claims and bears sole
responsibility for the manuscript, including its errors. None of these systems
is an author or holds authorship credit.

License.
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0
International License (CC BY-NC 4.0). The license text is available at
https://creativecommons.org/licenses/by-nc/4.0/.

1. Introduction

Knowledge, language, practical judgment, and creative organization can be used
without being consumed. One person’s use of a method leaves it available to
another, and one person’s learning of a language removes nothing from anyone
else. The conditions under which such forms are developed and exercised behave
differently. Time, funding, collaborators, recognition, institutional
permission, tools, infrastructure, priority, and channels of distribution are
scarce in several distinct ways, and they are frequently controlled by parties
other than those whose capability depends on them.

This asymmetry produces a situation that ordinary models of appropriation do not
describe. A participant can retain a method, a language, or a latent
organization in full while losing effective opportunity to initiate, continue,
redirect, or expand it. Nothing was taken in the sense that anything is missing.
What contracted is the set of futures reachable from that participant’s actual
position. A better-resourced party may occupy a recognized trajectory first; a
controller may open infrastructure, induce reliance, and then close it; a
capability assembled from broadly distributed contribution may be realized at
scale under narrow control. In each case the affected participant possesses
everything it possessed before, and can do less with it.

This paper asks when such a situation warrants a distinctive diagnosis. The
question has a boundary problem at its center. Competition lowers a rival’s
prospects; harm damages capacity; domination subjects a party to another’s
arbitrary power; displacement moves activity elsewhere; ordinary reciprocal
dependence constrains everyone. If generativity exploitation names something,
it must be separable from all of these, and the separation must be drawn in a
way that can return a negative verdict rather than absorbing every unfavorable
relation.

The paper’s response is a diagnostic with five conditions and a three-valued
report. A relation becomes a candidate instance when a participant with relevant
standing suffers a material disadvantage in effective generative capability
through an actor-linked relation, and a beneficiary’s specified advantage is
produced or sustained through that same affected position. Each condition is
evaluated against an independently admissible comparison arrangement, and the
complete result retains every admitted, excluded, and unresolved comparison
rather than collapsing into a verdict.

Four insights organize the argument and are stated here so that a reader can
locate them. The first concerns register: the loss in these situations is
representable only at the level of real opportunity, since possession is
unchanged and output may be unchanged or temporarily raised. The second concerns
what makes generative capability distinctive among capabilities. Its exercise
alters capability sets, so trajectories branch and a contraction forecloses not
only the excluded trajectories but every continuation they would have made
reachable; this is why the formal object is a set of paths over a horizon rather
than a quantity, and why later restoration does not restore the position. The
third concerns standing, and resolves a paradox that would otherwise defeat the
proposal. If a capability was assembled from the contribution of nearly
everyone, a contribution-based standing rule makes nearly everyone a
participant, and the category loses its contrast class. The paradox equivocates:
causal contribution is a historical fact admitting no threshold, while standing
is a normative status tracking who stands to lose. Standing is therefore
constituted by sustained participation together with reliance, which is
compatible with universal contribution and differentiates among contributors.
The fourth concerns what follows from a positive finding. When a beneficiary’s
advantage runs through an affected participant’s diminished position, the
beneficiary holds a standing interest in that diminution regardless of intent,
and that structural interest calls for justification. The finding therefore
shifts a burden without settling wrongfulness all things considered.

The paper’s contribution is a dependency structure rather than a verdict
procedure. It supplies a vocabulary in which a proposed diagnosis can be shown
to be supported, relatively rejected, baseline-sensitive, or unresolved, and it
makes explicit which evidential and normative obligation each status rests on.
Its most useful output in practice may often be an explicit unresolved profile,
which records where standing, comparison, measurement, or causal evidence is
missing and thereby converts a broad accusation into a specific research
demand. The paper establishes no prevalence, no wrongfulness verdict, no
liability, no remedy, and no universal metric, and it classifies no observed
relation.

The argument proceeds in nine further sections. Section 4
analyzes why depletion-based models fail and derives the requirements a
replacement must satisfy. Section 5 reviews exploitation
traditions as rival constraints. Section 6 argues the
theses just summarized. Section 7 describes the
political-economic mechanisms through which control of realization conditions
operates. Sections 8 and 9 construct the
capability object and the diagnostic. Section 10 tests both against
countermodels. Section 11 states the empirical protocol,
Section 12 collects open questions by register, and
Section 13 states limitations and defeat conditions.

The philosophical and formal parts stand in two distinct relations, and
conflating them causes avoidable confusion. They are formally
independent: every proposition in Sections 8
through 10 follows from declared model assumptions, and no
derivation cites a thesis. A reader may verify each proof while rejecting every
philosophical argument. They are justificatorily dependent: the formal
construction is a schema, and its claim to be worth constructing rests on the
theses. Section 6.9 records what fails if each is rejected.

2. Research Scope and Diagnostic Position

This section defines the paper’s target, limited thesis, and order of inquiry.
It begins from three motivating configurations, identifies the relational
object under study, and separates the proposed diagnostic from empirical,
moral, and legal conclusions.

2.1 Three Motivating Configurations

This subsection presents the situations that generate the research question.
Each is stated schematically. None is classified by this paper, and none is
offered as an empirical description of a particular person, firm, or industry.

Priority and funding asymmetry.
One participant originates a line of work and continues to possess it in full.
A better-resourced participant learns it, obtains the time, funding,
collaborators, and institutional permission required to realize it, publishes
or deploys it first, and occupies the recognized trajectory. The originator
loses nothing that can be inventoried. What contracts is the set of futures in
which the originator can still develop that work: the priority is taken, the
funding is allocated, the collaborators are committed, and the recognized
position is filled.

Open access followed by enclosure.
A controller develops a method and the infrastructure required to run it, then
opens both. Others learn the method, specialize their practices around it,
invest in complementary skills, and build trajectories that depend on continued
access. The controller subsequently closes access, restricts it, or reprices
it. The method remains known to everyone who learned it. The capability to
continue along the trajectories built around it contracts sharply, and it
contracts most for those who specialized most.

Distributed contribution with concentrated control.
A capability is produced historically through broadly distributed linguistic,
practical, and relational contribution. The infrastructure that realizes that
capability at scale is controlled narrowly. If nearly everyone contributed to
the field from which the capability was assembled, then either nearly everyone
stands in an exploited relation to its controllers, or the category of
exploitation dissolves for lack of a contrast class. Both conclusions are
unsatisfactory, and the tension is a reason to examine the concept rather than
to apply it.

2.2 Working Premise and Conceptual Target

This subsection states the premise the paper adopts and identifies the relation
it proposes to diagnose.

This paper adopts the working premise that language, knowledge, concepts,
research practices, and many creative forms can be non-rival in use: one
person’s use need not remove another person’s possession. The premise is
domain-sensitive and receives no general proof here. Their realization
nevertheless depends on time, education, collaborators, recognition,
institutional permission, tools, finance, infrastructure, channels, and
practical control. A person can therefore retain a language, method, or latent
organization while losing effective opportunities to initiate, continue,
redirect, or expand it. Resource depletion supplies an inadequate model for
this relation.

The present proposal uses generativity exploitation to name a possible
political-economic relation centered on this difference. Its conceptual
development revised several early formulations. Origination, copying, common
genesis, control of conditions, interruption, and intention were each rejected
as a complete essence. Capability loss alone was judged insufficient, as was the
coincidence of one party’s loss with another party’s gain. The surviving idea
places a benefit path through the affected party’s capability position at the
center of the diagnosis.

Generativity in a declared domain is a historically and relationally
conditioned organization capable, under suitable realization conditions, of
supporting a repertoire of transformations or trajectories. It is a model
object rather than an observed output, an automatically valuable property, or
a claim of authorship.

A relation is a candidate instance of generativity exploitation, relative to a
declared arrangement and comparison, only when an actor with relevant standing
suffers a material disadvantage in effective generative capability through an
actor-linked relation and a beneficiary’s specified advantage is produced or
sustained through that same affected position.

Claim ? is a proposed conjunction, not a theorem about the
ordinary or established meaning of exploitation. Its ingredients carry
conceptual, empirical, and normative burdens. In particular, standing,
comparison admissibility, materiality, and benefit selection prevent the
diagnostic from being value-free. Satisfaction also supplies no complete theory
of wrongfulness, domination, responsibility, liability, or remedy.

2.3 Source Use and Attribution

This subsection states the constraint governing every citation below.

Where a source is used, the text states the specific constraint it supplies and
the inference it withholds. The verified corpus establishes antecedents,
constraints, and disagreements; it establishes neither the novelty nor the
adequacy of the construction proposed here, and no source is cited in support of
a proposition or a countermodel.

3. Non-Depletion Conditions and the Structure of the Problem

This section performs the problem analysis that the diagnostic presupposes. Its
objective is to establish why depletion-based models of appropriation fail for
generative forms, to characterize the scarcity that remains, and to derive the
requirements a replacement diagnostic must satisfy.

3.1 Non-Rivalry in the Economic Tradition

This subsection identifies the verified economic antecedent for the working
premise and states the limits of that inheritance.

Romer’s account of technological change treats ideas as inputs that are neither
conventional private goods nor pure public goods: they are nonrival and only
partially excludable, and this combination is what permits increasing returns
and sustained growth (Romer, 1990). Nonrivalry in this sense is a
technical property of the input, and it entails that use by one agent leaves the
input available for use by another. The premise adopted in Section 2 is the
same property applied to a wider class of generative forms.

The knowledge-commons literature extends the analysis from an input in a growth
model to a governed social resource. Hess and Ostrom assemble analyses of
knowledge as a commons and of the institutional arrangements through which
communities define, protect, and build it (Hess & Ostrom, 2007). This work
supplies the vocabulary of provision, appropriation, and governance for
resources whose units are not subtracted by use.

Two limits apply. First, Ostrom’s original framework was developed for
common-pool resources whose units are subtractable and whose governance problem
is congestion or depletion (Ostrom, 1990). Its transfer to
knowledge is a contested extension rather than a settled result, and the
present paper relies on the extension only for vocabulary. Second, nonrivalry
is domain-sensitive. A method may be nonrival while the trained judgment
required to apply it is embodied, slow to acquire, and unavailable to
duplication. The premise is therefore stated per domain and never used to
license a general claim about all generative forms.

3.2 Mixed Scarcity in Realization Conditions

This subsection identifies the scarcity that persists once depletion of the
generative form is excluded. Its objective is to require that every foreclosure
claim name the specific scarcity carrying the causal effect.

A non-rival generative form is realized only through conditions that are
themselves scarce in several distinct ways:

• Rival conditions. Compute, capital, energy, laboratory access, and the researcher’s own time are consumed by use. Allocation to one trajectory withdraws them from another.
• Congestible conditions. Review capacity, mentorship, shared infrastructure, and channels of distribution degrade in quality as load increases without being exhausted outright.
• Excludable conditions. Institutional permission, credentials, data access, legal license, and organizational membership can be granted or withheld by an identifiable controller.
• Positional conditions. Priority, attention, recognition, reputation, and market position are constituted by relative standing. They cannot be expanded for everyone simultaneously, because the good consists in the ranking.

These four kinds carry different causal signatures and different remedies.
Congestion can be relieved by capacity expansion; positional scarcity cannot,
because expanding the supply of first place is incoherent. Exclusion identifies
a controller with a decision; rivalry may identify no one. A claim that a
participant’s capability was foreclosed is therefore incomplete until it names
which condition class carried the effect, and the diagnostic in Section 8
requires that specification through the arrangement components.

3.3 Retention with Foreclosure

This subsection states the central asymmetry the paper addresses and
distinguishes the objects it separates.

The configurations in Section 2 share a structure. In each, the affected
participant retains the generative form and loses effective opportunity to
realize it. Nothing was subtracted from what the participant knows or can do in
the abstract. What changed is the set of futures reachable from the
participant’s actual position under the actual arrangement.

Four objects must therefore remain distinct:

• fixed traces and artifacts, which can be copied without loss;
• latent generative organization, which is retained through copying;
• the participant’s socially situated position and trajectory, which copying can nonetheless alter;
• realization infrastructure and access conditions, which are controlled separately from all three.

Conflating these produces characteristic errors. Treating the artifact as the
object of appropriation yields the conclusion that non-rival copying is
harmless. Treating the latent organization as the object yields the conclusion
that retention settles the question. Only the third and fourth objects register
the configurations of Section 2, and only a capability construction indexed to
position and arrangement can represent them.

3.4 Requirements on a Replacement Diagnostic

This subsection converts the preceding analysis into the obligations discharged
by the formal construction.

The analysis yields four requirements. First, the diagnostic needs a capability
object distinct from possession and from observed output, indexed to a
participant’s position under a specified arrangement. Second, it needs a
comparison, because a capability level carries no diagnostic content in
isolation; and the comparison cannot be selected by the conclusion it is meant
to support. Third, it needs an actor link, because impersonal scarcity and
natural events can contract capability without implicating any party. Fourth,
it needs a path condition connecting the affected capability position to a
beneficiary’s advantage, because loss and gain coincide constantly under
ordinary competition.

Sections 8 and 8 discharge these requirements in order. Sections 5 through 6
first establish what existing traditions supply toward them.

4. Exploitation Traditions and Rival Diagnostics

This section locates the proposal among established accounts of exploitation.
Its objective is bounded positioning: each tradition supplies a constraint or a
contrast, the traditions disagree with one another, and the complete diagnostic
remains a repository-specific formulation. The section establishes antecedents
and disagreement; it establishes no novelty claim for the present proposal.

4.1 Production, Surplus, and Property Relations

This subsection presents the production-centered tradition and states what the
present proposal takes from it.

Marx’s analysis of labor-power, necessary labor, surplus labor, and surplus
value is the central production-oriented ancestor of political-economic
exploitation (Marx, 1992). Its structure is an extraction relation:
the worker sells labor-power, receives its value, produces more than that
value, and the difference is appropriated by the owner of the means of
production. The relation is located in production and is measured in a common
substance.

Roemer reconstructs exploitation without the labor theory of value. On his
game-theoretic account, a coalition is exploited when it would be better off
withdrawing with its per-capita share of productive assets, so that exploitation
tracks the distribution of property in productive assets rather than an unequal
exchange of labor (Roemer, 1982). Cohen argues separately that
Marx’s charge of exploitation does not require the labor theory of value at
all, and that the value-theoretic apparatus is detachable from the normative
claim (Cohen, 1979).

The present proposal takes three things from this tradition and declines a
fourth. It takes the idea that an exploitative relation is structural rather
than attitudinal; the idea that control over the conditions of production
matters independently of any transaction’s surface terms; and Roemer’s
demonstration that a withdrawal-based comparison can define exploitation
without a value substance. It declines the identification of generativity with
labor-power and the reduction of capability contraction to surplus-value
extraction. Generative capability has no established common substance in which
a surplus could be measured, and the paper introduces none.

4.2 Transaction, Fairness, and Mutual Benefit

This subsection presents the transactional tradition and the constraint it
imposes on comparison construction.

Wertheimer analyzes exploitation as taking unfair advantage in transactions and
relationships, and demonstrates that mutually beneficial transactions can still
be exploitative (Wertheimer, 1996). The demonstration matters
here for a specific reason: it blocks the inference from joint gain to
non-exploitation. A participant who is better off than before may still stand
in an exploitative relation, because the relevant comparison is not the
participant’s prior state.

This tradition therefore forces the comparison question into the open without
answering it. Wertheimer’s analysis supplies no uncontested benchmark, and the
choice among a prior state, a market outcome, an idealized arrangement, and a
feasible alternative is normatively loaded. The present proposal responds by
refusing to select a single benchmark: Section 8 constructs a family of
technically feasible arrangements, applies an independently justified
admissibility filter, and reports the resulting classification across every
admitted comparison. The two-stage construction is a repository-specific
response to Wertheimer’s problem and is not entailed by his account.

4.3 Degradation, Vulnerability, and Domination

This subsection presents the accounts that locate exploitation in a substantive
wrong and states why the present proposal separates that question.

Sample develops an ethically substantive account centered on degrading
interaction, vulnerability, and failures with respect to well-being
(Sample, 2003). On this view, exploitation is constitutively
wrongful, and market-price or gain comparisons do not exhaust the conceptual
field.

Vrousalis develops a relational account in which exploitation is the
self-enriching instrumentalization of another’s vulnerability, and in which
exploitation is a species of domination (Vrousalis, 2013). This
is the closest verified antecedent to the present proposal’s structure, because
it makes the beneficiary’s enrichment run through the affected party’s
constrained position. The resemblance is genuine and must be stated as such.

Pettit’s republican conception of freedom as non-domination supplies the
neighboring structure against which this can be contrasted (Pettit, 1997).
Domination on that account consists in subjection to another’s capacity for
arbitrary interference, and it obtains whether or not interference occurs and
whether or not the dominating party gains. The present proposal separates these:
an actor may hold arbitrary power over another’s realization conditions without
any advantage running through the affected position, and an advantage may run
through an affected position without arbitrary power. Domination and
generativity exploitation are therefore distinct predicates that frequently
co-occur.

The present proposal declines to make its diagnostic constitutively wrongful.
It records a structural relation and treats wrongfulness, consent, procedural
legitimacy, responsibility, and remedy as separate variables. This decision
requires an argument, since a classification with no normative consequence would
be idle. Section 5.8 supplies that argument and defends the intermediate
position that a positive diagnosis shifts a burden of justification onto the
beneficiary without settling whether the relation is wrongful all things
considered.

4.4 Structural Scope and Its Limits

This subsection presents the adversarial constraint on structural readings of
the proposal.

Zwolinski examines whether background injustice should determine the
exploitation status of a discrete transaction and argues for a more limited
role than strong structural accounts assign it (Zwolinski, 2012).
The objection applies directly here. A framework organized around control of
realization conditions invites the inference that any transaction occurring
within an unjust distribution of those conditions is thereby exploitative. That
inference would make the diagnostic vacuous, since every transaction occurs
within some background distribution.

The present proposal accepts the constraint. Structural position is neither
necessary nor sufficient for a positive row status. The diagnostic requires an
identified actor link and an identified benefit path in the specific relation
under assessment. Background injustice may enter through the admissibility
filter, where it belongs and where it must be defended explicitly, and it may
not enter as a substitute for the mechanism and witness clauses.

4.5 Comparative Summary of Traditions

This subsection consolidates the review. Table 1 records what
each tradition treats as the core relation, how it handles mutual benefit, what
normative status it assigns, and the specific constraint it imposes on the
present proposal.

| >p0.16

p0.21
p0.15
X

Tradition	Core relation	Normative status	Constraint imposed on the present proposal
Surplus extraction (Marx, 1992)	Appropriation of surplus labor through ownership of productive means	Constitutively wrongful	Supply a structural relation without importing a value substance in which surplus is measured
Property relations (Roemer, 1982)	Disadvantage relative to withdrawal with a per-capita asset share	Wrongful under stated distributive premises	Define comparison through a specified feasible alternative rather than an unequal exchange
Value-theory independence (Cohen, 1979)	Appropriation stated without the labor theory of value	Wrongful	Keep the normative claim separable from the measurement apparatus
Unfair advantage (Wertheimer, 1996)	Taking unfair advantage in a transaction or relationship	Wrongful, with mutually beneficial cases	Block inference from joint gain to non-exploitation and declare the comparison benchmark
Degradation (Sample, 2003)	Degrading interaction with a vulnerable party	Constitutively wrongful	State whether the proposed diagnostic is descriptive or itself carries wrongfulness
Domination and enrichment (Vrousalis, 2013)	Self-enriching instrumentalization of vulnerability	Constitutively wrongful	Acknowledge the closest antecedent and state the difference in benefit-path requirements
Non-domination (Pettit, 1997)	Subjection to a capacity for arbitrary interference	Freedom-based	Separate arbitrary power over conditions from advantage running through the affected position
Structural limitation (Zwolinski, 2012)	Background injustice, with limited transfer to transactions	Contested	Prevent structural position from substituting for the mechanism and witness clauses

Table. Exploitation Traditions and the Constraints They Impose

These accounts disagree about fairness, degradation, domination, vulnerability,
background structure, and the moralization of exploitation. The present paper
uses them as rival constraints rather than as a consensus, and it does not
claim that its diagnostic captures every accepted use of the term.

5. Philosophical Foundations of the Capability Object

This section argues for the four theses on which the framework depends. Its
objective is substantive rather than expository: the formal construction of
Sections 8 and 8 is a schema, and a schema acquires point only from an argument
that its distinctions track something worth distinguishing. Each subsection
states a thesis, argues for it, states the strongest objection known to the
author, and replies. Section 5.9 records exactly what fails if each thesis is
rejected.

The four theses answer four questions the formalism cannot answer for itself.
Why is capability the right object, given that possession and output are easier
to measure? Why does generative capability warrant separate treatment
among capabilities? Who has standing to be a subject of the diagnosis, given
that contribution to a shared practice is nearly universal? And what follows
normatively from a positive diagnosis?

5.1 Real Opportunity Against Achieved Output

This subsection identifies the source of the paper’s opportunity-set
orientation and the limits of that inheritance.

Sen distinguishes achieved functionings from the capability set: the
alternative combinations of functionings a person can feasibly achieve and
choose among (Sen, 1993). The distinction matters here because
the configurations in Section 2 are invisible at the level of achieved output.
The originator in the priority configuration may continue to produce work of
the same volume and quality. What contracted is the set of alternatives
available for choice.

Robeyns systematizes the distinction among means, conversion factors,
capabilities, and achieved functionings, and emphasizes the choices required to
operationalize any capability analysis: which dimensions to include, how to
weight them, and whether to permit incompleteness in the resulting ordering
(Robeyns, 2005). These choices are substantive and cannot be
delegated to a formalism. Section 7 accordingly leaves the dimension
specification, the partial order, and the materiality threshold as declared
case-level inputs.

Nussbaum develops the approach into a substantive account on which a threshold
level of central capabilities is what a decent political order owes each person,
and on which capability rather than achieved functioning is the appropriate
political goal precisely because it leaves the exercise to the person
(Nussbaum, 2000). The structural point matters here independently of
her particular list: an arrangement that forecloses an option wrongs a person
even when the person would have declined the option, because what is removed is
the standing to decide.

This yields the section’s first thesis.

The configurations of Section 2 involve a genuine loss, and that loss is
representable only at the level of real opportunity. Neither possession of a
generative organization nor achieved output registers it.

The argument is a process of elimination over the four objects distinguished in
Section 3.3. Possession is unchanged in all three configurations, so an account
locating the loss in possession must deny that any loss occurred. Achieved
output may be unchanged, and in the enclosure configuration it may rise for a
period as participants work harder to compensate; so an output-based account
either denies the loss or misdates it. The artifact is copied without
subtraction. What changed in every case is the set of continuations available
from the participant’s position. Opportunity is therefore the only one of the
four registers in which the phenomenon appears at all.

The strongest objection is that opportunity sets are cheap: any arrangement can
be described as removing some opportunity, so an opportunity-based account
proves too much and classifies ordinary competition as loss. The reply is that
Thesis ? establishes only where the loss is
representable, and carries no implication that every representable
contraction matters. Materiality thresholds, an actor link, and a benefit path
do the discriminating work, and Section 9.1 proves that coincident contraction
and gain can fail the benefit condition. The thesis secures the choice of
register and nothing more.

5.2 The Second-Order Character of Generative Capability

This subsection argues that generative capability warrants separate treatment
among capabilities, and thereby answers the question the capability literature
leaves open for this domain.

Capability theory supplies no reason to single out generative capability. Sen’s
framework is indexed to functionings a person has reason to value, and
Nussbaum’s to a list of central capabilities; neither contains a category
matching the object of this paper. A framework built on generative capability
therefore owes an argument that the object is distinctive.

Generative capability is the capability whose exercise alters capability sets.
Its contraction is therefore compounding in a way that contraction of a
first-order capability is not: what is foreclosed includes the conditions under
which further capability would have been acquired.

The argument runs as follows. A first-order capability is a set of functionings
a person can achieve. Exercising it produces the functioning and leaves the set
otherwise unchanged. Generative capability, as Definition ?
specifies it, is an organization that supports a repertoire of transformations.
Exercising it produces new transformations, and those transformations become
part of the position from which the next exercise occurs. Trajectories therefore
branch: what is reachable at time $t+1$ depends on what was realized at $t$.
Contracting the repertoire at $t$ removes not only the trajectories excluded at
$t$ but every continuation that those trajectories would have made reachable
later. This is why Section 7 represents capability as a set of paths over a
horizon rather than as a set of end states.

Three consequences follow, and each does work later. First, the harm of
foreclosure is not fully assessable at the time of foreclosure, which is why
Section 10 treats horizon declaration as a substantive rather than technical
choice. Second, the compounding structure explains the enclosure configuration’s
distinctive severity: those who specialized most had committed the most prior
branching to the foreclosed path. Third, it explains why restoring access later
does not restore the position, since the intervening branches were not taken.

The strongest objection is that the distinction is one of degree. Nearly any
capability alters some later capability set: literacy, health, and income all
condition what a person can later do, so the second-order property fails to
isolate a class. The reply concedes the premise and denies the conclusion. The
thesis claims a difference in degree that is large enough to matter, assessed
per case rather than assumed. This is precisely why the framework requires a
declared domain $d$ and a declared horizon $H$: the analyst must
show, for the case at hand, that the affected organization has a branching
structure over the stated horizon. A case in which the affected capability is
effectively first-order over the relevant horizon should return a negative or
indeterminate materiality judgment, and the framework provides the vocabulary to
say so.

A second objection holds that the argument proves the importance of generative
capability without showing that its contraction by another party is
wrongful. That is correct, and Section 5.8 addresses it separately.

5.3 Relational Constitution of Agency Coordinates

This subsection argues that autonomy, direction, exit, and attribution belong
inside the capability object rather than alongside it.

Anderson and Honneth treat autonomy as an acquired set of capacities whose
development and maintenance depend on relations of recognition, so that
autonomy is vulnerable to the disruption of those relations
(Anderson & Honneth, 2005). Mackenzie links relational autonomy,
vulnerability, and capabilities, and argues for obligations to foster autonomy
rather than merely to refrain from interference (Mackenzie, 2013).

This work supports the enriched capability profile of Section 7. If autonomy is
relationally constituted, then an arrangement can leave a participant’s
reachable trajectories unchanged while altering who directs them, who is
credited for them, and whether the participant can exit. Those are candidate
capability coordinates rather than private psychological properties, and they
can be impaired by an identifiable relation.

The argument for including these coordinates inside the capability object, in
place of treating them as a separate normative overlay, is that the alternative
misdescribes the enclosure and priority configurations. Suppose a participant’s
reachable trajectories are unchanged while direction over them transfers to
another party. On a separated account, capability is intact and something else
has gone wrong. But the participant can no longer choose among the trajectories,
and Section 5.1 established that capability is a set of alternatives available
for choice. A set from which one cannot choose is not the participant’s
opportunity set at all; it is a description of what will happen. Direction, exit,
and attribution are therefore conditions on the capability’s being the
participant’s, and they belong inside the object.

The support has strict limits. This literature does not establish that every
generative capacity is wholly socially produced, that every contributor to a
practice has standing with respect to it, or that relational genesis entails
common ownership. Path connectivity, causal contribution, authorship, standing,
and ownership remain separate predicates throughout this paper. Section 5.4
argues for the standing condition directly.

5.4 Contribution, Dependence, and Participant Standing

This subsection resolves the universal-contribution paradox of Section 2 and
supplies the argument that the standing input of Section 8 requires.

The paradox is genuine and is the strongest structural objection to the whole
proposal. If a capability is assembled from the linguistic, practical, and
relational contribution of nearly everyone, then a contribution-based standing
rule makes nearly everyone a participant, so that either the diagnosis applies
universally or the category loses the contrast class that gives it content. The
literature of Section 6.4 sharpens this without resolving it.

Participant standing with respect to a generative domain is constituted by
sustained participation that maintains or transforms the practice, together with
reliance such that the participant’s own trajectory is conditioned on continued
access. Causal contribution to the ambient field supplies evidence for standing
without constituting it.

The argument begins by identifying an equivocation. The paradox moves from
“contributed to the field from which the capability was assembled” to “has
standing with respect to that capability.” Those are different relations.
Causal contribution is a historical fact admitting no natural threshold, and
Section 6.4’s sources establish that it is broadly distributed. Standing is a
normative status determining who can be wronged by a given foreclosure. Nothing
requires these to have the same extension.

The thesis then supplies the differential condition the paradox needs. Consider
two people, each of whom contributed some quantum of language and practice to
the field from which a capability was assembled. One has built a working
practice that depends on continued access to that capability and participates in
maintaining it; the other has not. Foreclosure of access removes reachable
trajectories from the first and none from the second. There is a loss in the
first case and no loss in the second, and this difference is prior to any
question of desert. Standing tracks who stands to lose, and the dependence
condition captures exactly that.

This has three consequences that the framework relies on. Universal contribution
becomes compatible with differentiated standing, dissolving the paradox without
denying either conjunct. Standing becomes assessable from the participant’s
position and trajectory, which are the objects the formalism already represents,
rather than from an intractable historical attribution problem. And standing
becomes independent of origination: a participant who originated nothing but
whose practice is now constituted by the commons has standing, while an
originator who has moved on may lack it.

The strongest objection is that the thesis rewards specialization and thereby
protects those who committed most deeply to a controlled dependency, which seems
to reward imprudence and to grant the most protection to those who took the
largest voluntary risk. The reply is that this consequence is correct and is a
mark in the thesis’s favor. The enclosure configuration is paradigmatic
precisely because reliance was induced by the controller’s own decision to open
access. A framework that withheld standing from the most deeply reliant
participants would exempt the clearest case it was built to describe. Where
reliance was genuinely unreasonable, the appropriate treatment runs through the
admissibility filter and through separate questions of consent and assumption of
risk, which Section 8 keeps distinct from standing.

A second objection is that the thesis makes standing turn on a fact the
controller can manipulate, since a controller who never opens access creates no
reliance and so faces no participants with standing. The reply is that this is
a correct description of a real asymmetry rather than a defect: a controller who
never opens access forecloses nothing, and the diagnosis should return a
negative result. Section 4’s tradition review supplies the separate vocabularies
in which never opening access may still be criticized.

5.5 Recognition, Erasure, and Epistemic Standing

This subsection introduces a verified structure for the loss of standing that
accompanies the priority configuration and keeps it distinct from capability
contraction.

Fricker analyzes epistemic injustice as a wrong done to someone specifically in
their capacity as a knower, distinguishing testimonial injustice, in which
prejudice deflates the credibility accorded to a speaker, from hermeneutical
injustice, in which a gap in collective interpretive resources prevents a
person from rendering an experience intelligible (Fricker, 2007).

The priority configuration in Section 2 has an epistemic dimension that
capability language alone does not capture. When a better-positioned
participant occupies a recognized trajectory, the originator may lose not only
reachable futures but also the standing to be heard as the source. Attribution
and recognition therefore appear in this paper twice: as candidate capability
coordinates in Section 7, where they are measured, and as inputs to participant
standing in Section 8, where they bear on whether the participant is eligible to
be a subject of the diagnosis at all.

The two must remain distinct. Testimonial injustice can occur without any
change in reachable repertoire, and capability can contract without any
credibility deficit. The present paper records the connection as a frequent
co-occurrence with separate mechanisms, and it does not merge epistemic
injustice into the diagnostic.

5.6 Externalization and the Loss of Knowledge

This subsection identifies the closest verified philosophical antecedent for a
non-wage form of dispossession and states the difference.

Stiegler analyzes proletarianization as a loss of knowledge rather than as
impoverishment or as a class position (Stiegler, 2010). On this
account, grammatization is the process by which gesture, speech, and eventually
affect and reasoning are discretized, externalized into technical supports, and
thereby made reproducible and transmissible. The externalization is
constitutive of human technicity and is not in itself a loss. It becomes a loss
when the knowledge deposited in the technical support ceases to be recoverable
by those from whom it was drawn, so that they retain access to the system’s
outputs while losing the capacity that produced them.

This is the closest available antecedent to the present concern, and the
affinity should be stated plainly. Both analyses locate a form of dispossession
that operates without depletion, without a wage relation, and without
transferring any inventoriable object. Both treat the capacity to produce,
rather than the product, as what is lost.

Three differences are equally important. Stiegler’s target is a civilizational
process operating at the scale of technical epochs; the present proposal
requires an identified actor, an identified comparison, and an identified
benefit path in a specific relation. Stiegler’s analysis is diagnostic of a
condition; the present proposal is a test that can return a negative result.
And Stiegler’s framework treats the process as pharmacological, capable of both
poison and remedy, whereas the present diagnostic takes no position on whether
externalization is generally beneficial. The present paper therefore uses
Stiegler for the shape of the phenomenon and derives none of its conditions from
him.

5.7 The Relational and Temporal Object of Appropriation

This subsection argues for the paper’s account of what the relation acts upon,
and thereby justifies the form of the formal object in Section 7.

Generativity exploitation acts on the conditions under which a participant
continues to become capable. Its object is the reproduction of a capacity across
time from a particular position, and that object is relational and temporal
rather than a stock the participant holds.

The argument proceeds from a constraint on appropriation frameworks. Any account
of appropriation presupposes an object that can pass from one party to another,
because appropriation is a transfer notion. Apply the requirement to each of the
four objects of Section 3.3. Artifacts are copied, and copying is duplication in
place of transfer. Latent organization is retained in full, so nothing passes.
The participant’s situated position and the realization infrastructure do change
hands in the relevant sense; and neither is a thing. A position is constituted by
relations to other participants, institutions, and conditions, and infrastructure
is appropriable only as a set of control relations over access.

The conclusion is that if the configurations of Section 2 involve appropriation
at all, its object is relational. Adding the branching structure of
Thesis ? makes it temporal as well: what is at stake is the
participant’s continued becoming-capable across a horizon. This is the direct
justification for representing capability as a set of reachable trajectories
under an arrangement, in place of a quantity, a stock, or a set of end states.
The formalism has the shape it has because the object has this structure.

The strongest objection is that the thesis makes appropriation metaphorical,
since nothing is taken and the vocabulary of exploitation becomes strained
without a transferred substance. The reply accepts the observation and denies
that it defeats the thesis. Sections 5.1 and 4.2 record that the exploitation
literature already contains accounts without a common transferred substance:
Cohen argues that the charge survives the removal of the labor theory of value
(Cohen, 1979), and Roemer reconstructs the relation through property
positions and a withdrawal comparison (Roemer, 1982). The present
account continues that movement. What passes between the parties is the benefit
constituted through the affected position, and Section 8 requires a documented
rule or an identified causal contrast to establish it.

5.8 Justificatory Burden and Normative Significance

This subsection argues for what follows from a positive diagnosis. It answers
the question deferred in Sections 5.3 and 5.2, and it supplies the reason the
diagnostic is worth constructing.

Section 4.3 recorded that the paper declines to make its diagnostic
constitutively wrongful. That decision requires an argument, because a purely
descriptive relation with no normative consequence would be an idle
classification.

A positive diagnosis generates a defeasible reason for concern and places a
burden of justification on the beneficiary. It establishes neither
all-things-considered wrongfulness nor liability, both of which require
consent, procedure, alternatives, and distributive assessment as further inputs.

The argument for the burden’s existence turns on what the benefit-through
condition establishes. When the condition is satisfied, the beneficiary’s
advantage is conditioned on the affected participant’s diminished position:
restoring that position would reduce the advantage. The beneficiary therefore
holds a standing interest in the continuation of the diminution, whether or not
the beneficiary intends it, knows of it, or acts to maintain it. A structural
interest in another’s diminished capability is the kind of relation that calls
for justification, and this holds independently of the beneficiary’s mental
states. This is why Section 8 excludes intention from
Equation (5) while Section 10 returns it as a separate variable.

The argument for the burden’s limits turns on what the condition leaves open.
Anderson argues that egalitarian justice concerns the character of social
relations and the elimination of hierarchies of domination and status, in place
of a purely distributive criterion (Anderson, 1999). On such a view a
relation’s normative assessment depends on the terms on which parties stand to
one another, and the benefit-through condition constrains those terms without
fixing them. Two arrangements can satisfy the diagnostic identically while
differing in consent, in the availability of alternatives, in procedural
recourse, and in the distribution of the resulting gains. Those differences bear
directly on whether the relation is acceptable. A framework that returned a
verdict of wrongfulness from the diagnostic alone would suppress them.

Thesis ? therefore locates the diagnostic precisely. It is
stronger than a description, because it shifts a burden. It is weaker than a
verdict, because the burden can be discharged. This is also what distinguishes
the proposal from the accounts of Section 4.3: Sample and Vrousalis build
wrongfulness into the relation, and the present diagnostic separates the
structural finding from its normative resolution so that the two can be
contested independently.

The strongest objection is that a shifted burden of justification is either
trivial, since anyone can demand justification for anything, or covertly
equivalent to a wrongfulness verdict, since a burden that is rarely discharged
functions as a presumption of guilt. The reply distinguishes the two horns. The
burden is non-trivial because it is triggered by a specific and demanding
conjunction that Section 9 shows can fail in several ways, including cases of
coincident loss and gain. And it stops short of a verdict because Section 8
supplies the categories through which it is discharged: a defensible
admissibility argument, an alternative available to the affected participant, a
procedure through which the terms were set, or a distribution of gains that
addresses the diminution.

5.9 Dependence of the Framework on the Four Theses

This subsection records what the framework depends on, so that a reader can
locate a disagreement precisely.

The relation between Sections 6 and 7 through 8 has two components that should
be kept apart. Formal independence: every proposition in Sections 8 through 9 is derived from declared model assumptions, and no derivation cites a
thesis of this section. A reader who rejects all four theses can still verify
each proof. Justificatory dependence: the framework’s claim to be worth
constructing rests on these theses, and rejecting one narrows or defeats a
corresponding part of the proposal.

The dependencies are as follows. Rejecting Thesis ? defeats
the choice of register: if the configurations of Section 2 involve no loss
representable as opportunity, then possession or output suffices and the
capability object is unmotivated. Rejecting Thesis ? narrows
the paper to a general capability-and-benefit framework with no distinctive
subject matter, and the domain-and-horizon indexing loses its rationale.
Rejecting Thesis ? reopens the universal-contribution paradox,
which either universalizes the diagnosis or empties it; the formal apparatus
would survive with an undischarged standing input. Rejecting
Thesis ? removes the justification for the trajectory-set
form of the capability object, leaving the formalism intact but arbitrary.
Rejecting Thesis ? makes the diagnostic an idle classification
with no consequence for how a relation should be assessed.

Section 10 states the corresponding revision commitments.

6. Realization Conditions and the Political-Economic Register

This section develops the political-economic content that motivated the
research question. Its objective is to describe the mechanisms through which
control of realization conditions can operate, to supply verified vocabulary for
advantage obtained through position, and to state precisely why control alone is
insufficient for the diagnosis.

6.1 From Means of Production to Means of Generative Expansion

This subsection states the paper’s guiding political-economic conjecture and
the correction that limits it.

The production-centered tradition of Section 4.1 locates exploitation in
ownership of the means of production. The configurations of Section 2 suggest a
displaced structure: the party who benefits need control neither the
participant nor the participant’s generative organization, provided it controls
the conditions under which that organization can expand. Priority, funding,
collaborators, compute, institutional permission, and channels of distribution
are the operative holdings.

This suggests the conjecture that control has shifted toward the conditions
permitting generativity to expand. The conjecture requires an immediate
correction, and the correction is more important than the conjecture. Condition
control is one mechanism among several, and it is not the phenomenon’s essence.
The central relation is the contraction of another participant’s ability to
expand a generative organization, and condition control is one route by which
such contraction is produced. Interruption, redirection, dependency formation,
capture of direction, and foreclosure of initiation are others. A framework
built on condition control alone would misclassify every case in which
capability contracts through a mechanism other than control.

The conjecture is recorded here as a working conjecture with unresolved
empirical status. It is not used as a premise anywhere in the formal
construction.

6.2 Rent, Assetization, and Control Rights

This subsection supplies verified political-economic vocabulary for advantage
obtained through position rather than through production.

Birch analyzes contemporary technoscientific capitalism as organized around
rentiership: the appropriation of value through ownership and control rights,
monopoly conditions, and regulatory or market devices, rather than through
entrepreneurial commodity production (Birch, 2020). On this
account things are configured as assets, and accumulation strategies follow from
that configuration. Rent is analyzed as a social process involving actions,
knowledges, and transformations that make value realization possible over
extended periods.

This vocabulary matters to the present proposal for a specific reason. The
constitutive benefit-through witness of Section 8 requires a documented rule
through which a beneficiary’s advantage is constituted by the affected party’s
position: a transfer, an entitlement, a priority rule, an exclusivity relation,
or a control allocation. Rentiership is the political-economic category to
which such rules belong. Birch’s analysis supplies the account of how control
rights generate returns; the present paper supplies the additional requirement
that the return be traceable to a specific affected capability position.

6.3 Platform Bottlenecks and Infrastructural Position

This subsection supplies verified accounts of control exercised through
architectural position.

Srnicek analyzes platforms as an economic form defined by their position as
intermediaries between groups of users, their reliance on network effects, and
their capacity to extract and control data generated by the interactions they
mediate (Srnicek, 2017). The relevant feature here is architectural:
a platform’s advantage derives from occupying the position through which
interaction must pass.

Khan analyzes the gap between such architectural power and the antitrust
framework used to assess it, arguing that a consumer-welfare standard focused on
short-run price effects fails to capture power that operates through structural
position, vertical integration, and control of infrastructure on which
competitors depend (Khan, 2017).

Both accounts bear on the open-then-enclose configuration of Section 2. A
controller who supplies infrastructure on which others build acquires the
capacity to alter their reachable trajectories by altering access terms, and the
resulting power is not registered by price-centered tests. The present paper
uses these accounts to establish that such positions exist and are analytically
recognized. It does not infer from the existence of such a position that any
particular relation satisfies the diagnostic, and it makes no claim about any
firm discussed in those works.

6.4 Contributed Input Without Corresponding Return

This subsection addresses the distributed-contribution configuration and
supplies its verified antecedents.

Terranova analyzes free labor in the digital economy: the unpaid, often
voluntary, and frequently pleasurable activity through which users produce
value that is captured by firms, and which is simultaneously given freely and
appropriated (Terranova, 2000). Couldry and Mejias analyze data
appropriation as a form of colonization, in which human life is configured so
that data can be continuously extracted from it, and they connect this to the
historical structure of colonial appropriation (Couldry & Mejias, 2019).

These accounts supply the closest verified treatments of the third
configuration, in which a capability is assembled from broadly distributed
contribution while its realization infrastructure is narrowly controlled. They
establish that the pattern is real, analytically described, and normatively
contested.

They also sharpen the paradox rather than resolving it. If contribution is
genuinely universal, then either the exploitation category applies universally
or it loses its contrast class. The present framework’s response is to separate
contribution from standing: contribution may be broadly distributed while
standing with respect to a specific generative domain requires a defended
argument concerning participation, reliance, transformation, maintenance, or
entitlement. Section 8 accordingly treats standing as an external three-valued
input rather than as an inference from causal contribution. This response is a
proposal and is not established by the cited works.

6.5 Actor Separation in the Artificial-Intelligence Configuration

This subsection decomposes the configuration that motivated the research
question and states why treating it as a single actor makes the question
malformed.

The question that generated this research program asked whether artificial
intelligence makes exploitation more aggressive or changes its mechanism. The
question cannot be assessed while the relevant parties are treated as one actor.
Table 2 separates the positions that must remain distinct.

| >p0.24

X
p0.26

Position	Characteristic holding or role	Diagnostic relevance
Finance or capital	Supplies and directs investment; holds claims on returns	Candidate beneficiary; rarely the acting node
Provider or infrastructure controller	Controls the system, its access terms, and its deployment conditions	Candidate controller and beneficiary; site of constitutive rules
Adopting organization	Deploys the system within its own practices and reorganizes work around it	Candidate acting node with respect to affected practitioners
Labor or professional practice	Holds the domain generativity and depends on realization conditions	Candidate affected participant
Users, communities, and historical contributors	Supply interaction traces and the historical corpus from which capability was assembled	Candidate affected participants with contested standing
The system itself	Executes the mediating function	Standing, interests, and agency remain open
The living generative commons	The practice reproduced and transformed across participants	Candidate affected object with no single carrier
Mediating institutions	States, regulators, unions, and civil-society bodies	Alter admissibility and available comparisons

Table. Distinct Actor Positions in a Mediated Generative Configuration

Three consequences follow. The acting node, the technical mediator, the
controller of conditions, the affected participant, and the ultimate beneficiary
are distinct roles that a single description may collapse. Attribution of
exploitation to the system itself would require an account of its agency,
interests, standing, and control that this paper does not supply and treats as
open. And the comparative question of intensification is not a single question:
it decomposes into indexed hypotheses about scale, speed, opacity, reach,
reversibility, dependency formation, bargaining asymmetry, concentration, and
propagation rate, each requiring its own baseline and evidence.

6.6 Concentration Without Automatic Diagnosis

This subsection states and argues for the political-economic position the paper
defends.

Concentrated control of realization conditions confers a capacity to amplify one
participant’s trajectory while constraining another’s, and that capacity is a
structural feature of the arrangements described in Sections 7.2 through 6.4. A
relation becomes a candidate instance of generativity exploitation only through
participant standing, an independently admissible comparison, an actor-linked
mechanism, a material capability disadvantage, and a typed benefit-through
witness.

The argument for the first half is the reviewed literature of Sections 7.2 through 6.4 together with Thesis ?. Rentiership, intermediary
position, and appropriation of distributed contribution are described mechanisms
by which returns accrue to control over conditions. The compounding structure of
generative capability makes control over those conditions consequential beyond
the period in which it is exercised, because it shapes which branches are taken.
The paper therefore treats concentration as a serious matter and declines the
view that non-rivality renders it harmless.

The argument for the second half is that the capacity to constrain and the
relation being diagnosed have different truth conditions. A controller may hold
the capacity and never exercise it; may exercise it against a party with no
standing; may exercise it while every admissible comparison leaves the affected
party no better off; or may exercise it while its own advantage arises through
an independent path. Section 9.1 proves the last case formally. Inferring the
diagnosis from the capacity would collapse these distinctions and reproduce
exactly the error that Section 4.4 attributes to strong structural accounts.

The strongest objection is that the second half renders the framework
politically inert, since the evidentiary demands of standing, comparison,
mechanism, and witness will rarely be met, so that concentration continues
undiagnosed. The reply is that indeterminate results are informative under the
reporting rule of Definition ?: an indeterminate
profile records which specific input is missing, and that record is a research
and disclosure demand directed at the party positioned to supply it. Section 10.1
makes this the protocol’s standard output. A framework that returned positive
findings on weaker evidence would purchase its political usefulness by becoming
unable to be wrong.

The paper classifies no observed labor, academic, platform, commons, or
industrial relation.

7. Reachable Repertoires and Effective Capability

This section constructs the paper’s primary capability object in continuous
dependency order. It separates the analytical registers the preceding sections
motivated, declares the trajectory space and arrangement, defines a reachable
viable repertoire and an enriched profile, and derives the limited conditions
under which one arrangement expands another’s repertoire.

7.1 Analytical Registers

This subsection fixes the objects that the remaining definitions connect. The
separation is substantive: an arrangement can change feasible actions or control
while leaving latent organization and observed output unchanged.

Table 3 records each register, its formal object, the use it
licenses, and the warrant that remains outstanding. The registers implement the
four-object separation of Section 3.3 and the mixed-scarcity requirement of
Section 3.2, which enters through the arrangement’s realization conditions.

| >p0.19

p0.20
X
X

Register	Formal object	Licensed use	Outstanding warrant
Latent organization	$G\in\mathfrak G_d$	Index a candidate generative organization	Carrier, domain, and empirical interpretation
Complete arrangement	$r\in\mathfrak R$	Supply actions, institutions, relations, kernels, and viability	Scope, consistency, and transition feasibility
Reachable repertoire	$\Pp_Y(G;r,h)$	Represent viable trajectories reachable at declared resolution	Horizon, metric, action family, probability threshold
Enriched capability	$\K_Y^\Theta(r)$	Add predeclared autonomy, direction, attribution, continuity, or exit dimensions	Measurement, partial order, and materiality
Comparison family	$(\B^F,\Lambda)$	Retain technically feasible, admitted, excluded, and unresolved alternatives	Technical feasibility and non-circular admissibility
Relational mechanism	$M_{X\to Y}^{\Theta}(m;b)$	Connect an actor-linked relation to the capability difference	Causal, institutional, and coordinate evidence
Benefit-through witness	$w=(\Psi,Z_Y,\text{model})$	Connect the affected capability state to a beneficiary advantage	Causal identification or institutional verification
Diagnostic profile	$\Pi(m)$	Preserve baseline-specific support, rejection, and uncertainty	Decision use, normative interpretation, and external validity

Table. Analytical Registers in the Proposed Diagnostic

7.2 Arrangements and Trajectory Space

This subsection declares the comparison units and the measurable structure the
repertoire definition requires.

Fix an affected actor $Y$, domain $d$, history $h$, horizon $H>0$, and latent
organization $G\in\mathfrak G_d$. A case-relative complete arrangement is a
tuple

$$r=\langle C^r,\A^r,\mathbb P^r,\V^r,\mathcal I^r,
\mathcal R^r,\beta^r\rangle\in\mathfrak R,$$

containing the realization conditions, policy family, trajectory law, viability
event, actor and institutional positions, constitutive rules, and benefit map
required by the declared question. Completeness is relative to the case and
horizon rather than metaphysical exhaustiveness. The realization conditions
$C^r$ carry the mixed-scarcity distinctions of Section 3.2, so that a
comparison records which condition class differs.

Let $(\T_{d,H},\rho)$ be a metric trajectory space and let $\F$ be a
$\sigma$-algebra containing its Borel sets, so every closed metric
neighborhood used below is measurable. For each complete arrangement
$r\in\mathfrak R$, let
$\A_Y(r,h)$ be a nonempty feasible policy set,
$\V_Y(r,h)\in\F$ a viability event, and
$\mathbb P_{Y,G}^{r,a}(,\cdot\mid h)$ a probability kernel on $\T_{d,H}$ for
each $a\in\A_Y(r,h)$. Fix a resolution $\delta>0$ and probability threshold
$\epsilon\in(0,1]$, and write

$$\N_\delta(\tau)={\tau’\in\T_{d,H}:\rho(\tau,\tau’)\leq\delta}.$$

7.3 Reachable Viable Repertoire and Enriched Profile

This subsection defines the capability object and the enrichment that
Section 5.2 motivates.

The effective reachable repertoire of $Y$ under arrangement $r$ is

$$\Pp_Y^{H,\epsilon,\delta}(G;r,h)

\left{
\tau\in\V_Y(r,h):
\exists a\in\A_Y(r,h):
\mathbb P_{Y,G}^{r,a}
\bigl(\N_\delta(\tau)\cap\V_Y(r,h)\mid h\bigr)
\geq\epsilon
\right}.$$

The indices $(H,\epsilon,\delta)$ are part of the object. A deterministic model
is the special case in which each kernel is concentrated on one trajectory.
Equation (1) expresses effective possibility under a model;
it neither reconstructs every future nor assigns value to every reachable
trajectory. It is the formal counterpart of the set of alternatives available
for choice discussed in Section 5.1.

A trajectory set may omit capability-relevant relations when two arrangements
permit the same paths but allocate direction, exit, attribution, or control
differently. Let $q_Y^\Theta(r)$ contain only such independently declared
dimensions, with specification $\Theta$ stating their measurement rules. The
admissible coordinates are those Section 5.2 supports and Section 5.3 requires
to be distinguished from standing.

The enriched capability profile is

$$\K_Y^\Theta(r)

\left(
\Pp_Y^{H,\epsilon,\delta}(G;r,h),
q_Y^\Theta(r)
\right)
\in\mathfrak K_\Theta.$$

The specification $\Theta$ equips $\mathfrak K_\Theta$ with a partial preorder
$\succeq_\Theta$. Write $k_1\succ_\Theta k_0$ when
$k_1\succeq_\Theta k_0$ and $k_0\not\succeq_\Theta k_1$. A materiality rule
$\mu_\eta^\Theta(k_1,k_0)\in{1,0,?}$ supplies declared resolution
thresholds, and

$$k_1\succ_{\Theta,\eta}k_0
\quad\Longleftrightarrow\quad
k_1\succ_\Theta k_0
\ \text{and}
\mu_\eta^\Theta(k_1,k_0)=1.$$

Incomparability and unresolved materiality are admissible results. No scalar
aggregation is presumed.

The materiality rule may contain domain-specific resolution thresholds, a
Pareto requirement over protected dimensions, or another justified decision
rule. It must be specified before classification and tested for sensitivity.
Robeyns’s warning about dimension selection and weighting applies directly
here (Robeyns, 2005).

7.4 Conditional Monotonicity and Separation

This subsection derives the section’s two results. The first states the
conditions under which one arrangement’s repertoire includes another’s; the
second separates possession from capability.

Arrangement $r’$ expands $r$ for the declared model when

• $\A_Y(r,h)\subseteq\A_Y(r’,h)$;
• $\V_Y(r,h)\subseteq\V_Y(r’,h)$; and
• for every $a\in\A_Y(r,h)$ and $\tau\in\V_Y(r,h)$, $$\mathbb P_{Y,G}^{r’,a}
  \bigl(\N_\delta(\tau)\cap\V_Y(r’,h)\mid h\bigr)
  \geq
  \mathbb P_{Y,G}^{r,a}
  \bigl(\N_\delta(\tau)\cap\V_Y(r,h)\mid h\bigr).$$

Under Assumption ?,

$$\Pp_Y^{H,\epsilon,\delta}(G;r,h)
\subseteq
\Pp_Y^{H,\epsilon,\delta}(G;r’,h).$$

Take $\tau\in\Pp_Y^{H,\epsilon,\delta}(G;r,h)$. Definition
? gives $\tau\in\V_Y(r,h)$ and a witnessing policy
$a\in\A_Y(r,h)$ with qualifying probability at least $\epsilon$.
Assumption ?(2) places $\tau$ in $\V_Y(r’,h)$.
Conditions (1) and (3) keep $a$ available and preserve its qualifying
probability. The same policy therefore witnesses membership under $r’$.

The proposition is deliberately conditional. An additional resource or formal
permission can alter transition kernels, impose control, reduce autonomy, or
change viability; ordinary language about “more conditions” supplies none of
Assumption ?. Its converse also fails. This is the formal
reason why the political-economic observation of Section 6.1 cannot be
converted directly into a diagnosis: supplying conditions and expanding
capability are distinct relations.

There exist two arrangements with the same latent organization $G$ and
different reachable repertoires; equality of $G$ therefore does not entail
equality of effective capability.

Let $\T={\alpha,\beta}$ have the discrete metric, let both trajectories be
viable, take $0<\delta<1$ and $\epsilon=1$, use deterministic kernels, and keep
$G$ fixed. Arrangement $m$ permits only a policy reaching $\alpha$;
arrangement $b$ permits policies reaching $\alpha$ and $\beta$. Then
$\Pp_Y(G;m,h)={\alpha}$ and
$\Pp_Y(G;b,h)={\alpha,\beta}$. Thus fixed latent organization permits
different capability. Even with equal repertoires, unequal
$q_Y^\Theta$ can distinguish actor-controlled from externally directed
continuation.

Proposition ? is the formal statement of the
retention-with-foreclosure asymmetry described in Section 3.3. It is
elementary, and its elementary character is the point: once capability is
represented as a set of reachable futures under an arrangement, the
independence of possession from capability follows immediately.

8. Comparison Arrangements and Benefit-Through Witnesses

This section assembles the diagnostic after the capability object has been
defined. It separates technical feasibility from normative admissibility,
introduces standing, an actor-linked relational mechanism, and material
disadvantage, distinguishes two benefit-through witness types, and returns a
baseline-indexed three-valued profile.

8.1 Technically Feasible and Admitted Comparison Families

This subsection constructs the comparison family in two stages. The
construction responds to the benchmark problem stated in Section 4.2 and to the
structural-scope constraint stated in Section 4.4.

Let $m\in\mathfrak R$ denote the arrangement under assessment. A baseline
cannot be selected by calling it “non-exploitative,” because that would
presuppose the desired result.

The technically feasible, independently admitted, and unresolved comparison
families are

$$\B^{F}(m)
&={b\in\mathfrak R:\Tech(b\mid m)=1},\
\B_\Lambda(m)
&={b\in\B^{F}(m):\Adm_\Lambda(b\mid m)=1},\
\B_\Lambda^{?}(m)
&={b\in\B^{F}(m):\Adm_\Lambda(b\mid m)=?}.$$

The specification $\Lambda$ discloses resource, institutional, rights,
standing, and transition constraints. Its admissibility rule may not invoke
$\GEX$ or the conclusion of the present diagnostic. The complete report keeps
every $b\in\B^F(m)$, including rows with admissibility $0$ or $?$. An empty
admitted family yields comparator underdetermination.

The two-stage construction places background injustice where Section 4.4
requires it. A judgment that the prevailing distribution of realization
conditions is unjust may enter $\Lambda$, where it must be stated and defended
as a comparison principle. It may not enter as a substitute for the mechanism
and witness clauses below.

8.2 Standing, Mechanism, and Material Disadvantage

This subsection defines the three row-level statuses that precede the benefit
path. Standing is the external input discussed in Sections 6.3 and 6.4.

For each $b\in\B^{F}(m)$, the external input
$S_Y^\Lambda(G,d,h;m,b)\in{1,0,?}$ records whether $Y$ has relevant
participant standing under $\Lambda$. The status
$M_{X\to Y}^{\Theta}(m;b)\in{1,0,?}$ records whether an actor-linked
intervention, rule, or institutional relation connects the focal arrangement
to the capability difference evaluated under $\Theta$: $1$ records a supported
actor link, $0$ records an adequate result of independence, and $?$ records an
unresolved relation or affected coordinate. The
material-disadvantage status is

$$D_Y^{\Theta,\eta}(m;b)=
\begin{cases}
1,
&\K_Y^\Theta(b)\succ_\Theta\K_Y^\Theta(m)
\text{ and }\mu_\eta^\Theta(\K_Y^\Theta(b),\K_Y^\Theta(m))=1,\
0,
&\K_Y^\Theta(m)\succeq_\Theta\K_Y^\Theta(b),\
0,
&\K_Y^\Theta(b)\succ_\Theta\K_Y^\Theta(m)
\text{ and }\mu_\eta^\Theta(\K_Y^\Theta(b),\K_Y^\Theta(m))=0,\
?,
&\substack{\text{when profiles are incomparable, or materiality}\
\text{or evidence remains unresolved.}}
\end{cases}$$

Contested standing, an unresolved actor link, and unidentified capability are
recorded as $?$, not as $0$. A natural disaster or impersonal scarcity can
produce $D=1$ while $M=0$ or $?$, and therefore supplies no candidate result
through disadvantage alone.

The standing input $S_Y^\Lambda$ is where Thesis ? enters the
construction. The thesis supplies its content: standing is constituted by
sustained participation that maintains or transforms the practice, together with
reliance such that the participant’s trajectory is conditioned on continued
access. Genealogy, creation, and causal contribution supply evidence for that
status without constituting it, and common origin entails neither equal
contribution nor a specific ownership result. This is the framework’s response
to the universal-contribution paradox of Sections 3 and 6.4: broad causal
contribution is compatible with differentiated standing because the two track
different relations.

The value remains a declared external input for a reason. Thesis ?
fixes the kind of condition standing is; it does not settle, for a given case,
whether a participant’s reliance and participation reach the threshold. That
judgment requires domain evidence and admits of reasonable disagreement, and
$?$ is the correct value when it is unresolved.

8.3 Typed Benefit and the Two Witness Forms

This subsection defines beneficiary advantage and separates the two ways a
benefit can run through an affected capability position.

Fix a benefit specification

$$\Psi=\langle J_\Psi,\mathcal Y_X^\Psi,\succeq_\Psi,\eta_B\rangle,
\qquad
\mathcal Y_X^\Psi=\prod_{j\in J_\Psi}\mathcal Y_{X,j},$$

where the index set lists admitted benefit types, $\succeq_\Psi$ is a partial
preorder, and $\eta_B$ is a typed materiality schedule. Money, control,
priority, recognition, and bargaining power remain distinct coordinates unless
an aggregation rule is independently justified. Let
$\beta_X^\Psi(r)\in\mathcal Y_X^\Psi$, and write
$u\succ_{\Psi,\eta_B}v$ when $u$ strictly exceeds $v$ under the declared
partial order by a typed material amount. The displayed causal form uses one
predeclared scalar coordinate $B_{X,j}^\Psi$ and its declared threshold
$\eta_{B,j}\geq 0$.

For every baseline row $b$, declare a row-specific state map $Z_Y^b(r)$ for
$r\in{m,b}$. It names the concrete capability coordinate, subordination,
capture, or redirection used in the $D$ comparison. When $D=1$, its contrast
must underwrite that same disadvantage judgment. The binary value $D$ is not
itself a mediator.

A causal witness $w$ specifies an SCM or potential-outcome model in which

$$\Delta_{\BT}^{w}(m,b)

\mathbb E!\left[
B_{X,j}^\Psi\bigl(m,Z_Y^b(m)\bigr)
-B_{X,j}^\Psi\bigl(m,Z_Y^b(b)\bigr)
\right]>\eta_{B,j}.$$

The witness records the meaning of interventions, consistency and composition
assumptions, interference scope, identification conditions, and sensitivity
analysis. Equation (4) is a mediation-style estimand; observed
loss and gain do not identify it. Let $W^c_{X,Y}(m;b)$ be $1$ when the material
contrast is supported, $0$ when an adequate design supports its material
absence, and $?$ when the intervention, model, or identification remains
inadequate.

A constitutive witness $w$ is a verified institutional rule, transfer map,
exclusivity relation, priority rule, or control allocation $\rho_Z^m$ through
which $X$’s specified advantage is constituted by the affected position
$Z_Y^b(m)$. The witness specifies a complete matched rule-replacement
arrangement $m^{-\rho_Z;b}$ that deactivates or replaces that link, records
every induced change needed for coherence, and satisfies

$$\beta_X^\Psi(m)\succ_{\Psi,\eta_B}\beta_X^\Psi(m^{-\rho_Z;b}).$$

The witness is a rule-relative constitutive comparison rather than an estimated
mediated effect or an automatic ceteris-paribus substitution. Let
$W^k_{X,Y}(m;b)$ be $1$ when the exact rule, matched arrangement, and material
benefit change are supported, $0$ when governing rules establish independence,
and $?$ when the rule or comparison remains unresolved.

The constitutive witness is the point at which the rentiership vocabulary of
Section 6.2 becomes operative. Priority rules, exclusivity relations, transfer
maps, and control allocations are precisely the devices through which returns
are appropriated by position, and a constitutive witness requires that such a
device be documented and its removal specified as a complete alternative
arrangement.

Causal mediation supplies the discipline for the other branch. Imai, Keele, and
Tingley distinguish a causal mediation estimand from its identification,
estimation, and sensitivity analysis (Imai et al., 2010). That distinction
is adopted here: a constitutive institutional rule is treated separately from a
mediated causal effect, and neither correlation nor temporal order is promoted
to a benefit-through result.

The two witnesses answer different questions and cannot substitute for one
another. Let $W_{X,Y}^{\Psi}(m;b)\in{1,0,?}$ be $1$ when at least one
admissible witness is supported, $0$ when a predeclared complete witness family
has been tested and refuted, and $?$ when the path remains unidentified or the
witness family is incomplete.

8.4 Row Status and Diagnostic Profile

This subsection assembles the row predicate and defines the reporting rule.

For every $b\in\B^F(m)$, define

$$g_b(m)

\Adm_\Lambda(b\mid m)
\wedge
S_Y^\Lambda(G,d,h;m,b)
\wedge
M_{X\to Y}^{\Theta}(m;b)
\wedge
D_Y^{\Theta,\eta}(m;b)
\wedge
W_{X,Y}^{\Psi}(m;b),$$

using the strong three-valued conjunction: the result is $1$ when every clause
is $1$, $0$ when a clause is demonstrably $0$, and $?$ otherwise. A value of
$1$ is a candidate generativity-exploitation diagnosis relative to
$(b,\Lambda,\Theta,\eta,\Psi,w)$. A row excluded by $\Lambda$ receives $0$
under this reporting rule and retains its exclusion reason; the value does not
assert that the excluded arrangement is itself non-exploitative.

The complete result retains every technically feasible row:

$$\Pi(m)=
{(b,\Adm_\Lambda,S_Y,M_{X\to Y},D_Y,
W^c_{X,Y},W^k_{X,Y},W_{X,Y},g_b):
b\in\B^F(m)}.$$

Let $\mathcal A=\B_\Lambda(m)$ and $\mathcal U=\B_\Lambda^{?}(m)$. Apply the
following order. The profile is indeterminate when $\mathcal A$ is
empty, $\mathcal U$ is nonempty, or any admitted row has $g_b=?$; it is
baseline-robust when every admitted row has $g_b=1$; it is
relatively rejected when every admitted row has $g_b=0$; and it is
baseline-sensitive when resolved admitted rows include both values.
No voting rule collapses the profile, and excluded rows remain visible.

Intention, awareness, genealogy, consent, domination, wrongfulness,
responsibility, and remedy remain external variables. Their exclusion from
Equation (5) means that none is automatically decisive. They can
still change standing, admissibility, capability ordering, witness credibility,
and downstream judgment. The separation of domination from the diagnostic is
the formal counterpart of the contrast drawn in Section 4.3.

9. Diagnostic Consequences and Boundary Cases

This section stress-tests the construction with finite countermodels and
thought experiments. Its role is discriminative: the cases reveal which clause
does work and where the framework returns uncertainty. They supply no evidence
that a current institution or industry instantiates the diagnosis.

Table 4 summarizes the main cases. Its statuses concern the
proposed diagnostic; neighboring descriptions such as harm, domination, or
competition require their own criteria.

| >p0.22

p0.23
p0.23
X

Case structure	Capability relation	Mechanism and benefit-path relation	Diagnostic status
Independent innovation	$Y$ may be worse relative to a comparator	$X$’s gain is independent of $Y$’s disadvantage	Candidate rejected at that baseline; competition or displacement remains open
Impersonal scarcity	Material disadvantage can be supported	Actor-linked capability mechanism is absent or unresolved	Candidate rejected or indeterminate; deprivation remains open
Malicious suppression	Material disadvantage can be supported	Beneficiary advantage is absent	Candidate rejected; harm, violence, or domination remains open
Predatory exclusion	Standing, relational mechanism, and material disadvantage are supported	Verified transfer, priority, or causal path benefits $X$ through the affected state	Candidate supported relative to the declared baseline
Captured continuation	Output or repertoire can remain stable while autonomy, exit, direction, or attribution declines	Control allocation sustains $X$’s declared advantage	Enriched-profile candidate when all clauses are supported
Mutually enabling cooperation	Actual capability matches or exceeds admissible comparators	Reciprocal gains may occur	Disadvantage clause fails; local constraint is insufficient
Comparator plurality	Actual capability lies between two admissible alternatives	Witness status is held fixed	Baseline-sensitive profile
Evidence limitation	Profile order, standing, or path is unresolved	Causal or constitutive witness remains unidentified	Indeterminate profile

Table. Boundary Cases Under Capability, Mechanism, and Benefit-Path Evidence

9.1 Coincidence Without a Benefit Path

This subsection establishes that capability loss and beneficiary gain can
co-occur with no benefit running through the affected position.

A strict capability disadvantage for $Y$ and an observed gain for $X$ can
coexist while every causal benefit-through contrast in a declared model equals
zero.

Let $R\in{0,1}$ index arrangements, with focal arrangement $m=1$ and
admitted baseline $b=0$. In the scalar capability space with its usual order,
set $\K_Y^\Theta(R)=1-R$ and choose a materiality threshold below $1$.
Thus $D_Y(1;0)=1$. Let the actor-linked relation setting $R$ also set the
capability state $Z_Y^0(R)=R$, so its mechanism status is $1$. Define
$B_{X,j}^\Psi(R,z)=R$, independent of $z$. The observed benefit increases
from $0$ to $1$, while

$$B_{X,j}^\Psi(1,Z_Y^0(1))-B_{X,j}^\Psi(1,Z_Y^0(0))=1-1=0.$$

Declare the witness family complete and stipulate that the arrangement contains
no transfer, entitlement, priority, or control rule constituting $X$’s benefit
through $Z_Y^0$. The causal and constitutive branch statuses are therefore both
$0$, hence $W_{X,Y}^\Psi(1;0)=0$. With admissibility and standing fixed at $1$,
Equation (5) returns $g_0(1)=0$. Loss and gain co-occur while the
affected state carries none of the gain.

This countermodel formalizes the distinction between independent innovation
and predatory exclusion. If a new practice lowers another practice’s market
prospects while its benefit arises independently, displacement may occur but
the proposed exploitation predicate is not satisfied. If an exclusivity rule
transfers the excluded party’s access, priority, collaborators, or controlled
branches to the beneficiary, a constitutive witness may be available. The
action’s subject matter need not match the affected generativity; the path does
the classificatory work.

The result also answers the concern raised in Section 4.4 from the opposite
direction. Just as structural position cannot substitute for a benefit path,
temporal coincidence within a concentrated arrangement cannot either.

9.2 Baseline Sensitivity

This subsection establishes that admissible comparisons can disagree, and that
the profile must therefore be preserved.

Two independently admissible baselines can yield different candidate statuses
for the same assessed arrangement.

Let $\B_\Lambda(m)={b_1,b_2}$ and $\B_\Lambda^?(m)=\varnothing$, with all
row inputs resolved. Choose the capability ordering
$Y^(b_1),_Y^(m)
Y^(b_2)$. Hold admissibility, standing,
relational mechanism, and an admissible benefit-through witness fixed at $1$.
Then $D(m;b_1)=1$ and $D(m;b_2)=0$, so
$g{b_1}(m)=1$ and $g{b_2}(m)=0$. Definition
? reports a baseline-sensitive profile.

The result is elementary and consequential. A raw before–after comparison, an
ideal arrangement, a market outcome, and a rights-respecting feasible
alternative can rank the same state differently. The framework requires the
analyst to preserve this plurality until an independent argument selects or
orders comparisons. This is the operational form of the benchmark problem that
Section 4.2 attributed to the transactional tradition.

9.3 Captured Continuation and the Limits of Output Measures

This subsection establishes that equal output is compatible with material
capability disadvantage.

Let $\Pp_Y(G;m,h)=\Pp_Y(G;b,h)$, while $q_Y^\Theta(m)$ records transferred
direction and blocked exit. A repertoire-only or productivity-only measure
reports equality; the enriched profile can report a material disadvantage.
This result depends on the prior legitimacy and validation of the added
dimensions. It prevents output growth from deciding the case while also
preventing autonomy language from entering without measurement.

The case connects directly to Section 5.2. If autonomy is relationally
constituted, an arrangement that reallocates direction and forecloses exit
alters a capability coordinate even when it leaves productive output intact.
The formal apparatus registers this only because $\Theta$ was declared in
advance to include those coordinates, and a case that becomes positive only
through an opportunistically chosen coordinate remains unsupported.

9.4 Mechanism Families and Their Diagnostic Neutrality

This subsection records that mechanism type does not determine status.

The research program distinguishes actions involving the same generativity,
its application, its realization conditions, an apparently unrelated action,
and cross-generativity effects. These categories form a useful stress test but
no hierarchy. Any can produce harm; any becomes a candidate here only through
standing, an admissible comparison, an actor-linked relational mechanism,
material disadvantage, and benefit-through. Interruption, slowing, redirection,
condition control, dependency, and capture remain mechanism families.

This is the formal counterpart of the correction recorded in Section 6.1.
Condition control is one mechanism among several, and the framework treats it
with no special weight.

10. Empirical Protocol and Normative Interfaces

This section translates the formal dependencies into a research protocol. It
preserves the distinction between a proposed estimand and an identified effect,
and it states how a returned profile combines with the evaluations the
diagnostic deliberately excludes.

10.1 Case-Study Protocol

This subsection orders the measurement, comparison, pathway, and interpretation
tasks required for any applied use.

A case study should first preregister or otherwise disclose the affected actor,
domain, history, horizon, trajectory space, resolution, probability threshold,
viability rule, capability dimensions, materiality rule, benefit coordinate,
and normative specification. These choices determine the object. They cannot be
selected retrospectively to force a preferred classification.

Second, the study should construct $\B^F(m)$ and apply $\Lambda$ separately.
Technical feasibility needs evidence about resources, transition costs,
institutions, and behavior. Admissibility needs an independent argument about
rights, standing, procedure, or another declared standard. At least two
plausible comparisons should be retained when available; admitted, excluded,
and unresolved rows should all appear in the full profile.

Third, the study should estimate or bound $\K_Y^\Theta(r)$. Finite observed
trajectories rarely reveal a reachable repertoire. Interventions, natural
experiments, policy variation, longitudinal histories, simulation, and
qualitative process evidence may constrain the model, while exact latent
reconstruction remains a separate identification problem. Equal outputs across
arrangements do not establish equal capability.

Fourth, the study should establish or refute an actor-linked relation to the
same affected capability state and then type its benefit-through evidence. A
causal witness requires a concrete treatment or arrangement, the exact affected
state as mediator, a benefit outcome, an interference scope, identifying
assumptions, and sensitivity analysis. A constitutive witness requires
documentary or institutional evidence for the operative rule and its
beneficiary. The mediation literature warns that definition and identification
are separate (Imai et al., 2010).

Fifth, the study should return $1$, $0$, or $?$ for each baseline and preserve
the witness type. The result should then be combined with separate evaluations
of consent, domination, distributive effects, procedural legitimacy,
intentionality, responsibility, and remedies. A candidate edge diagnosis may
be embedded in a system that compensates, propagates, or reverses its effects;
those network and regime questions exceed the dyadic scope of this paper.

11. Open Questions Across Registers

This section collects the questions the paper leaves unresolved, organized by
the register in which each would be answered. Its objective is to separate
questions that further conceptual or formal work could settle from questions
requiring empirical evidence or independent normative argument, and to mark
which of them the paper’s own theses have made pressing.

11.1 Conceptual and Formal Questions

Develop transparent procedures for constructing technically feasible and
independently admissible baseline families without circularity, opportunistic
selection, or suppression of contested alternatives. The two-stage construction
separates feasibility from admissibility; it supplies no procedure for either.

Determine which restricted model classes, interventions, and observation
regimes can identify or bound reachable viable repertoires and enriched agency
dimensions from finite histories. Observed output underdetermines every
component of the capability object.

The enriched profile admits autonomy, direction, exit, attribution, continuity,
and recognition as coordinates. These can overlap and can conflict. Which
independence conditions must hold before a coordinate is admitted, and how are
dependencies between the repertoire component and the agency coordinates to be
declared?

Materiality is a declared threshold rule over a partial order. Which rules avoid
covertly reintroducing a scalar aggregation, and does any defensible rule
preserve genuine incomparability rather than resolving it by convention?

11.2 Normative Questions

Standing is constituted by sustained participation together with reliance. What
threshold on each is defensible, and does the account produce acceptable results
for a participant with deep reliance and negligible participation, or the
reverse?

A positive finding shifts a burden onto the beneficiary. What discharges it?
The paper names admissibility argument, available alternatives, procedure, and
distribution of gains as candidates, and supplies no account of their
sufficiency.

The diagnostic is deliberately weaker than a wrongfulness verdict. Under which
further conditions does a positive finding become sufficient for
all-things-considered wrongfulness, and is that question answerable in general
or only per case?

Extend the dyadic diagnostic to collective carriers, distributed
beneficiaries, interference, feedback, and changing actor positions while
preserving explicit standing and benefit paths.

11.3 Political-Economic Questions

The mixed-scarcity taxonomy distinguishes rival, congestible, excludable, and
positional realization conditions. Positional scarcity cannot be relieved by
capacity expansion, since the good consists in a ranking. What follows for
remedy when the operative scarcity is positional?

Concentration of realization conditions confers a capacity to constrain. What
measure of that concentration is appropriate, given that the relevant
conditions are heterogeneous and that control is exercised through terms rather
than through holdings alone?

11.4 Empirical Questions

No empirical case supports any claim in this paper. The priority questions are
whether any domain permits construction of an admissible comparison family that
survives adversarial scrutiny; whether a constitutive witness can be documented
in a setting where the governing rules are contested; whether the enriched
coordinates admit measurement that participants would recognize; and whether a
determinate negative has ever been produced. A framework that never returns a
negative finding in practice would be unfalsifiable regardless of its formal
capacity to do so, and Section 13 records this as a defeat
condition.

12. Limits and Revisable Research Program

This section consolidates the paper’s contribution, states its defeat
conditions, and assigns the questions it does not address.

12.1 Contribution and Its Limits

This subsection consolidates the paper’s strongest defensible contribution
across its four registers.

The mathematical content is modest. Definition ? constructs
a model-relative opportunity set; Proposition ? proves
inclusion under strong preservation assumptions; Proposition
? separates latent organization from effective capability;
and Propositions ? and ? establish two
countermodel separations. None proves that generativity exploitation occurs in
the world or that the diagnostic captures every accepted use of
exploitation.

The analytical contribution is the problem statement of Section 3: the
separation of non-rival generative forms from the mixed scarcity of their
realization conditions, and the resulting asymmetry between retention and
foreclosure. The reviewed traditions of Section 4 supply antecedents,
constraints, and rivals for this problem without resolving it.

The philosophical contribution is the four-thesis argument of Section 5. It
establishes where the loss in the motivating configurations is representable
(Thesis ?), why generative capability warrants separate
treatment through its compounding branch structure
(Thesis ?), how standing can be differentiated under
near-universal contribution (Thesis ?), why the object of the
relation is relational and temporal (Thesis ?), and what
normative consequence a positive diagnosis carries
(Thesis ?). Theses ? and ? are the
load-bearing results: the first dissolves the universal-contribution paradox
that would otherwise defeat the proposal, and the second supplies the reason the
diagnostic is worth having. These are arguments offered for assessment, and none
is established by the cited literature.

The political-economic contribution is the mechanism vocabulary of Section 6,
the actor separation of Table 2, and the argued refusal in
Thesis ? to convert concentration into diagnosis. The
reviewed literature establishes that control through position, rent, and
distributed uncompensated contribution are real and analytically recognized
patterns. It establishes nothing about the prevalence of the relation this
paper defines.

The conceptual contribution is a dependency chain:

$$G+r
\longrightarrow
(\A,\mathbb P,\V)
\longrightarrow
\Pp
\longrightarrow
\K^\Theta
\longrightarrow
\B^F
\longrightarrow
\Lambda
\longrightarrow
S_Y
\longrightarrow
M_{X\to Y}
\longrightarrow
D
\longrightarrow
w_{\BT}
\longrightarrow
g_b
\longrightarrow
\Pi.$$

Each arrow names an obligation rather than a discovered law. The chain
preserves the proposal’s central conceptual corrections: non-rivality removes a
depletion requirement; retained generativity does not secure effective
capability; interruption and condition control are mechanisms; capability loss
and coincident gain are insufficient; and intention can vary independently of
the proposed structural diagnosis.

12.2 Defeat Conditions and Deferred Questions

This subsection states the findings that would narrow or defeat the proposal
and assigns the questions this paper does not address.

The formal framework should be revised or abandoned if its capability dimensions
cannot be defined independently of the desired conclusion, if admissible
comparisons remain incurably circular, if benefit-through cannot be separated
from ordinary competitive interdependence, or if its classifications add no
discriminating content beyond established accounts of exploitation, harm,
domination, and capability deprivation. Section 4 establishes that several such
accounts exist and disagree; it does not establish that the present conjunction
is absent from them. A systematic originality review remains outstanding, and
the verified corpus establishes antecedents and disagreement rather than
novelty.

The philosophical theses carry separate and more consequential defeat
conditions, stated in Section 5.9 and repeated here as commitments. Defeating
Thesis ? removes the motivation for a capability object and
returns the analysis to possession or output. Defeating
Thesis ? reduces the paper to a general capability-and-benefit
framework with no distinctive subject matter, and the appropriate response is to
retitle and rescope it accordingly. Defeating Thesis ? reopens
the universal-contribution paradox; if no differential standing condition
survives, the proposal should be withdrawn, since a diagnosis that applies to
everyone or to no one has no use. Defeating Thesis ? leaves
the formalism intact and unmotivated, requiring a different justification for the
trajectory-set construction. Defeating Thesis ? makes the
diagnostic idle, and the honest response is to present it as a descriptive
taxonomy with no normative consequence. The author regards
Theses ? and ? as the most vulnerable and would
welcome direct challenge to either.

The political-economic register should be narrowed if the mechanisms of
Section 6 prove to be adequately handled by existing accounts of market power,
foreclosure, and rent without any capability construction.

Several subjects are deliberately deferred. Observation and partial
identification require a separate formal paper. Propagation, cascades, cycles,
buffers, and recovery require network and dynamic models. Commons standing,
ownership of realization conditions, the distribution and accumulation of
expansion infrastructure, AI-specific applications, legal interests, and
remedies require their own evidence and normative arguments. A field-theoretic
register would become relevant only if it supplies explanatory or inferential
gain beyond the set-valued, causal, and institutional models used here.

The resulting position is conservative. Generativity exploitation is retained
as a promising research concept when a participant’s material capability
disadvantage is linked to another party’s advantage through a specified path
relative to independently admissible comparisons. The framework’s most useful
outcome may often be an explicit unresolved profile. Such an outcome records
where standing, comparison, measurement, or causal evidence is missing and
thereby turns a broad accusation into a revisable research program.

References

Elizabeth S. Anderson. What is the point of equality? Ethics, 109(2):287–337, 1999.

Joel Anderson and Axel Honneth. Autonomy, vulnerability, recognition, and justice. In John Christman and Joel Anderson, editors, Autonomy and the Challenges to Liberalism: New Essays, pages 127–149. Cambridge University Press, Cambridge, 2005.

Kean Birch. Technoscience rent: Toward a theory of rentiership for technoscientific capitalism. Science, Technology, & Human Values, 45(1):3–33, 2020.

G. A. Cohen. The labor theory of value and the concept of exploitation. Philosophy & Public Affairs, 8(4):338–360, 1979.

Nick Couldry and Ulises A. Mejias. The Costs of Connection: How Data Is Colonizing Human Life and Appropriating It for Capitalism. Culture and Economic Life. Stanford University Press, Stanford, CA, 2019.

Miranda Fricker. Epistemic Injustice: Power and the Ethics of Knowing. Oxford University Press, Oxford, 2007.

Charlotte Hess and Elinor Ostrom, editors. Understanding Knowledge as a Commons: From Theory to Practice. MIT Press, Cambridge, MA, 2007.

Kosuke Imai, Luke Keele, and Dustin Tingley. A general approach to causal mediation analysis. Psychological Methods, 15(4):309–334, 2010.

Lina M. Khan. Amazon’s antitrust paradox. The Yale Law Journal, 126:710–805, 2017.

Catriona Mackenzie. The importance of relational autonomy and capabilities for an ethics of vulnerability. In Catriona Mackenzie, Wendy Rogers, and Susan Dodds, editors, Vulnerability: New Essays in Ethics and Feminist Philosophy, pages 33–59. Oxford University Press, New York, 2013.

Karl Marx. Capital: A Critique of Political Economy, volume 1. Penguin Books, 1992. Translated by Ben Fowkes; original German edition published in 1867.

Martha C. Nussbaum. Women and Human Development: The Capabilities Approach. The Seeley Lectures. Cambridge University Press, Cambridge, 2000.

Elinor Ostrom. Governing the Commons: The Evolution of Institutions for Collective Action. Canto Classics. Cambridge University Press, Cambridge, 2015. Canto Classics reissue; originally published in 1990.

Philip Pettit. Republicanism: A Theory of Freedom and Government. Oxford Political Theory. Oxford University Press, Oxford, 1997.

Ingrid Robeyns. The capability approach: A theoretical survey. Journal of Human Development, 6(1):93–117, 2005.

John E. Roemer. A General Theory of Exploitation and Class. Harvard University Press, Cambridge, MA, 1982.

Paul M. Romer. Endogenous technological change. Journal of Political Economy, 98(5, Part 2):S71–S102, 1990.

Ruth J. Sample. Exploitation: What It Is and Why It’s Wrong. Rowman & Littlefield, Lanham, MD, 2003.

Amartya Sen. Capability and well-being. In Martha C. Nussbaum and Amartya Sen, editors, The Quality of Life, pages 30–53. Oxford University Press, Oxford, 1993.

Nick Srnicek. Platform Capitalism. Theory Redux. Polity Press, Cambridge, 2017.

Bernard Stiegler. For a New Critique of Political Economy. Polity Press, Cambridge, 2010. Translated by Daniel Ross.

Tiziana Terranova. Free labor: Producing culture for the digital economy. Social Text, 18(2 (63)):33–58, 2000.

Nicholas Vrousalis. Exploitation, vulnerability, and social domination. Philosophy & Public Affairs, 41(2):131–157, 2013.

Alan Wertheimer. Exploitation. Princeton University Press, Princeton, NJ, 1996.

Matt Zwolinski. Structural exploitation. Social Philosophy and Policy, 29(1):154–179, 2012.