Cycles of Generativity Exploitation - Stability, Distribution, and Procedural Evaluation
Abstract
When exploitative relations close into a circuit, every participant can occupy
both the affected and the beneficiary position. Aggregate capability can be
unchanged or rising, roles can be symmetric, and the structure can reproduce
itself indefinitely. This invites an inference that the paper’s source
discussion made explicitly and then withdrew: that a sufficiently reciprocal
circulation cancels the exploitative character of its constituent relations, so
that if everyone is exploited then no one is. The withdrawal was forced by a
counterexample. In a closed cycle of robberies where each participant takes and
is taken from in equal measure, aggregate holdings are unchanged and every
constituent robbery remains wrongful. This paper reconstructs that dialectic and
develops what survives it. It argues that distributive outcome, relational
procedure, and dynamical stability are three independent evaluative registers,
none of which determines another, and that stability in particular belongs to
none of them: slave systems, monopolies, and predator–prey systems are all
stable. It then dissolves the universal-participation paradox by locating an
equivocation. The paradox assumes that exploitation requires a contrast class of
persons, so that universal instantiation is self-defeating. The
diagnostic’s contrast class is a class of arrangements: each relation is
assessed against an admissible feasible alternative, not against other
participants. Universal instantiation is therefore coherent and describes a
regime in which everyone would be better off under an admissible alternative
while each still benefits through someone. Two formal results support the
reconstruction. The edge predicate contains no aggregate term, so aggregate
neutrality, symmetry, and closure are logically independent of it. And aggregate
capability can strictly increase over time while every participant remains
materially below an admissible alternative, because growth is measured against
the past and the diagnostic against a feasible counterfactual. The paper then
separates two structures that aggregate description conflates: a closed cycle in
which every participant benefits through someone, and a systemic trap in which
everyone is below a feasible alternative and no benefit-through witness exists
anywhere. These are mutually exclusive and jointly non-exhaustive, so a third
mixed class exists and is probably where most real configurations fall. They
require different remedies: procedural intervention at edges for the first, and
alteration of the feasible set for the second. The paper establishes no
prevalence, classifies no observed regime, and supplies no aggregation rule for
combining edge-level findings.
Keywords: generativity exploitation; cycles; procedural
Document Status, Preparation, and Reuse
This unnumbered section records the manuscript’s status, research genealogy,
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
normative arguments are offered for assessment. Objections, counterexamples,
corrections, alternative formulations, and relevant literature are welcome at
mailto:huangwanhong@serendip.ngohuangwanhong@serendip.ngo.
Research genealogy.
The dialectic reconstructed in Section 3 is researcher-origin
throughout its decisive moves. The researcher closed the propagation chain into
a circuit, asked whether a stable cycle is necessarily objectionable, and posed
the universal-participation paradox in the form everyone is exploited,
therefore no one is. When an over-permissive resolution was proposed in
response, the researcher supplied the closed-cycle robbery counterexample
together with the question of procedural justice, which forced its withdrawal.
The stability taxonomy, the three-register separation, the paradox-dissolution
argument, the formal non-cancellation results, the growth construction, and the
cycle–trap separation predicate are later agent-added reconstructions. Their
inclusion records development of the proposal and attributes these
constructions neither to the initiating notes nor to the cited literature.
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
When exploitative relations form a chain, the structure has ends. Someone at the
head is only a beneficiary and someone at the tail is only affected, and the
asymmetry makes the configuration easy to assess. Closing the chain removes both
ends. Every participant becomes both an affected party and a beneficiary,
aggregate capability may be unchanged or rising, and the structure may reproduce
itself indefinitely. What such a configuration warrants is the question this
paper addresses.
The question is not artificial. It arose in the source discussion when a
propagation chain was closed into a circuit, and it immediately produced a
paradox stated there directly: if the cycle is stable and everyone in it is
exploited, someone will ask what is wrong with it, since everyone being
exploited seems to leave no one exploited. The response initially proposed was
that universal local exploitation may cease to be exploitation when relations
close into sufficiently reciprocal circulation. That response was withdrawn
under a counterexample. If reciprocal circulation cancelled, then a cycle in
which each participant robs and is robbed of equal value would contain no
robbery. It plainly does, and the accompanying question named what the response
had ignored: procedural justice.
This paper reconstructs that dialectic and develops what survives it. The
retraction is preserved rather than replaced by the corrected position, because
the inference that failed is natural and its failure is the result.
Three insights organize the argument. The first is a separation. Distributive
outcome, relational procedure, and dynamical stability are independent
evaluative registers, and none determines another. The robbery cycle scores
neutrally on outcome and badly on procedure; a procedurally impeccable
arrangement can progressively exhaust a participant; and stability belongs to
neither register, since slave systems, monopolies, and predator–prey systems
are all stable and differ in everything that matters evaluatively.
The second dissolves the paradox. The inference from everyone is
exploited to no one is exploited is valid only if exploitation requires
a contrast class of persons. The diagnostic’s contrast class is a class of
arrangements: each relation is assessed against an admissible feasible
alternative rather than against other participants. A predicate whose comparison
class is arrangements can hold of every participant without becoming empty.
Universal instantiation is therefore coherent, and what it describes is a regime
in which everyone would be better off under an admissible alternative while each
still benefits through someone. That is a determinate and unattractive condition
rather than a contradiction. A major reconstruction in the exploitation
literature defines the relation through a withdrawal comparison against a
counterfactual distribution, which shows that arrangement-relative comparison is
available and not invented here to escape a difficulty.
The third explains why the aggregate facts cancel nothing. The edge predicate
contains no aggregate term: it is a conjunction over standing, comparison
admissibility, an actor-linked mechanism, a material capability disadvantage,
and a benefit-through witness, none of which references the sum of anything, the
symmetry of roles, or the persistence of the configuration. Aggregate neutrality
and closure therefore do not fail to cancel the predicate because they are
outweighed; they are logically independent of it, and the inference attempted
was never available. The same structure explains why growth is no defence:
growth compares a configuration with its own past, while the diagnostic compares
each participant with a feasible alternative, and the paper exhibits a
configuration in which aggregate capability strictly rises at every period while
every participant remains materially below an admissible alternative.
The paper then separates two structures that aggregate description conflates. In
circulating appropriation every participant is affected on some edge carrying a
supported benefit-through witness. In a systemic generative trap everyone is
below an admissible alternative and no witness exists anywhere. These are
mutually exclusive and jointly non-exhaustive, so a third mixed class exists and
is probably where most real configurations fall. They require different
remedies: procedural intervention at edges for the first, alteration of the
feasible set for the second. Treating a trap as circulating appropriation
produces a search for an appropriator who does not exist; treating circulating
appropriation as a trap erases the benefit-through guardrail and dissolves
responsibility that is locatable.
The paper’s contribution is therefore a set of structural separations rather
than a verdict procedure. It establishes no prevalence, classifies no observed
regime, supplies no rule for aggregating findings across levels, and recommends
no institution.
The argument proceeds in nine further sections. Section 3 presents
four structures and reconstructs the dialectic. Section 4
analyzes closure, feedback, and stability, and states the level decomposition.
Section 5 reviews the traditions the problem engages.
Section 6 argues the four theses.
Section 7 develops the political economy of
circulating appropriation. Sections 8 and 9
supply the formal results. Section 10 states countermodels,
Section 11 collects open questions, and Section 12
states limits and defeat conditions.
The philosophical and formal parts are formally independent, since no
proposition cites a thesis, and justificatorily dependent, since the
propositions are elementary and their significance rests on the arguments that
make them worth stating. Section 6.5 records what fails if each
thesis is rejected.
2. Cycle Scope and the Source Correction
This section states the paper’s target, reconstructs the dialectic that
generates it, and fixes the position and order of inquiry.
2.1 Four Structures
This subsection presents the configurations the paper must distinguish. Each is
schematic and none is a description of an observed regime.
Closed appropriative circuit.
Four participants stand in a chain of relations, each of which satisfies a
diagnostic for generativity exploitation, and the chain closes: the last
participant stands in such a relation to the first. Every participant is both
an affected party and a beneficiary.
Reciprocal generative dependence.
Four participants mutually constrain one another under terms each accepts, and
the constraints jointly expand what each can reach. The topology is the same
circuit. The relations are not exploitative, and were not from the outset.
Systemic generative trap.
Every participant’s effective capability is materially below what an admissible
feasible alternative would supply, the configuration reproduces itself, and no
participant’s advantage runs through any other’s diminished position. There is
no beneficiary anywhere.
Growing circuit.
A closed appropriative circuit in which aggregate capability rises period after
period, so that every participant is better off than it was before, while each
remains below an admissible alternative.
The first and second share a topology and differ in diagnosis. The first and
third share an aggregate description and differ in structure. The fourth blocks
the most common defence. Distinguishing them is the paper’s work.
2.2 The Dialectic
This subsection reconstructs the sequence of moves that generates the paper’s
central problem. The reconstruction preserves a retracted position, because the
retraction is the result.
The starting point is a chain of relations in which an upstream relation raises
the probability of a downstream one. The researcher closed the chain:
$$X\leadsto Y\leadsto Z\leadsto W\leadsto X.$$
Closure changes the object. A chain has a first party who is only a beneficiary
and a last who is only affected. A circuit has neither.
The researcher then asked whether a stable cycle is necessarily objectionable.
The initial answer was that stability is a dynamical property carrying no
normative content, since slave systems, monopolies, and predator–prey systems
are all stable, and that assessment requires separate questions about viability,
distribution, and process.
The researcher then posed the paradox directly: if the cycle is stable and
everyone in it is exploited, someone will ask what is wrong with it, since
everyone being exploited seems to leave no one exploited.
The response proposed was over-permissive. It held that universal local
exploitation may cease to be exploitation when the relations close into
sufficiently reciprocal circulation, qualified only by asymmetry or by the cycle
trapping everyone below feasible possibilities.
The researcher rejected this with a counterexample and a question. If reciprocal
circulation cancels, then a cycle in which each participant robs and is robbed
of equal value would contain no robbery. It plainly does. The accompanying
question named what the over-permissive answer had ignored: procedural justice.
The retraction is the paper’s foundation. What the counterexample shows is that
an evaluation running only through distributive outcome cannot reach the
property that makes each relation objectionable, because that property concerns
how the relation was transacted.
2.3 Position and Source Use
This subsection states what the paper claims and the constraint governing its
citations.
Closure, dynamical stability, symmetric role occupancy, aggregate neutrality,
and aggregate growth are individually and jointly insufficient to cancel the
diagnostic status of a constituent relation. A cyclic topology is
diagnostically inert on its own, and a regime in which everyone is below an
admissible alternative requires a further distinction between circulating
appropriation and a beneficiary-free trap.
Where a source is used, the text states the specific constraint it supplies and
the inference it withholds. No source in the verified corpus treats exploitative
cycles, states the paradox, or supplies its dissolution, and no source is cited
in support of a proposition or a countermodel. The corpus supplies the traditions
the problem engages and the constraints they impose.
3. Closure, Feedback, and Stability Classes
This section performs the problem analysis. Its objective is to separate three
things that the word cycle runs together, and to establish that none of
them carries an evaluative verdict.
3.1 Closure Is Not a Cycle
This subsection distinguishes a topological from a dynamical object.
A directed circuit is a topological fact about a relation graph: following
edges from a participant returns to that participant. It is compatible with the
edges having occurred once, in sequence, with no influence of the later on the
earlier.
A dynamic cycle requires more: a later relation must affect an earlier one, so
that the structure reproduces or modifies itself. Without feedback to an earlier
edge, a closed circuit is a path that happens to return, and its evaluation is
the evaluation of a path.
The distinction matters because the paradox of Section 3 draws its
force from reproduction. A configuration that occurred once and stopped invites
no question about whether its persistence is acceptable.
3.2 Stability Classes
This subsection records that stability admits several distinct meanings, each
with different implications and none with evaluative content.
A reproducing configuration may be stationary, with edge intensities
approximately constant; damped, with intensities declining toward
absence; amplifying, with intensities rising; periodic, cycling
through states; intermittent, with irregular activation;
metastable, persisting under small perturbation and collapsing under
large; or absorbing, having reached a state from which no exit is
available.
These classes are not interchangeable for any purpose. A metastable
configuration and an absorbing one differ precisely in whether exit exists,
which bears directly on the participants’ position, while both may be described
as stable in ordinary usage. The network literature makes the same point about
connectivity: a structure can be stabilizing under small shocks and
destabilizing under large ones (Acemoglu et al., 2015), and integration has
non-monotone effects on the extent of a cascade (Elliott et al., 2014).
3.3 Stability Carries No Verdict
This subsection states the first thing the paper needs and the reason for it.
That a configuration reproduces itself says nothing about whether it should. A
configuration can be stable because its participants are satisfied, because
their alternatives have been foreclosed, or because the cost of exit exceeds
the cost of remaining. Slave systems, monopolies, and predator–prey systems are
all stable, and they differ in every respect that matters evaluatively.
The observation generalizes to the aggregate quantities that stability analyses
produce. A subcritical cascade parameter reports that expected propagation is
finite; it reports nothing about who bears the residual or how each relation was
transacted. Threshold models establish that aggregate outcomes depend
sensitively on the distribution of individual response rules
(Granovetter, 1978), which is a further reason that an aggregate
summary cannot substitute for the distribution it summarizes.
3.4 Four Levels of Analysis
This subsection states the structural decomposition the cycle problem requires.
Its objective is to identify where each of the questions raised so far is
answered, and to record that answers at one level do not transfer to another.
The capability object at issue is an opportunity set rather than an achieved
output, following the distinction between achieved functionings and the
alternatives a participant could feasibly achieve and choose
(Sen, 1993). Its specification requires declared dimensions,
weights, and an admissible treatment of incompleteness
(Robeyns, 2005). Both features matter below: the disadvantage
comparison is between opportunity sets under two arrangements, and the
comparison may be partial.
Table 1 records four levels. The decomposition is not a hierarchy
in which higher levels aggregate lower ones. Each level has its own objects, its
own evidence, and its own characteristic error.
| >p0.17
p0.25
X
| Level | Object | Characteristic error |
|---|---|---|
| Edge | One relation: standing, mechanism, disadvantage, benefit-through | Treating a supported edge as a verdict on the configuration |
| Path | Whether one relation causally changes another | Reading temporal succession or closure as causal reproduction |
| Actor | Cumulative position across occupancy: share lost, frequency of each role, exit availability | Reading symmetric role occupancy as equal position |
| Regime | Distribution, stability class, feasible alternatives, presence or absence of any beneficiary | Reading aggregate neutrality or growth as cancellation |
Table. Four Levels of Analysis for a Closed Configuration. Answers at one
level do not transfer to another. The final column records the error produced by
attempting the transfer.
The actor level is the one the cycle problem makes newly visible, and it has no
counterpart in the open-path case. In a chain, a participant occupies one role.
In a circuit, a participant occupies both, and the question of how its
occupancies compare is a question about that participant rather than about any
edge or about the regime. Section 7 develops it, and
Section 11 records that its proper formalization is unresolved.
The regime level is where the over-permissive resolution operated, and
Section 8 proves that its inference to the edge level is
unavailable.
4. Reciprocity, Procedure, and Trap Traditions
This section locates the paper among the traditions its problem engages. Its
objective is bounded positioning: each supplies a constraint or a contrast, and
none supplies a treatment of exploitative cycles.
4.1 Exploitation Traditions Under Closure
This subsection asks what each account of exploitation says when the relation
closes.
Marx’s account locates exploitation in the appropriation of surplus labor
through ownership of productive means (Marx, 1992). Its structure is
asymmetric by construction: the classes are distinct. Closure has no natural
place in it, since a capitalist who is also a worker in the same relation is not
a case the account addresses.
Roemer’s reconstruction is more instructive. Exploitation there is defined
through a withdrawal comparison: a coalition is exploited when it would be
better off withdrawing with its per-capita share of productive assets
(Roemer, 1982). This is a comparison against a
counterfactual arrangement rather than against another party, and it is
therefore compatible in principle with every coalition being exploited relative
to some alternative. Section 6 builds on exactly this feature.
Cohen’s argument that the exploitation charge survives removal of the labor
theory of value (Cohen, 1979) matters here for the same reason: it
detaches the charge from a conserved quantity, and a conserved quantity is what
would make circulation cancel.
Wertheimer establishes that mutually beneficial transactions can still be
exploitative (Wertheimer, 1996). A cycle is a limiting case of
mutual benefit, and the account gives no reason to treat it differently.
Sample’s degradation-centered account (Sample, 2003) and
Vrousalis’s domination-based account (Vrousalis, 2013) both
locate the wrong in features of the relation rather than in the distribution it
produces, so both are prima facie unaffected by closure. Zwolinski’s constraint
against inferring a transaction’s status from background structure
(Zwolinski, 2012) applies in the opposite direction here: it warns
against inferring edge status from regime-level facts, which is what the
over-permissive resolution attempted.
Wollner’s analysis of anonymous exploitation, which is non-individual,
non-agential, and structural (Wollner, 2019), is the closest
verified treatment of the trap branch, and Section 9
distinguishes it from circulating appropriation.
4.2 Procedural Justice
This subsection supplies the register the counterexample invokes.
Rawls distinguishes perfect, imperfect, and pure procedural justice
(Rawls, 1999). In the first two there is an independent criterion for a
just outcome and the procedure is assessed by its tendency to produce it. In
pure procedural justice there is no independent criterion: the outcome is just
because the procedure was correctly followed. The distinction matters here
because it establishes that procedure and outcome are separately assessable, and
that in at least one important case the direction of determination runs from
procedure to outcome rather than the reverse.
Tyler’s research on procedural justice and legitimacy establishes empirically
that judgments about the fairness of procedures influence acceptance and
cooperation substantially, and can do so independently of the favorability of
outcomes (Tyler, 2003). The paper uses this carefully: it concerns
perceived fairness and its behavioral consequences, which is evidence that the
registers come apart in judgment, not a normative argument that they come apart
in fact. The normative argument is given in Section 6.
Anderson’s relational-equality argument holds 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). Pettit’s conception of freedom as non-domination
supplies a further relational property that distributive facts do not settle
(Pettit, 1997). Both support the claim that a distributive
summary omits something, without establishing what the omitted property is in
this framework.
4.3 Reciprocity
This subsection blocks an inference from reciprocity to legitimacy.
Gouldner analyzes the norm of reciprocity as a generalized moral norm requiring
that people help those who have helped them, and identifies its function as a
starting mechanism that initiates and stabilizes relations before more specific
obligations are established (Gouldner, 1960). He also distinguishes
homeomorphic reciprocity, in which returns are equivalent in form, from
heteromorphic reciprocity, in which they differ in form while being judged
equal in value.
Two consequences follow for this paper. Reciprocity is a normatively loaded
notion doing stabilizing work, so calling a configuration reciprocal does not
describe it neutrally. And the homeomorphic case is where the over-permissive
resolution has most intuitive appeal, since each participant appears to return
what it received. The robbery cycle is homeomorphic reciprocity, which is why it
is the decisive counterexample.
4.4 Traps and Structural Injustice
This subsection supplies the antecedent for the beneficiary-free branch.
The poverty-traps literature analyzes self-reinforcing configurations in which
individuals, groups, or whole economies remain at low levels through mechanisms
that reproduce the condition, drawing on economics, economic history, and
sociology (Bowles et al., 2006). The relevant structural feature is that such
configurations can persist without any party appropriating from the trapped, and
that the reproduction mechanism is the object of explanation.
Young’s social connection model characterizes responsibility arising from
participation in structural processes producing injustice, and distinguishes it
from liability that isolates a perpetrator (Young, 2006). This
supplies the form responsibility takes when no beneficiary edge exists, which
Section 9 identifies as the trap branch’s characteristic
situation.
4.5 Comparative Summary
Table 2 records what each tradition supplies and what it
leaves open for the cycle problem.
| >p0.20
p0.26
X
| Tradition | What it supplies | Constraint imposed here |
|---|---|---|
| Surplus appropriation (Marx, 1992) | Asymmetric class structure | Offers no account of a participant occupying both positions |
| Withdrawal comparison (Roemer, 1982) | Comparison against a counterfactual arrangement | Universal instantiation is coherent when the contrast is an arrangement |
| Value-theory independence (Cohen, 1979) | Charge detached from a conserved quantity | Circulation cannot cancel what is not conserved |
| Mutual benefit (Wertheimer, 1996) | Beneficial transactions can be exploitative | A cycle is a limiting case and gets no exemption |
| Degradation and domination (Sample, 2003; Vrousalis, 2013) | Wrong located in the relation | Closure does not touch relational properties |
| Structural limitation (Zwolinski, 2012) | Background does not settle a transaction | Regime-level facts do not settle edge status |
| Anonymous exploitation (Wollner, 2019) | Structural wrong without an agent | Requires separation from circulating appropriation |
| Procedural justice (Rawls, 1999; Tyler, 2003) | Procedure assessable independently of outcome | Distributive summary omits a register |
| Relational equality (Anderson, 1999; Pettit, 1997) | Relations as the object of justice | Aggregate neutrality is not sufficiency |
| Reciprocity (Gouldner, 1960) | Reciprocity as a loaded stabilizing norm | Calling a cycle reciprocal is not neutral description |
| Poverty traps (Bowles et al., 2006) | Self-reproducing low-level configurations | Persistence without an appropriator is a distinct object |
| Structural responsibility (Young, 2006) | Forward-looking responsibility without a perpetrator | Supplies the form responsibility takes in the trap branch |
Table. Traditions Engaged by the Cycle Problem. No tradition reviewed supplies
a treatment of closed exploitative circuits. The constraints they impose are
listed in the third column.
5. The Universal-Participation Paradox and Three Registers
This section argues the four theses on which the paper depends. Each subsection
states a thesis, argues for it, states the strongest objection known to the
author, and replies.
5.1 Register Independence
This subsection states the separation the counterexample forces.
Distributive outcome, relational procedure, and dynamical stability are three
independent evaluative registers. None determines another, and a complete
assessment of a reproducing configuration reports all three.
The argument establishes independence pairwise. Outcome does not determine
procedure: the robbery cycle is distributively neutral and procedurally
objectionable throughout. Procedure does not determine outcome: an arrangement
can be procedurally impeccable, with full consent, disclosure, and recourse at
every step, while producing a distribution in which one participant’s capability
is progressively exhausted. Stability determines neither: a configuration
persists for reasons ranging from satisfaction to foreclosure of exit, and
Section 4 records that the standard stability classes cut across
every evaluative distinction.
The traditions of Section 5 support the first two
separations from opposite directions. Rawls’s category of pure procedural
justice (Rawls, 1999) shows a case where procedure is prior to outcome,
and relational-equality accounts (Anderson, 1999) show that a
distributive summary omits properties of the relation.
The strongest objection is that three registers with no aggregation rule
produce no verdict, so the thesis purchases honesty at the price of usefulness.
The reply concedes the cost and denies that the alternative is available. An
aggregation rule that collapsed the registers would have to assign a rate of
exchange between procedural wrong and distributive gain, and the counterexample
shows what happens when that rate is implicitly set to make distribution
decisive: the robbery cycle comes out acceptable.
Section 11 records the construction of a defensible rule as an open
problem rather than pretending one exists.
5.2 Dissolution of the Universal-Participation Paradox
This subsection resolves the paradox that forced the dialectic’s central error.
The inference from everyone is exploited to no one is exploited
equivocates on the contrast class. It is valid only if exploitation requires a
contrast class of persons. The diagnostic’s contrast class is a class of
arrangements, so universal instantiation is coherent and describes a determinate
condition.
The argument identifies the equivocation precisely. The paradox reasons that
exploitation is a comparative notion, that comparatives require a contrast, and
that if the predicate holds of everyone the contrast is empty. The second premise
is correct and the third depends on what the comparison ranges over.
The diagnostic evaluates each relation against an admissible feasible
arrangement: would this participant have materially greater effective
capability under an alternative that is technically feasible and independently
admissible? Nothing in that comparison mentions other participants. A predicate
whose comparison class is arrangements can hold of every participant without
becoming empty, in the way that every person in this room is shorter than
they would be with better nutrition is not self-defeating.
Roemer’s withdrawal comparison has the same structure and the point is visible
there (Roemer, 1982): a coalition is assessed against what it
would obtain by withdrawing with a per-capita share, and it is not incoherent
for every coalition to be worse off than under some alternative distribution of
assets.
The condition universal instantiation describes is therefore determinate. Every
participant would be better off under an admissible alternative, and each
participant’s advantage runs through some other participant’s diminished
position. That is a coherent and unattractive state of affairs, not a
contradiction.
The strongest objection is that this drains the term of comparative force,
making exploitation a synonym for regime that could be improved.
The reply is that the benefit-through condition prevents the collapse. Universal
instantiation requires that each participant benefit through another’s
diminished position, which is a substantive structural requirement that most bad
regimes fail. Section 9 exhibits the class that fails it: a
trap in which everyone is below an admissible alternative and no one benefits
through anyone. Universal exploitation and universal deprivation are different
conditions, and the framework distinguishes them.
A second objection is that an admissible alternative may not exist for every
participant simultaneously, so universal instantiation may be infeasible even if
coherent. The reply concedes this and treats it as a substantive constraint
rather than an objection: the companion diagnostic requires comparison
arrangements to be jointly realizable, so universal instantiation is available
only where the alternatives can be jointly realized. Where they cannot, the
correct finding is that the comparison family is incoherent, which is an
unresolved status rather than a negative one.
5.3 Non-Cancellation
This subsection states what the robbery counterexample establishes.
Closure, symmetric role occupancy, aggregate neutrality, and aggregate growth
leave the diagnostic status of each constituent relation unchanged.
The argument is the researcher’s counterexample together with a structural
diagnosis of why it works. Consider a closed set of participants in which each
takes a fixed value from its successor and has the same value taken by its
predecessor. Aggregate holdings are unchanged, every participant occupies both
roles, and the configuration can repeat indefinitely. Every constituent taking
remains wrongful.
The structural diagnosis is given formally in Section 8 and
stated here informally: the edge predicate contains no aggregate term. It is a
conjunction over standing, comparison admissibility, an actor-linked mechanism,
a material capability disadvantage, and a benefit-through witness. None of these
clauses references the sum of anything, the symmetry of roles, or the persistence
of the configuration. Facts about aggregates are therefore not merely
insufficient to cancel the predicate; they are logically independent of it. The
inference the over-permissive resolution attempted was never available.
The strongest objection is that robbery is a poor analogue, since it involves
discrete wrongful acts with clear victims while exploitation on this framework
is a structural relation that may be entered without intention. The reply is
that the counterexample targets one inference and requires no closer analogy.
The inference under test moves from aggregate neutrality and role symmetry to
cancellation. Any case in which those premises hold and cancellation fails
refutes it, and the formal result shows the failure is structural rather than
a feature of the example.
5.4 Compensation Does Not Cancel
This subsection states the practical form of the preceding thesis.
Downstream compensation can restore a participant’s capability position without
cancelling the character of the upstream relation that diminished it.
The argument follows from Thesis ?. Compensation operates in
the distributive register: it alters the outcome. The property that made the
upstream relation objectionable is a property of how the relation was
transacted, which compensation does not reach. A configuration in which each
participant is diminished and then made whole is distributively equivalent to
one in which nothing happened, and procedurally quite different.
The thesis matters because compensation is the natural remedy proposal for a
cycle, and because it is genuinely valuable. Its limits should be stated
precisely rather than by dismissal: compensation is an adequate response to the
distributive component and no response at all to the procedural one. Where the
procedural component is the objection, compensation changes the subject.
The strongest objection is that this makes remedy impossible, since the past
transaction cannot be altered. The reply is that procedural remedies exist and
differ from compensation: altering the terms on which future relations are
transacted, supplying recourse, and removing the mechanism are all available and
all operate in the register where the objection lies.
5.5 Dependence of the Framework on the Four Theses
This subsection records what fails if each thesis is rejected.
Rejecting Thesis ? permits a distributive summary to settle the
assessment, which the robbery counterexample refutes, and licenses reading
stability as a normative result. Rejecting Thesis ? leaves the
universal-participation paradox standing, which either voids the diagnostic
under universal instantiation or forces the over-permissive resolution the
source discussion withdrew. Rejecting Thesis ? reinstates
cancellation by circulation. Rejecting Thesis ? permits
compensation to close cases whose objection it does not reach.
Theses ? and ? are load-bearing.
Thesis ? is the paper’s principal contribution and its most
contestable claim, since it rests on a diagnosis of an equivocation that a
reader may not accept. Every proposition below survives the rejection of all
four.
6. The Political Economy of Circulating Appropriation
This section develops the political-economic content of the cycle problem. Its
objective is to identify what a circuit changes about position, defence, and
remedy.
6.1 Symmetric Roles, Asymmetric Positions
This subsection states why role symmetry conceals difference.
That every participant occupies both positions does not entail that
participants are alike. They may differ in the magnitude appropriated, in the
share of their capability the loss represents, in the availability of exit, in
the cost of remaining, and in the duration of occupancy of each role. A circuit
in which one participant loses a small fraction and another loses most of its
remaining capability is symmetric in role and radically asymmetric in position.
The distinction between metastable and absorbing configurations from
Section 4 is the operative one here. A participant with an
available exit stands differently from one without, and the difference does not
appear in any description of the topology.
Three quantities constitute cumulative position, and none is recoverable from
the edge statuses. The share a participant loses on the edges where it is
affected, measured against its remaining capability rather than in absolute
terms, since equal absolute losses fall unequally. The frequency and
duration of occupancy in each role, since a participant affected in most
periods and beneficiary in few stands differently from one with the reverse
pattern. And the availability and cost of exit, which distinguishes
participation from confinement.
A configuration can be symmetric in role occupancy and maximally asymmetric in
all three. This is the actor level of Table 1, and it is the
level at which the intuition behind the universal-participation paradox has real
content. The paradox asks what is wrong with a configuration in which everyone
is exploited. Part of the answer given in Section 6 is that
the question misdescribes the comparison class. The remaining part is that
everyone conceals differences that matter, and that role symmetry is
precisely the description under which those differences vanish.
The framework does not currently supply a cumulative-position measure.
Section 11 records its construction as an open problem and notes
that whether it belongs to the distributive register or constitutes a fourth
register is itself unsettled.
6.2 Growth as a Defence
This subsection identifies the most common defence of a circulating
configuration and states why it fails.
The defence holds that participants are better off than they were, so the
configuration cannot be objectionable. Proposition ? shows the
inference fails, and the reason is instructive: growth is measured against the
past, and the diagnostic is measured against a feasible alternative. These are
different comparison classes, and a configuration can improve on its own history
while every participant remains below what an admissible alternative would
supply.
The defence therefore answers a question the diagnostic does not ask. Its appeal
comes from the fact that historical comparison is easy and counterfactual
comparison is hard, which is an epistemic asymmetry rather than a normative
argument.
6.3 Where Remedy Attaches
This subsection states the practical consequence of the register separation.
Remedies attach to registers. A distributive remedy transfers capability or its
conditions and addresses the outcome. A procedural remedy alters the terms on
which relations are transacted, supplies recourse, or removes a mechanism, and
addresses how the relation is conducted. A dynamical remedy alters the feedback
that reproduces the configuration and addresses its persistence.
The three are not substitutes, and the cycle case makes this visible. Compensation
leaves the mechanism in place and the configuration reproducing. Removing the
mechanism at one edge leaves the remaining edges and may simply relocate the
circuit. Breaking the feedback halts reproduction without addressing any
completed relation. A complete response specifies which register each measure
addresses and what it leaves untouched. The paper recommends no institution and
classifies no regime.
7. Cycle Objects and Non-Cancellation
This section supplies the formal core. Its objective is to state the cycle
objects precisely and to prove that the aggregate facts invoked by the
over-permissive resolution are independent of the edge predicate.
7.1 Declared Objects
Let $\Vv$ be a finite set of participants and let each ordered pair carry at most
one relation position at each time. For an edge $e$ from an affected participant
to a candidate beneficiary at time $t$, let
$$\GEX(e,t)\in{1,0,?}$$
be the edge diagnostic supplied by the companion framework: a strong three-valued
conjunction over comparison admissibility, participant standing, an actor-linked
mechanism, a material capability disadvantage, and a typed benefit-through
witness, each evaluated relative to a declared admissible comparison arrangement
$b_e$ drawn from a feasible family $\Bb_e$.
Let $\Gamma_i(r)$ denote participant $i$’s effective capability under
arrangement $r$, with the partial comparison and materiality rules of the
companion framework. The construction below uses no property of $\Gamma$ beyond
its being indexed to an arrangement.
A relation set exhibits closure when its edges form a directed circuit on
$\Vv$. It is a dynamic cycle when in addition some later edge causally
affects an earlier one, so that
$$e_{r-1}(t_{r-1})\leadsto e_0(t_r),
\qquad t_r>t_{r-1}.$$
The stability class of a dynamic cycle is its qualitative long-run behavior
under the declared update: stationary, damped, amplifying, periodic,
intermittent, metastable, or absorbing.
7.2 Aggregate Independence
This subsection proves the structural fact underlying
Thesis ?.
Let $\Phi$ be any predicate on a configuration definable from the multiset of
capability levels ${\Gamma_i(\cdot)}_{i\in\Vv}$, the role-occupancy pattern,
and the stability class. Then $\Phi$ is logically independent of
$\GEX(e,t)$ for every edge $e$: no value of $\Phi$ entails any value of
$\GEX(e,t)$, and no value of $\GEX(e,t)$ entails any value of $\Phi$.
$\GEX(e,t)$ is a conjunction whose clauses are: admissibility of $b_e$ under a
declared filter; standing of the affected participant; an actor-linked mechanism
connecting the assessed arrangement to the capability difference; a material
disadvantage comparing $\Gamma$ under the assessed arrangement with $\Gamma$
under $b_e$; and a benefit-through witness connecting the affected position to
the beneficiary’s advantage. Each clause is evaluated on the pair
(assessed arrangement, comparison arrangement) for the single edge $e$.
No clause is a function of the multiset of capability levels across
participants, since the disadvantage clause compares one participant across two
arrangements rather than participants with one another. No clause is a function
of role occupancy, since the predicate is evaluated per edge without reference
to the participants’ positions on other edges. No clause is a function of the
stability class, since the predicate is evaluated at a time without reference to
the configuration’s long-run behavior. Hence $\Phi$ and $\GEX(e,t)$ are
functions of disjoint arguments, and Section 10 exhibits
configurations realizing each combination of their values.
The proposition is elementary and that is its point. Aggregate neutrality, role
symmetry, closure, and stability do not fail to cancel the edge predicate
because the cancellation is outweighed by something. They fail because the
predicate never referred to them.
A configuration in which aggregate capability is unchanged, every participant
occupies both roles, and the structure reproduces itself is compatible with
$\GEX(e,t)=1$ for every edge.
7.3 Growth Against Counterfactual Comparison
This subsection proves that the strongest form of the aggregate defence also
fails.
There exist configurations in which aggregate capability strictly increases at
every period while every participant is materially below its admissible
comparison arrangement at every period.
Take $\Vv={1,2,3}$ arranged in a circuit, and let capability be represented on
a scalar coordinate for the construction, with materiality threshold below one
unit. Under the assessed arrangement $m$, set
$$\Gamma_i(m,t)=10+t,\qquad i\in\Vv,\ t=0,1,2,\ldots$$
For each $i$ let the admissible comparison $b_i$ be the arrangement in which the
edge into $i$ is absent and all other components are held fixed, and set
$$\Gamma_i(b_i,t)=12+t .$$
Aggregate capability under $m$ is $3(10+t)$, which strictly increases in $t$.
For every $i$ and every $t$, $\Gamma_i(b_i,t)-\Gamma_i(m,t)=2$, which exceeds
the materiality threshold, so the disadvantage clause holds for every edge at
every period. Stipulating standing, admissibility, an actor-linked mechanism,
and a benefit-through witness for each edge yields $\GEX(e,t)=1$ throughout.
The construction is deliberately simple, and its simplicity carries the content.
Growth compares the configuration with its own past. The diagnostic compares
each participant with a feasible alternative. Nothing connects the two
comparisons, so improvement over the past is consistent with universal
disadvantage relative to what was available.
8. The Cycle–Trap Separation
This section distinguishes two structures that aggregate description conflates.
Its objective is to preserve the benefit-through requirement while recognizing
regime-level suppression.
A configuration exhibits circulating appropriation when it is a dynamic
cycle and every participant is the affected party on some edge with
$\GEX=1$, so that every participant both suffers a material disadvantage
through an actor-linked relation and holds an advantage running through some
other participant’s affected position.
A configuration is a systemic generative trap when every participant is
materially below its admissible comparison arrangement and no edge admits a
supported benefit-through witness.
Under a common specification, circulating appropriation and a systemic
generative trap cannot both hold. They are not jointly exhaustive: there exist
configurations that are neither.
Circulating appropriation requires a supported benefit-through witness on at
least one edge, since every participant must be affected on an edge with
$\GEX=1$ and that value requires a supported witness. A systemic trap requires
that no edge admit a supported witness. The two conditions are contradictory.
For non-exhaustiveness, take a configuration in which some edges carry supported
witnesses and others do not, while some participants are above their admissible
comparisons. Such a configuration fails the universality clause of
Definition ? and the no-witness clause of
Definition ?.
The residual class matters more than the dichotomy. Most configurations of
interest are likely to be mixed, with appropriative edges embedded in a
generally suppressed regime, and the dichotomy is an analytical device for
separating two pure cases rather than a taxonomy of actual regimes. A finding
that a configuration is neither is informative, and reporting it as such is
required by the framework’s three-valued discipline.
The separation has a practical consequence. In circulating appropriation, remedy
can attach to edges: the mechanism through which one participant’s advantage
runs through another’s position can be altered or removed. In a trap there is no
such edge, and remedy must alter the feasible set itself. Treating a trap as
circulating appropriation produces a search for an appropriator who does not
exist; treating circulating appropriation as a trap erases the benefit-through
guardrail and dissolves responsibility that is locatable. The forward-looking
structural responsibility of the social connection model
(Young, 2006) is the appropriate form for the trap branch,
where no party performed a locatable wrong, and it is the wrong form for the
appropriative branch, where parties did.
9. Countermodels
This section supplies the cases that discriminate the framework’s clauses.
Table 3 summarizes them. They are formal or hypothetical
structures and classify no observed regime.
| >p0.20
p0.27
X
| Structure | Aggregate description | Diagnostic outcome |
|---|---|---|
| Robbery circuit | Holdings unchanged, roles symmetric, reproducing | Every edge positive; cancellation fails |
| Growing circuit | Aggregate strictly rising, roles symmetric | Every edge positive; growth is the wrong comparison class |
| Reciprocal dependence | Same topology, aggregate rising | Disadvantage clause fails at every edge; not exploitative from the outset |
| Systemic trap | Everyone below admissible alternative, reproducing | No benefit-through witness anywhere; trap branch |
| Mixed regime | Some participants above, some edges witnessed | Neither pure class; report the profile |
| Compensated circuit | Holdings restored each period | Distributive register addressed; procedural register untouched |
| Absorbing circuit | Same aggregate as metastable case | Exit unavailable; position differs while topology does not |
Table. Cycle Countermodels and Their Diagnostic Outcomes. Each case holds the
aggregate description fixed and varies the structure, or holds the topology
fixed and varies the diagnosis.
Let each participant take a fixed value from its successor at each period and
have the same value taken by its predecessor. Aggregate holdings are unchanged,
role occupancy is symmetric, and the configuration is stationary. Stipulate
standing, an admissible comparison in which the taking is absent, an
actor-linked mechanism, a material disadvantage, and a benefit-through witness
for each edge. Then $\GEX(e,t)=1$ throughout, and by
Proposition ? the aggregate facts entail nothing to the
contrary.
Let each participant accept a constraint that expands the others’ reachable
repertoires, with the same circuit topology. Suppose each participant’s
capability under the arrangement weakly dominates its capability under every
admissible comparison. Then the disadvantage clause fails at every edge and
$\GEX(e,t)=0$ throughout. A cyclic topology therefore carries no diagnostic
content on its own.
Countermodels ? and ? share a topology and an
aggregate direction while differing in every diagnostic value. Together they
establish that neither closure nor reciprocity is evidence either way.
Extend Countermodel ? with a transfer at the end of each period
restoring every participant’s holdings. The distributive register is now
identical to a configuration in which nothing occurred. Every edge retains
$\GEX(e,t)=1$, since no clause of the predicate is a function of subsequent
transfers. This is the formal counterpart of Thesis ?.
Take two configurations with identical edge diagnostics, identical aggregate
capability paths, and identical role occupancy, differing only in that exit is
available in one and unavailable in the other. Every aggregate and topological
description coincides. The participants’ positions differ, which establishes
that such descriptions underdetermine position.
10. Open Questions
This section collects unresolved questions by register.
10.1 Formal and Aggregation Questions
The framework reports a vector of edge statuses together with distributive and
stability information. What reporting rule preserves the three registers,
avoids double counting when a participant appears on several edges, and remains
usable? Proposition ? establishes that no rule can be
derived from the aggregate facts alone.
Which of the stability classes of Definition ? are well
defined for a system whose edge states are three-valued and whose capability
objects are set-valued and partially ordered?
Thesis ? requires the admissible comparisons of all participants
to be jointly realizable for universal instantiation to be available. Under what
conditions on the feasible family does joint realizability hold?
10.2 Normative Questions
Is there a defensible rule for cases in which the registers conflict, or is the
correct conclusion that conflicts must be reported rather than resolved?
A participant occupying the affected position more often, or losing a larger
share of remaining capability, stands differently from one that does not. What
account of cumulative position across a cycle is defensible, and does it belong
to the distributive register or to a fourth?
Circulating appropriation supports locatable responsibility and a trap supports
forward-looking structural responsibility. What form applies to the mixed class
that Proposition ? shows to be non-empty?
10.3 Empirical Questions
No empirical case is classified here. The priority questions are whether any
domain exhibits a configuration meeting Definition ? under
independently constructed edge diagnostics; whether the trap branch can be
distinguished empirically from a configuration whose appropriator is merely
unidentified; and whether exit availability can be measured well enough to
separate absorbing from metastable configurations.
11. Limits and Revisable Research Program
This section consolidates the contribution and fixes the conditions under which
the paper should be narrowed or withdrawn.
11.1 Contribution
The conceptual contribution is the dissolution of the universal-participation
paradox by locating its equivocation on the contrast class, together with the
three-register separation and the statement of what compensation does and does
not reach.
The formal contribution is Proposition ?, which explains
why aggregate facts cannot cancel an edge predicate that never referred to them;
Proposition ?, which shows that growth and universal
disadvantage are compatible because they use different comparison classes; and
Proposition ?, which separates circulating appropriation
from a systemic trap and establishes that the two are not jointly exhaustive.
The provenance is stated in the document-status section: the closure, the
stability question, the paradox, and the robbery counterexample are
researcher-origin, and the reconstruction is agent-added.
The paper establishes no prevalence, classifies no observed regime, supplies no
aggregation rule, and recommends no institution.
11.2 Defeat Conditions
The conceptual contribution loses support if the equivocation diagnosis in
Thesis ? is rejected, since the paradox would then stand and
either void the diagnostic under universal instantiation or force the
over-permissive resolution the source discussion withdrew.
The formal contribution loses support if the edge predicate turns out to contain
an aggregate term after all, which would make Proposition ?
false; or if the admissible comparisons required by
Proposition ? cannot be jointly realized, which would make the
construction vacuous.
The separation loses support if the trap branch cannot be distinguished
empirically from a configuration whose appropriator is merely unidentified,
since the distinction would then be unavailable in practice however clear it is
in principle.
The paper loses its point entirely if the three registers admit a defensible
aggregation rule that makes distribution decisive, since the robbery
counterexample would then have to be rejected rather than accommodated.
11.3 Deferred Questions
The edge diagnostic itself belongs to a companion project, as does open-path
propagation, which stops where this paper begins. Identification of the
capability objects belongs to a third. Recovery, remedy design, and institutional
evaluation belong to a fourth, and legal responsibility to a fifth. This paper
supplies the structural separations such accounts would need and takes no
position on their content.
The resulting position is conservative. A cyclic topology is diagnostically
inert. Closure, symmetry, stability, aggregate neutrality, and aggregate growth
cancel nothing, because the predicate they are supposed to cancel never referred
to them. Universal instantiation is coherent and describes a determinate
condition rather than a contradiction. And a regime in which everyone is below
an admissible alternative may be either circulating appropriation or a
beneficiary-free trap, which are different structures requiring different
remedies and are distinguished by whether a benefit-through witness exists
anywhere.
References
Daron Acemoglu, Asuman Ozdaglar, and Alireza Tahbaz-Salehi. Systemic risk and stability in financial networks. American Economic Review, 105(2):564–608, 2015.
Elizabeth S. Anderson. What is the point of equality? Ethics, 109(2):287–337, 1999.
Samuel Bowles, Steven N. Durlauf, and Karla Hoff, editors. Poverty Traps. Princeton University Press, Princeton, NJ, 2006.
G. A. Cohen. The labor theory of value and the concept of exploitation. Philosophy & Public Affairs, 8(4):338–360, 1979.
Matthew Elliott, Benjamin Golub, and Matthew O. Jackson. Financial networks and contagion. American Economic Review, 104(10):3115–3153, 2014.
Alvin W. Gouldner. The norm of reciprocity: A preliminary statement. American Sociological Review, 25(2):161–178, 1960.
Mark Granovetter. Threshold models of collective behavior. American Journal of Sociology, 83(6):1420–1443, 1978.
Karl Marx. Capital: A Critique of Political Economy, volume 1. Penguin Books, 1992. Translated by Ben Fowkes; original German edition published in 1867.
Philip Pettit. Republicanism: A Theory of Freedom and Government. Oxford Political Theory. Oxford University Press, Oxford, 1997.
John Rawls. A Theory of Justice. Harvard University Press, Cambridge, MA, revised edition, 1999.
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.
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.
Tom R. Tyler. Procedural justice, legitimacy, and the effective rule of law. Crime and Justice, 30, 2003. Published in the Crime and Justice review series; reported page ranges differ across secondary sources and no range is asserted here.
Nicholas Vrousalis. Exploitation, vulnerability, and social domination. Philosophy & Public Affairs, 41(2):131–157, 2013.
Alan Wertheimer. Exploitation. Princeton University Press, Princeton, NJ, 1996.
Gabriel Wollner. Anonymous exploitation: Non-individual, non-agential and structural. Review of Social Economy, 77(2):143–162, 2019.
Iris Marion Young. Responsibility and global justice: A social connection model. Social Philosophy and Policy, 23(1):102–130, 2006.
Matt Zwolinski. Structural exploitation. Social Philosophy and Policy, 29(1):154–179, 2012.