# The Economics of Legibility: Reading Costs, Operator Order, and the Endogenous Formation of the Choice Set

**Diogo Carlotto**  
*Independent Researcher*

*Initial version: August 2026 · Editorial revision: October 2026 · Creative Commons CC BY 4.0 license*

> English translation of the Brazilian Portuguese original, `The_Economics_of_Legibility_PT-BR.md`, prepared with AI assistance. Operator symbols retain the Portuguese notation, including B for *busca* (search) and the subscript IA for *inteligência artificial* (artificial intelligence).

---

## Abstract

This essay proposes separating two institutional functions historically coupled within a single mechanism: reading, the informational operation that makes a contribution understandable and retrievable; and choice, the allocative operation that commits scarce resources under uncertainty. Its central thesis is that the order selection → preservation → search, here called Regime A, is not a logical necessity. It is a rational response to the historically high aggregate cost of reading, which could make broad preservation of contributions economically unviable. This cost combines preservation ($c_P$), retrieval ($c_R$), and interpretation ($c_I$): $c_L = c_P + c_R + c_I$.

The chain through which the choice set is formed ($U \to U_X \to U_P \to U_L \to \mathcal{C}$) distinguishes the potential universe of contributions ($U$), the externalized universe ($U_X$), the preserved universe ($U_P$), the legible universe ($U_L$), and the choice set ($\mathcal{C}$). By formalizing these stages, the essay argues that the boundary of what a decision-maker can consider is endogenous to their informational infrastructure for reading.

Reductions in the cost of contextual inference through artificial intelligence may make the inversion to Regime B (preservation → search → selection) economically defensible. This inversion allows the available alternatives to expand and problem formulations and partial solutions to precede their author's formal admission, thereby decoupling the reading function from the choice function. The separation does not eliminate material scarcity or the fiduciary responsibility of selection. Sustaining it at scale depends on the activities in which the informational advantage of preserving before allocating outweighs total coordination costs, including the mechanisms of provenance, verification, and retrospective attribution pertinent to each context.

*Keywords: operator order, transaction costs, economics of legibility, endogenous formation of the choice set, decentralized coordination, Friedrich Hayek.*

### Methodological note

This essay describes how preserving, retrieving, and interpreting contributions condition the alternatives available for choice. The set inclusions ($\mathcal{C} \subseteq U_L \subseteq U_P \subseteq U_X \subseteq U$) and functional decomposition of the aggregate cost of reading ($c_L = c_P + c_R + c_I$) constitute the structural basis of the problem, developed in Sections 2 and 3: they make explicit the layers of compression and precedence relations that determine the endogenous formation of the choice set. The consequences of these relations are distinguished from the behavioral and performance hypotheses presented in Section 6. The parameterization of dynamic equilibria and utility functions, together with econometric estimation, is deliberately deferred to a subsequent empirical agenda.

---

## Introduction

Platforms, rankings, résumé screening, manuscript submission, performance appraisal — for more than a century we have specialized in creating instruments that treat as a single mechanism what are actually two: making legible (giving alternatives an understandable, contextualized, retrievable form) and deciding (committing scarce resources to an alternative). But why have the formation of legibility and allocative decision-making been, and why do they remain, coupled within a single mechanism? And what becomes apparent when undoing this coupling becomes broadly possible?

The answer lies in the cost structure and is disconcertingly simple: faced with the need to select under restrictive reading costs, we resorted to instruments that, by reducing the universe of alternatives in advance, coupled reading and choosing into a single operation. Historically, this response was so economically rational and, for that reason, so indiscriminately adopted that, institutionally and individually, we stopped asking under what conditions these costs would still justify such coupling. Thus, the question almost no one asks would reveal, if asked, an obvious yet uncomfortable fact: where this coupling is no longer necessary, economic activity is operating with a fraction of the potential universe of alternatives available for selection, as though that fraction were the whole.

---

## 1. The question: why were reading and choosing coupled?

### 1.1. The naturalized convention

The history of economic institutions is, to a large extent, the history of how practical arrangements, born in response to transient material contingencies, crystallize into unquestioned dogmas. When a convention persists long enough, theory tends to forget its instrumental origin and treat it as an intrinsic feature of reality itself.

In numerous institutional processes of admission and allocation, access to visibility and consideration is subordinated to a prior positional contest:

- **In the labor market:** résumé screening conditions reading, which in turn conditions an interview and possible hiring.
- **In peer-reviewed scientific publishing:** editorial screening and reviewers' assessments determine whether the manuscript is published through that channel.
- **In large organizations:** annual performance appraisal and calibration committees determine which local initiatives gain visibility with senior management.
- **In digital markets:** proprietary curation and ranking algorithms determine which products or creators reach the consumer's attention.

In channels organized this way, we encounter the same architecture: to reach that channel's audience, an individual capability, a hypothesis, or a concrete solution must win a competition for admission. Being read and being selected have become coupled within the same operation. We must distinguish, however, a filter that merely delimits a reading from one that also conditions preservation: it is in this second function that present screening may foreclose future consideration.

This architecture has a fundamental analytical consequence: when reading capacity is restricted, access to consideration itself becomes a scarce resource. Agents then compete not only for the material resources to be allocated, but for the prior opportunity to be read by decision-makers. Credentials, formal positions, titles, and rankings function in this context as admission filters rationing access to a reading capacity unable to accommodate the heterogeneity of existing contributions. The scarcity an agent experiences ex ante does not necessarily arise from a lack of merit or an inability to generate value, but from the narrowness of the entry channel determining what gets considered.

**We must ask: why should this coupling be an unavoidable necessity?**

The response of economic common sense is usually: "It follows naturally from scarcity; one cannot choose without first inspecting the alternatives, and inspecting everything that exists is unfeasible." This apparent obviousness confuses two functions belonging to entirely different orders:

1. **The reading function (cognitive and informational):** making a contribution understandable, contextualized, and retrievable. It is an informational operation whose cost depends on technology and the organization of records.
2. **The choice function (allocative and decisional):** deliberating whether to commit scarce resources — capital, time, authority, or legal responsibility — to an alternative. It is an allocative operation conditioned by material scarcity and uncertainty, even though carrying it out also draws on informational resources.

This distinction is the logical core of the argument that follows. Selecting and preserving are distinct operations, conceptually and institutionally separable, that were historically coupled because reading was expensive. The fundamental economic problem is not that institutions or individuals need to select — material scarcity and uncertainty make choice an imperative of economic life. The real problem is that, under past technical constraints, selection was used as a criterion for informational preservation, determining what deserved to be kept and, therefore, what could be discovered in the future.

Constraints on reading may coexist with other reasons for controlling entry: responsibility, verification, protection of information, and institutional power. The following analysis isolates the role of reading costs and asks when these other objectives can be met without making allocative approval a condition of preservation.

### 1.2. The mechanism of ontological selection

The apparent naturalness of the sequence read -> choose conceals the fact that the first link is not neutral. If reading capacity is limited, then:

$$U_L \subseteq U$$

where $U$ represents contributions potentially externalizable within the scope considered, while $U_L$ represents the subset that, after externalization and preservation, can actually be made legible. The critical point is that allocative choice applies strictly to $U_L$, not $U$ — it is within this legible domain that alternatives can reach deliberation.

Consequently, the cost of reading is not merely transactional friction. It operates as a mechanism of ontological selection for the decision process — where "ontological" refers not to determining what exists in the world, but strictly to determining which entities or contributions become objects of deliberation. The compression required to make the world cognitively manageable under scarce reading capacity may exclude potentially relevant capabilities and contributions through *lossy* compression before they reach the choice stage — losses that are irrecoverable when the original record does not remain available. The allocative system may operate with perfect rationality conditional on its representation ($U_L$), but the representation is a lossy version of the potential universe ($U$).

We must therefore distinguish the efficiency of choice given a legible space from efficiency in constructing the space over which choices are made. We do not choose among everything that exists; we choose among what our infrastructure can make legible.

---

## 2. The historical constraint: the economic rationality of prior screening

We should not make the mistake of classifying present and past institutional funnels merely as bureaucratic pathologies. Under high constraints and costs of processing and interpreting heterogeneous data, prior screening is an economically rational response, particularly in relevant classes of information-intensive organizations.

### 2.1. Reading costs and the rationality of prior screening

The aggregate cost of reading ($c_L$) can be decomposed into three functional components:

$$c_L = c_P + c_R + c_I$$

where:
- **$c_P$** is the cost of relational preservation — maintaining the integrity, context, and provenance of a record over time;
- **$c_R$** is the cost of retrieval and search — locating and extracting relevant information when a demand arises;
- **$c_I$** is the cost of contextual inference and interpretation — examining, comparing, and interpreting the relevance and expected usefulness of retrieved alternatives in relation to a specific problem.

Each component of $c_L$ responds differently to technological development. Cloud infrastructure substantially reduced the marginal cost of preservation ($c_P$) and made large-scale preservation economically viable; semantic search algorithms expanded retrieval capacity ($c_R$), making it possible to locate formulations through semantic and contextual relationships rather than literal keyword matching alone; and, more recently, generative neural architectures have been advancing on the cost of contextual inference and interpretation ($c_I$). These advances must not, however, be confused with the disappearance of functional costs: cheaper storage is not sufficient to preserve context and provenance, just as low-cost inference does not guarantee the quality of interpretation.

This decomposition should be read for the same time horizon, volume of records and queries, and quality standard. Preservation generates maintenance costs over time; retrieval and interpretation generate costs as queries arise. The components interact: preserving context can reduce later interpretive effort, while enlarging the archive can increase search costs. The viability of keeping contributions available for subsequent readings depends on the total cost of these functions and the quality achieved, not merely the price of storage or an isolated inference. Nor does this availability require interpreting the entire archive deeply upon entry: some of that effort can be deferred until a demand justifies it.

As long as the aggregate cost of reading ($c_L$) remains high enough for the costs of broad preservation and on-demand retrieval and interpretation to exceed the expected benefits of reconsideration, the use of external validation markers, costly signals (Spence, 1973), and other instruments of prior screening — even operating through lossy compression of informational dimensionality — remains economically rational.

### 2.2. Distinguishing classes of scarcity

When the constraint imposed by aggregate reading costs ($c_L$) is overcome, its scope needs to be delimited by decomposing the generic notion of scarcity into three fundamental analytical categories:

1. **Intrinsic scarcity (physical and temporal):** arising from the insurmountable limits of matter, energy, and time. A particular individual's synchronous attention, for example, is scarce; no institutional arrangement or computational advance can multiply it.
2. **Scarcity through contractual commitment:** concerning, for example, the allocation of material resources, mobilization of financial capital, and assumption of fiduciary and legal responsibility. A committed resource ceases to be freely available for other uses, and undoing the commitment may entail real, even prohibitive, costs. This is a legitimate domain of the firm and hierarchical decision-making, though not exclusive to them.
3. **Institutionally constructed scarcity (positional):** including, for example, an opening in the organizational chart, a promotion, a ranking position, or a place in formal validation. This rivalry is created by the architecture of the decision mechanism (Hirsch, 1976).

This decomposition reveals a fundamental asymmetry between the categories. Whereas intrinsic scarcity (time and matter) and contractual scarcity (capital and legal responsibility) impose limits on resource availability and costs on reallocation, positional scarcity is different: its configuration depends on institutional engineering itself. The form of organizational positions, credentials, and validation spaces does not follow a law of physics, although those positions may mobilize genuinely scarce resources. What does not follow from that scarcity is the necessity of limiting preservable contributions to the number of people who can be hired. Faced with the practical impossibility of accommodating and processing all potential demands and formulations, institutions resorted to positional scarcity as a deliberate instrument of ordering and prior filtering. This is therefore constructed scarcity: a mechanism designed to restrict inflow and protect the organization's finite time and resources.

### 2.3. The economic basis of separation: non-rivalry of records and reading costs

If material and temporal scarcity and the costs of commitments remain, is changing the architecture of admission economically legitimate and viable?

In institutional arrangements that subordinate memory and deliberation to formal admission, the possibility of a human capability being used and preserved depends on the prior granting of a formal position. Before a contribution can enter an organization's memory and deliberations, the agent must be hired, given a position, or admitted to an authorized channel. Without that positional endorsement, the individual lacks even an institutional means of bringing their contribution into that horizon of reading; their capabilities remain invisible to that decision-making process, even when expressed elsewhere.

This architecture responds largely to aggregate reading costs ($c_L$). As long as relational preservation ($c_P$), contextual search ($c_R$), and in-depth interpretation ($c_I$) impose costs greater than the expected benefits of openness, maintaining active channels to process dispersed formulations is not economically rational. To prevent deliberative collapse through overload, decision systems tied listening to hiring: granting a formal position functions as the admission valve that rations in advance what the institution is willing to read and preserve.

The viability of breaking this dependency rests on a fundamental economic distinction between resources and information: unlike resources committed to wages, budgets, and positions — whose allocation to one use restricts their availability for others — recorded informational content is non-rival in use (Arrow, 1962; Romer, 1990). Consulting a record does not prevent another agent from using it, and preserving a contribution does not, in itself, require hiring its author or funding its execution. Production, custody, and consultation still consume resources; it is the informational use of the content that can be shared without exhausting it.

When reductions in preservation ($c_P$) and retrieval ($c_R$) costs are sufficient, something previously economically prohibitive becomes viable: a formulation can be externalized and retained relationally and retrievably before any decision about hiring and formally allocating the agent or using their formulation. An operational separation that institutionally constructed scarcity had historically prevented thus becomes possible:

$$\text{Preserving a contribution} \;\neq\; \text{Committing resources to its execution}$$

Breaking this coupling does not eliminate remuneration, contracts, or hierarchical authority: its strict effect is to shift the allocative decision from the entry point to the moment a demand justifies committing resources — allowing judgment to address what has already been formulated, rather than requiring *ex ante* credential screening. This judgment may precede execution; recognition of consequences, in turn, follows use. The decision system need not first admit an agent to a formal position merely to preserve and see what they can express.

**Broad preservation does not mean the institution must deliberate or decide on everything.** In some contexts, retaining records may cost less than mobilizing personnel and capital to assess every contribution immediately. The hypothesis depends on the benefits of keeping alternatives available outweighing preservation, search, protection, and maintenance costs. Interpretation ($c_I$) continues to consume resources. Deferring in-depth judgment and contractual formalization may be viable when a concrete need directs that effort; verifying this advantage requires comparing costs and outcomes in the context of use.

---

## 3. The change in the dominant regime: the order of decision operators

Although large language models are not the subject of this essay, their potential to reduce inference costs ($c_I$) makes a change in the dominant ordering of decision operators plausible. The logic and properties of each regime require analytical examination.

The following comparison examines when selecting a contribution for use also determines whether it remains in memory — understood here as the archive of records available for subsequent readings.

### 3.1. Regime A: select to preserve

The traditional decision model is structured in the following sequence:

$$S \;\to\; P \;\to\; B$$

In this regime, the decision system first applies the Selection operator ($S$) to the incoming flow. Only alternatives, contributions, or agents approved by this screening have their information incorporated into an active relational record (Preservation, $P$). When deliberation requires information, a Search ($B$) is performed within that pre-filtered archive.

The exclusion effected by Regime A is economic in nature and may change the very way data is collected and retained. Under high reading costs ($c_L$), systems tend to standardize and compress information at entry — reducing complex trajectories to rigid forms, simplified credentials, and linear metrics — to keep processing feasible.

The fraction that passes this initial screening enters the system's active memory and becomes useful for subsequent searches ($B$), potentially already stripped of some original texture and dimensionality. The remainder meets one of two fates: it is discarded and erased, or relegated to inert storage whose retrieval and interpretation costs ($c_R + c_I$) exceed its expected value under the conditions of that reading.

Whether through deliberate destruction or economic neutralization of search, the immediate effect converges: potentially relevant contributions are excluded from the decision system's horizon of choice. We must, however, distinguish the loss of a record from its present economic unavailability: a surviving record may become legible again under other demands or technologies.

### 3.2. Regime B: preserve to find

When a decline in the components of $c_L$ makes establishing a new coordination flow economically viable, the decision arrangement can adopt the following sequence:

$$P \;\to\; B \;\to\; S$$

In Regime B, Preservation ($P$) precedes screening: the trail of formulations, drafts, dependencies, and contributions is retained as it is produced — whether by the system itself, in shared infrastructures, or in agents' own custody — without retention depending on prior approval for a particular use. Custody seeks to maintain the texture of work — its context, relationships, and provenance — in retrievable form. When a specific demand arises — whether a question about an existing archive or the formulation of an open problem inviting new proposals and approaches — contextual searches and inferences, grouped within operator $B$, are conducted over this universe preserved with context and relationships, identifying connections, solutions, and capabilities not captured by simplified credentials.

Only after these contextual matches are identified is the Selection operator ($S$) activated — that is, the irreplaceable judgment of committing scarce resources, entering contracts, and assigning remuneration or authority to the alternatives found. Assessment can therefore address an already formulated contribution: the attribution of merit, relevance, or value follows an interpretive judgment of the contribution itself — the texture of what was concretely formulated — rather than an advance credential endorsement of its author. This assessment follows formulation but may precede execution and its results. Recognizing an actual consequence is another act, subsequent to use; selecting a hypothesis for development remains a decision under uncertainty.

This transition is not merely a chronological permutation of factors: it changes the institutional articulation of the operators and the universe over which they can act:

1. **Preservation ($P$)** no longer depends on approval for a particular use, allowing continuous, relational custody of the living texture of work and knowledge, including when no decision has yet been made to use it;
2. **Search ($B$)** operates over contributions whose retention did not depend on earlier allocative approval. Both regimes can employ semantic search and AI; the structural difference lies in the universe and relationships those technologies can retrieve;
3. **Selection ($S$)** is relieved of its function as memory's gatekeeper — where it had the additional task of rationing alternatives in response to reading costs — to serve its primary economic purpose: sovereign allocation of scarce resources over an already legible horizon, with the fiduciary and contractual responsibilities each arrangement imposes.

The designation Selection ($S$) is retained in both regimes because the analysis addresses precisely this dual scope of choice: deciding what will be used and, when coupled to preservation, also deciding what will remain available for future readings. In Regime A, selection also acts as memory's gatekeeper: its screening conditions what is preserved for subsequent searches. In Regime B, preservation ceases to depend on that screening and selection performs its allocative function over an already legible horizon. The change in the scope of $S$ thus constitutes the very decoupling being investigated. The diagrams express this precedence relation; they do not enumerate every evaluative operation involved in a decision.

The regimes characterize the relationship between selection and preservation within a defined flow. The same institution may preserve rejected proposals in one channel and eliminate records in another. Diagnosis must identify the archive, readers, and access conditions: a refusal to use a contribution acts as gatekeeping when it also prevents that contribution from remaining available for other economically viable readings.

Search continues to select candidates for each reading. This contextual selection delimits what will be examined on that occasion without determining what may continue to exist in the archive. The condition of Regime B is that a contribution need not be chosen to become available for consideration; no particular decision not to use it should, by itself, end its availability for other readings.

```
REGIME A (Traditional — Scarce Reading Capacity):
[Flow of Capabilities] → [SELECTION / FILTER] → [Active Preservation] → [Restricted Search]

REGIME B (Emerging — Falling Reading Costs):
[Flow of Capabilities] → [RELATIONAL PRESERVATION] → [Contextual Search] → [SELECTION / ALLOCATION]
```

### 3.3. The formation of the choice set

The decisive point can be expressed simply: the universe of alternatives a decision system can consider does not necessarily coincide with the broader universe of potentially relevant capabilities and contributions. Between them operates an infrastructure of preservation and search constituting the chain through which the choice set is formed:

- **$U$ (Potential universe of contributions):** the set of possible recordable manifestations of human capabilities, bounded by a scope of agents, time, and material conditions. It is an analytical universe of possibilities, not a complete inventory known to the decision system;
- **$U_X$ (Externalized universe):** the subset of $U$ actually formulated, expressed, or externalized by an agent, regardless of whether a particular system captured it. It includes problem formulations, hypotheses, and solutions without requiring completed work;
- **$U_P$ (Preserved universe):** contributions from $U_X$ whose records remain retained and indexable by the infrastructure — whether in common repositories or in the agent's own custody;
- **$U_L$ (Legible universe):** preserved contributions that are concretely retrievable and interpretable in relation to a specific demand;
- **$\mathcal{C}$ (Choice set):** the subset of $U_L$ admitted as concrete alternatives for deliberation.

The chain through which the choice set is formed is therefore expressed by the progression:

$$U \;\longrightarrow\; U_X \;\longrightarrow\; U_P \;\longrightarrow\; U_L \;\longrightarrow\; \mathcal{C}$$

This formally implies the successive inclusions:

$$\mathcal{C} \subseteq U_L \subseteq U_P \subseteq U_X \subseteq U$$

where each set tracks units of contribution in different states of availability — from externalization to admission into deliberation. These inclusions do not identify a person or their entire capability with the record representing them.

What makes a record effectively legible ($U_L$) requires clarification: this condition follows not from the record's mere physical or digital existence, but from the system's practical capacity to locate and interpret it at viable cost when a need arises.

Two conditions guide this legibility. **Situated demand** determines a record's relevance to the question guiding its reading, including when the inquiry is exploratory and has no defined application yet. **Technology** conditions its retrieval and interpretation: a dense report may remain beyond the reach of mechanical keyword search (literal matching), yet become legible and actionable when the system has semantic inference tools capable of understanding context.

Subject to each reader's access conditions, the same informational particle may thus be fully legible for one purpose under a given technology and remain invisible under different conditions.

To make this situated character explicit, we can write:

$$\mathcal{C}_{i,q,t} \subseteq U_L(i,q,t) \subseteq U_P(t)$$

where $i$ identifies the reader or decision system, $q$ the demand, and $t$ the time of reading. The preserved archive may support different choice sets, subject to each reader's access conditions. A contribution absent from $\mathcal{C}_{i,q,t}$ may enter another choice set without needing to be produced again or readmitted to preservation. For simplicity, these indices remain implicit in the essay's other expressions.

Demand $q$ may itself be formulated and revised. A participant can make legible a problem the organization has not yet recognized, state what they consider important, or propose another criterion of success. The axiological dimension here refers to these criteria of value assumed by participants. Openness to contribution may therefore expand both the available responses and the questions and purposes submitted to deliberation. Defining $q$ in a query is an analytical delimitation; it does not require treating the formation of demands as external to coordination.

The chain describes a passage, not an immutable catalogue. Reading a contribution can prompt new questions, experiments, and combinations. Once externalized, these formulations enter another passage through $U_X$ and $U_P$. Dynamic expansion of the formulated universe must be distinguished from retrieval of existing records: the two effects can reinforce each other, but require different empirical observations.

A parallel note is needed on the chain's initial link ($U \to U_X$): no formulation arises in a vacuum of demand. If an individual devoted time and cognitive energy to expressing and externalizing a capability ($U \to U_X$), that externalization may occur, among other motivations, through two paths relevant to the argument: on their own initiative, responding to an internal or idiosyncratic demand (as in Raymond's [1999] formulation that good software projects often begin with a developer's personal need); or in response to an open problem formulated by the coordination environment.

In the open arrangement examined here, both paths allow externalization without requiring prior authorization or screening of who has the prerogative to formulate. A record's coordination value rests on its capacity to respond to subsequent external demands.

Although the strict dynamics of formulation are not this essay's central subject, an agent's investment in externalizing a capability beyond their private circuit responds in part to the expectation that their contribution will find a reading channel able to accommodate it. Under Regime A, knowing that reading is rationed through prior screening, the agent may anticipate the cost of invisibility and disinvest in externalizing heterogeneous capabilities, concentrating on those fitting standardized validation formats (a relationship detailed in Section 4.2).

In Regime A, prior selection restricts what reaches $U_P$ and may alter its representation — inducing compression into standardized formats — doubly narrowing the horizon over which future searches can operate. In Regime B, broader preservation frees $U_P$ from this initial screening, allowing the texture of formulations to be retained and shifting the decisive role in delimiting the boundaries of $\mathcal{C}$ to retrieval and contextual interpretation operators.

> **The informational boundary of the choice set ($\mathcal{C}$) is endogenous to the infrastructure's regime of legibility.**

### 3.4. The conditions for regime inversion

Inverting the order of operators does not follow from automatic technological determinism, but from deliberately establishing three economic and architectural requirements:

1. **Economic viability of broad custody ($c_P$ under control):** coordination infrastructure must make it sustainable to retain a broad, granular archive of formulations without prohibitive storage and integrity costs. This viability can expand when preservation is distributed — through common infrastructures or creators' own custody — relieving the decision system of centralized physical storage costs and allowing resources to concentrate on indexing and archive integrity. As preservation costs ($c_P$) are compressed by scalable data architectures, retaining work in its living texture becomes economically defensible, enabling future mobilization ($c_R$). Distributing custody, however, does not eliminate its cost: assessment must include resources borne by creators and archive maintainers.
2. **Precision in contextual retrieval architecture ($c_R$ and $c_I$ calibrated):** retaining large information volumes is insufficient; semantic search and inference mechanisms must be designed with enough precision to prevent archive expansion from becoming interpretive congestion. If ambiguity, false positives, or dispersion produced by search are excessive, gains from broad preservation may be canceled by overloading decision-makers. Inversion therefore requires instruments capable of identifying alternatives with sufficient contextual relevance, without imposing excessive redundancy or systematically excluding relevant contributions.
3. **Sufficient reconsideration value relative to additional costs:** contributions must exist that early screening would exclude and whose subsequent use can compensate for the additional costs of preservation, retrieval, interpretation, and verification, including possible losses in present decisions. The comparator is a Regime A with access to the same technology, over the same analytical horizon and with the pertinent quality criteria. Diversity of demands and cross-domain complementarities may raise this value. Its existence and magnitude are hypotheses to examine in each environment, not automatic consequences of archive expansion.

These three conditions delimit the assessment of inversion's economic viability. Its sustainability, however, faces additional challenges, notably the degree of **accountability** — the possibility of linking acts to those responsible and of that linkage producing consequences when information is deliberately manipulated.

There is also a distributional question: those bearing the costs of formulation and preservation may not receive the future benefits. An architecture may expand coordination possibilities while still offering insufficient incentives to participants or maintainers. Falling costs make certain arrangements viable; the distribution of costs, access, and recognition conditions their adoption and continuity.

Linking declarations to persistent identities, contractual relationships, or reputational consequences can raise the expected cost of producing adversarial noise for its issuer, thereby reducing incentives to manipulate. The effect depends, however, on the probability of detection, application of consequences, and benefits of manipulation. The mere existence of identity or signature does not ensure that the expected cost of adversarial behavior exceeds its benefit.

The situation changes when accountability is insufficient or absent. This may happen when creating new identities permits escape from the consequences of misconduct — regardless of whether the network is centralized or distributed. In these environments, the low cost of producing and disseminating records, when below the expected benefit of manipulation and unaccompanied by sufficient consequences, generates an adverse incentive: strategic agents have reasons to inject large quantities of "formulated noise" — adversarially fabricated artifacts designed to manipulate semantic search and capture the system's attention. This risk of **informational pollution** can degrade the preserved base ($U_P$) and inflate *ex post* contextual inference costs ($c_I$), threatening to overload deliberation.

Choosing an accountability mechanism involves costs and exclusionary effects. Formal identification, persistent reputation, attestations, and economic commitments may provide useful signals, but none reduces manipulation in every context. Their requirements may also restrict participation by people with few resources, without credentials, or with a legitimate need for privacy. The design question is how to obtain informative signals for situated assessment without turning every credibility requirement into prior authorization to preserve.

In environments with low native accountability, distributed preservation can be combined with local criteria for retrieval, verification, and trust. A reader need not examine or host everything others preserve. The economic criterion is that protecting one reading need not eliminate a record's availability for others. Provenance graphs, attestations, and consequence records are design possibilities; their effectiveness against manipulation and congestion depends on implementation and incentives.

When these three conditions hold and strategic pollution is institutionally contained, the informational advantage of Regime B ($P \to B \to S$) lies in retaining, for different readings, contributions that Regime A's screening ($S \to P \to B$) would remove from subsequent searches. This expanded availability does not imply that every query produces a larger or better choice set: realizing its potential depends on retrieval and contextual interpretation. This qualification is essential: traditional prior selection served not only to ration reading but also to distinguish participant attributes under asymmetric information (Spence, 1973). Regime B's sustainability therefore depends on new validation mechanisms compensating for the credibility and signaling functions prior screening performed. This essay delimits the conditions under which inversion can be viable and economically rational along the informational dimension; demonstrating its full sustainability in open networks is a subsequent architectural design question.

It is precisely this conditionality that distinguishes the present thesis from a naive defense of technological abundance. The optimal operational order depends on the design of reading infrastructure, not computational automatism.

Finally, these conditions need not arise simultaneously: preserving broad archives ($U_P$) may become established in a system long before a complete inversion of the selection regime is formally justified.

### 3.5. Institutional approximations to Regime B

The inversion proposed in this essay does not arise in an institutional vacuum. Contemporary coordination arrangements already reduce, to varying degrees, the dependency between agents' prior authorization and the discovery of relevant capabilities. These empirical experiences should not, however, be confused with the full realization of Regime B. Their analytical usefulness lies precisely in distinguishing a reduction in the compression of an agent's profile from a structural separation between preserving a contribution and selecting its author.

So-called *internal talent marketplaces* are a telling example of the first category within corporations. The literature on internal markets examines, among other questions, how promotions influence applications (Keller & Dlugos, 2023); the relevant focus here is the architecture through which capabilities are represented. These platforms seek to expand the representation of individuals beyond their formal duties, recording skills, experiences, interests, and availability, and using that base to bring workers together with specific projects or demands. The architecture partially shifts the traditional allocation process:

$$\text{formal position} \;\longrightarrow\; \text{candidate} \;\longrightarrow\; \text{prior selection} \;\longrightarrow\; \text{allocation}$$

to an attribute-based flow:

$$\text{capability profile} \;\longrightarrow\; \text{demand} \;\longrightarrow\; \text{matching} \;\longrightarrow\; \text{allocation}$$

The potential economic gain is clear: a capability that would remain invisible because it was confined to a particular department or role becomes retrievable when a compatible need arises. Technology thus reduces the cost of searching for and matching agents.

Richer profiles are insufficient to classify an architecture as Regime A or B. An internal market may preserve information before selecting someone for a project while restricting its universe to workers already admitted. Where only profiles are preserved, another limitation remains: the system locates people with compatible attributes but does not directly retrieve autonomous formulations produced before allocation. We should distinguish the openness of the participant universe, the granularity of representation, and the dependency between preservation and choice. The profile-based sequence is:

$$\text{individual} \;\longrightarrow\; \text{attribute indexing} \;\longrightarrow\; \text{search} \;\longrightarrow\; \text{selection / allocation}$$

In Regime B, by contrast, the primary unit of custody and retrieval is the **contribution** or formulated artifact:

$$\text{contribution} \;\longrightarrow\; \text{relational preservation } (c_P) \;\longrightarrow\; \text{contextual search } (c_R, c_I) \;\longrightarrow\; \text{deliberative selection}$$

The distinction is not merely terminological; it redefines the formation of the choice set. This dynamic applies both to **spontaneous** contributions (arising from the agent's exploratory initiative or idiosyncratic demands) and to contributions **elicited by the decision system's open formulation of a problem**. The critical point is that articulating a challenge does not require selecting in advance who has the prerogative to address it. In arrangements where only previously authorized agents can formulate, admission already restricts the externalized universe ($U_X$). Opening a problem removes this restriction but is insufficient to characterize Regime B: rejecting a proposal must also not end its availability for subsequent readings. Furthermore, a solution structured for a specific present demand acquires informational optionality — the capacity to be reconsidered under other demands, developed in Section 3.7: its record remains preserved so that future external demands, entirely unforeseen at conception, can rediscover and mobilize it.

From this perspective, certain structures of open production, notably *free software* (*open-source*) communities, approximate Regime B's mechanics more directly (Benkler, 2002; Raymond, 1999). In practices where proposals and discussions remain recorded before integration, contributions emerge through both paths: purely voluntary and exploratory developer initiatives, and decentralized responses to problems, bugs, and challenges openly catalogued by the community. The decision system or project governance formulates demand — the report of a problem to overcome — without needing to grant the author eventual integration of their solution in advance. Submission and integration rules may continue to exist; their existence does not require discarding rejected proposals. A contribution enters the preserved archive ($U_P$) autonomously. Subsequently, contextual assessment ($c_R, c_I$) confronts formulated proposals with the stated need, forming the set of alternatives ($\mathcal{C}$) for deliberation about possible integration of an artifact into the system's core.

The causal sequence thus approaches full inversion:

$$\text{contribution} \;\longrightarrow\; \text{preservation } (U_P) \;\longrightarrow\; \text{contextual retrieval } (U_L) \;\longrightarrow\; \text{alternatives for deliberation } (\mathcal{C})$$

The decisive analytical point is that the system's prior admission of the author ceases to be a condition for work to exist and be preserved — whether in common infrastructures or in the agent's own custody. A formulation can enter the horizon of legibility and await deliberation without prior institutional endorsement of its producer. Assessment can address an already formulated contribution, while recognition of consequences depends on what its use actually produces. In arrangements subordinating memory to admission, the decision system must arbitrate in advance which agents deserve to have their ideas recorded; in structures of broad preservation, both allocative deliberation and the attribution of value can be deferred until the contribution's expected usefulness can be confronted with a real need.

This does not imply that *open source* fully realizes Regime B. Externalization remains conditioned by asymmetric incentives; decentralized retrieval faces noise and integration costs; and final incorporation of the artifact remains subject to scarce attention and fiduciary control. These arrangements offer examples in which preserving a contribution can precede its author's admission to an organization. They illustrate the possibility of that separation, but do not demonstrate its economic advantage in every context or, by themselves, validate Legible's performance.

The structural differences between these arrangements are summarized in Table 1:

**Table 1**  
*Comparison of Architectures of Legibility and Selection*

| Architecture | Primary unit of representation | Dominant question | Timing of selection | Impact on the formation of $\mathcal{C}$ |
|---|---|---|---|---|
| **Regime A** | Agent or contribution subject to retention screening | *What will be admitted to the archive?* | Conditions preservation and, in some channels, formulation itself | $\mathcal{C}$ is confined to the archive resulting from earlier screening |
| ***Internal Talent Marketplace*** | Individual and vector of attributes/skills | *Who has capabilities compatible with this opportunity?* | May preserve profiles before project selection; participant entry may depend on prior hiring | Expands the reading of admitted participants' capabilities; openness to autonomous contributions depends on design |
| **Open production (*Open Source*)** | Contribution / formulated artifact (*patch*, module) | *Which proposal can solve this demand?* | Integration may be decided after preserving the proposal | $\mathcal{C}$ may admit spontaneous formulations or responses to open problems; rejected proposals may remain retrievable |
| **Regime B** | Retrievable contextual contribution, including problems and partial solutions | *What can be mobilized or developed in response to this demand?* | Allocative choice follows preservation and reading; recognition of consequences follows use | $\mathcal{C}$ depends on demand, reader, access, and retrieval quality; rejection does not require discarding |

The comparison shows why contemporary technological transformation cannot be reduced to mere optimization of *matching*. On platforms where matching operates over a universe whose preservation depended on prior authorization, improving search does not undo that bottleneck. Regime B operates at a deeper institutional level: when declining custody ($c_P$) and contextual reading ($c_R, c_I$) costs make adoption viable, it relieves selection of its historical gatekeeping function over what deserves to be remembered, restoring its legitimate function of fiduciary allocation over what has proved relevant.

Corporate talent marketplaces establish the viability of making profiles more flexible; open production demonstrates the viability of preserving artifacts independently of prior authorization. Regime B generalizes these partial manifestations into an economic design principle: when reading infrastructure is sufficiently efficient, the decision system need not decide in advance what has the right to be recorded. It can preserve broadly and allow subsequent external demands to determine each formulation's relevance.

The economic question then shifts from *"whom should we select so we can discover what they can do?"* to a structurally prior inquiry: *"what can we preserve so that its usefulness can be discovered when an external demand arises, without having to select in advance who has the right to be recorded?"*

### 3.6. Agency, provenance, and retrospective attribution

As aggregate legibility costs ($c_P, c_R, c_I$) fall, other coordination costs become relatively more salient. Transaction cost literature (Coase, 1937; Williamson, 1979) traditionally identifies costs associated with search, negotiation, monitoring, and verification. This essay proposes that expanding the preserved universe and postponing validation until a demand justifies examination may shift the economic bottleneck toward retrospective attribution — making the provenance of formulations, their dependencies, and the consequences recognized by beneficiaries retrievable in distributed production processes. This task draws on preservation, retrieval, interpretation, and verification; its cost should not be added again to reading components already counted.

As Alchian and Demsetz (1972) showed, in team production the aggregate product makes each input's marginal productivity difficult to observe. In the classical firm, this asymmetry is mitigated by a central monitor who supervises inputs and establishes contracts *ex ante*. In Regime B, however, the absence of entry screening and the possibility of distributed custody — or custody by creators themselves — make retrospective (*ex post*) recognition of value and relevance a possible basis for remuneration, reputational credit, and new agreements, compatible with wages, grants, and advance funding. The broader the preserved archive and the more fluid the recombination of ideas, the greater the effort of reconstructing the genealogy of formulations may be: reconstructing who conceived what, which dependencies were declared, and which consequences were recognized downstream, while distinguishing this reconstruction from causal demonstration of added value.

Retrospective attribution does not presuppose a universal measure of merit or canonical distribution of value. Records of authorship, dependency, and consequence provide elements for situated judgments; a declaration of dependency does not, by itself, demonstrate causality, just as a consequence recognized by one beneficiary does not establish the contribution's value for everyone else. Verification and interpretation of these elements remain necessary for decisions that rely on them.

It is useful to distinguish a relationship inferred by the system from a declaration that a participant explicitly stands behind. Inferring that an idea influenced a work and recording that its author declared a dependency on it produce information of different kinds. The same holds for inferring a preference versus recording that someone formulated a problem or recognized a consequence. An agency record adds to the archive the declaration through which a participant takes responsibility for a formulation, dependency, or value judgment; its occurrence and content can be examined separately from the reader's inference.

An AI agent can help formulate, relate, and retrieve these records. When the object recorded is a human position, attributing it requires the participant's corresponding act; a prediction of what they would say does not constitute that act. A declaration's authenticity, its content's truth, and its economic importance are distinct questions. The declaration may be mistaken or strategic, and recording it neither demonstrates causality nor establishes a universal measure of merit.

Opening preservation and recording these acts are complementary dimensions. Regime B, by its minimal definition, requires no specific agency protocol; an institution with restricted channels can also record declarations. The additional hypothesis is that combining broad preservation, context, declared dependencies, and recognized consequences improves particular readings and forms of coordination. Its effect must be distinguished from the gain of merely increasing data volume.

The present thesis limits itself to identifying this possible shift in the coordination bottleneck and its economic functions. Designing institutional and protocol arrangements capable of lowering retrospective attribution costs — without restoring the gatekeeping filter that makes approval a condition of memory — constitutes the next frontier of this research program.

### 3.7. Informational optionality

Interpreting the regime change as a simple reduction in Stiglerian search costs (Stigler, 1961) is insufficient to characterize the proposed transition. Lower search costs improve choice over a given archive. The mechanism described here operates at a structurally prior layer: it changes the conditions under which the searchable universe itself is constituted.

Regime B's fundamental economic justification rests on the concept of informational optionality — the capacity to reconsider a contribution under demands not yet known — an application, proposed in this essay, of the intuition of option value to coordination archives and informational memory (Dixit & Pindyck, 1994). Option value follows from a fundamental asymmetry of irreversibility: whereas retaining a record leaves the decision about its use open — preserving does not require executing — discarding or reductive compression under Regime A is often irreversible, irrecoverably destroying the texture of the original formulation. Low-cost preservation — whether in common infrastructures or creators' own custody — preserves the capacity to deliberate in the future. This asymmetry makes neither retention costs recoverable nor unrestricted disclosure reversible: preservation and granting access are distinct decisions.

This optionality manifests over two distinct deliberative horizons: **intertemporally**, by retaining formulations whose value becomes observable only when future external demands or inquiries reveal their relevance (the value of waiting under uncertainty; Dixit & Pindyck, 1994); and **across concurrent present problems**, by allowing the decision system to accommodate and preserve plural approaches in $U_P$, exploring gains from parallel experimentation and postponing the decision about which proposals to develop until contextual reading ($B$) permits examination of their suitability. Parallel experimentation may broaden this examination but also requires resource allocation, whose costs enter the comparison (Nelson, 1961).

There is also a cross-domain dimension: a contribution can acquire usefulness when associated with a problem or another record outside its original context. Consider an extraction technique conceived for a mechanical problem and later interpreted as a hypothesis for another field. The connection requires preserving aspects of the formulation that a summary oriented only toward its first use might eliminate. A record's value thus also depends on the readers, questions, and complements it may encounter.

This complementarity permits gains disproportionate to the number of records added: a contribution may make a previously unavailable combination viable. Automatic combinatorial growth of economic value does not follow; locating, testing, and integrating the combination remain costly. The analytical consequence is that assessing contributions solely by their isolated usefulness to a present demand may underestimate the value of preserving them for other coordination.

Informational optionality also contains a distributional question. Those bearing preservation costs may not share in the benefits other readers obtain from a record. Under these conditions, discarding can be rational for the institution maintaining the archive while still foreclosing possibilities whose use by third parties would justify retention. The private value of custody and the value of its uses across coordination need not coincide. Cheaper reading expands preservation's technical viability; sustaining memory also depends on the distribution of its costs and benefits.

The conditionality established in Section 3.4 must nevertheless be emphasized: an informational option's existence does not guarantee economic viability if its exercise costs are prohibitive. Exercising the preserved option requires retrieving and interpreting the record ($c_R + c_I$). If archive expansion occurs in environments where adversarial issuance offers benefits exceeding its costs and consequences, proliferating noise may raise retrieval and interpretation costs until the option's net value disappears. The rationality of broad preservation therefore depends on the robustness of the validation architecture protecting the indexed base from interpretive congestion.

Regime A thus faces a structural trap: under high reading costs, it must anticipate future value at the entry point — preserving means predicting — preemptively discarding contributions whose retention is not justified by present criteria, whether credentials, topics, or estimated usefulness. Regime B enables temporal separation: preserve at low cost to keep contingent valuation possible — preserving means retaining the capacity to discover.

Under genuine uncertainty — where the space of future states and the contingent relevance of each contribution do not admit reliable probabilization (Knight, 1921) — informational optionality can be appreciated without reliable stochastic pricing. The thesis does not require the decision system to know probability distributions over the future: low-cost preservation retains the capacity to discover uses that irreversible discarding may eliminate. Demonstrating optimality under a minimax regret criterion (Savage, 1951), however, requires specifying states, costs, and losses; it does not follow from falling $c_P$ alone. Management, access, and congestion costs also enter this comparison, under the conditions in Section 3.4.

---

## 4. A formal consequence of coupling: the formation of the choice set

### 4.1. The neglected economic variable

Before modeling search behavior (Stigler), decentralized coordination (Hayek), or the firm's boundary (Coase), much economic analysis assumes — implicitly or explicitly — that agents deliberate over a given set of alternatives. A broad class of choice and strategic interaction models takes the set of alternatives as part of the problem's pre-established conditions; even in information economics subfields that endogenize agents' attention — such as rational inattention — the universe of objects to which attention may be directed is often still treated as given.

This is the variable the chain of choice-set formation, defined in Section 3.3, makes analytically explicit. Formalizing the transition from the initial universe to the effective choice set reveals that the boundary of what can be chosen is not an exogenous property of the economic environment, but the result of reading infrastructure conditioning every link in the chain.

A direct implication of the distinction established in Section 1 is this set's endogeneity: a decision system's reading capacity and the order of its operators ($S$, $P$, $B$) determine which capabilities and contributions become admissible alternatives for economic decision-making.

This proposition is not trivial. Its analytical distinction is not a generic assertion that choice sets are endogenous — a dynamic already formalized at the individual level in rational inattention and competitive marketing literature — but an explicit account of how the decision system's informational boundary is endogenous to its reading infrastructure. It implies that infrastructure can change not only search efficiency within a fixed space, but the dimensionality of alternatives — their diversity of attributes and relationships — that the decision system can admit to deliberation. A decision system operating under Regime B does not simply choose faster or more cheaply; it may gain access to alternatives and relationships that, under Regime A, remained structurally invisible to deliberation.

### 4.2. The anatomy of informational loss

To understand the anatomy of informational loss, we must distinguish precisely the states in which a human capability may exist within the social fabric — and, within them, two distinct causes of non-externalization that should not be conflated into a single residual category:

- **(1) The contribution formulated and made legible to the decision system:** it has been preserved and can be retrieved and interpreted in relation to the demand considered; it belongs to $U_L$, although it may still not enter that occasion's choice set.
- **(2a) The tacit dimension not fully codified:** forms of practical knowledge whose enactment is not reducible to the available explicit description (Polanyi, 1966). The essay does not assume their complete conversion into records, but allows that observable manifestations of this knowledge can be preserved;
- **(2b) The potentially externalizable capability whose formulation never occurs:** the agent possesses a capability that could, in principle, be articulated, but the expected benefit of externalization, given available legibility infrastructure, does not compensate for the effort of formulation. Unlike category (2a), the mechanism isolated here is institutional: it belongs to $U$ but does not pass into $U_X$, precisely in anticipation of the prevailing legibility architecture.
- **(3) The capability formulated and externalized (belonging to $U_X$), but whose institutional legibility never formed or did not persist:** solutions actively produced for specific problems and discarded by Regime A's credential funnel without their texture being examined, as well as trails of engineering drafts, local routines, and prototypes dissipated in the limbo of dead files. Category (3) covers both cases where legibility existed and was lost and cases where it simply never formed institutionally.

The primary object of this thesis is categories (2b) and (3). Complete codification of tacit knowledge (2a) is not assumed; the argument is that some potentially relevant capabilities, whose magnitude requires investigation, may remain outside the choice set through two distinct mechanisms — because they were never formulated, given the anticipated absence of a reception channel (2b), or because they were formulated and their institutional legibility did not form or persist (3) — rather than because of inherently low productivity.

The distinction from (2a) concerns the full content of knowledge, not the exclusion of its holder. A case description, formulated diagnosis, or recorded result may make a manifestation of competence retrievable without codifying the entire experience that produced it. This partial legibility can help others find the person and establish cooperation while preserving the need for their situated participation.

Category (2b) focuses on the present decision to externalize a capability. Its mechanism can be expressed as a participation condition: the agent formulates when expected benefits — material, relational, intellectual, or recognition-related — compensate for effort, exposure, and other costs. A more credible prospect of being read can change this condition for some agents; AI assistance may also reduce formulation effort. Neither change guarantees that everyone crosses the threshold. The relevant hypothesis is that some additional contributions are produced because the channel of consideration has changed.

This focus connects with research on exposure to innovation and inventors' trajectories (Bell et al., 2019). It requires neither reducing their mechanisms to skill formation nor rigidly opposing formation and opportunity. To isolate the proposed effect, we must observe whether agents with available capabilities start externalizing them when conditions of preservation, reading, and recognition change.

The hypothesis that legibility infrastructure affects investment in formulation can be examined in dialogue with models of statistical discrimination, attention, and costly verification (Coate & Loury, 1993; Echenique & Li, 2025; Townsend, 1979). The mechanism to identify is the effect of changing the viability of preservation and reconsideration on the decision to formulate, accounting for beliefs, incentives, and verification costs. These mechanisms may coexist. Formulating an equilibrium model and identifying its effects belong to the subsequent agenda.

This dynamic allows a circuit of persistent invisibility. The expectation of not being read discourages formulation; a shortage of observed contributions may in turn be interpreted as evidence of scarce capabilities, legitimizing continuation of the restricted channel. The architecture thus helps produce the evidence that appears to justify it. A consequence for evaluating openness follows from this circularity: some of what the new arrangement can make legible is not yet in the archive, because it first needs reasons to be formulated. The prospect of reconsideration and recognition must become credible to change this participation condition.

### 4.3. Delimitation in relation to adjacent literatures

Formulating the choice set's endogeneity does not oppose adjacent theoretical traditions — it organizes them. Each illuminated a dimension of the informational problem; this essay makes explicit the dependency between formulation, preservation, and consideration, taking operator order as a design variable. The Economics of Legibility proposes articulating these contributions within a unified syntax:

- **In relation to Search Theory (Stigler, 1961; Rothschild, 1973):** classical search literature investigates optimal stopping behavior given a pre-existing space of alternatives. Legibility theory shifts the analysis to the endogenous determination of the searchable space itself.
- **In relation to Organizational Learning (March, 1991; Levitt & March, 1988):** the tension between exploration and exploitation illuminates the problem of retaining experience without foreclosing discovery of new possibilities. This essay analyzes how falling $c_L$ changes the boundary between internal archives and open ecosystems.
- **In relation to Real Options Theory (Dixit & Pindyck, 1994):** the logic of optionality under uncertainty is applied to preserving formulations whose future uses may not have been anticipated — keeping open the decision to mobilize them in response to new demands.
- **In relation to Matching and Search-and-Matching Theory (Gale & Shapley, 1962; Mortensen & Pissarides, 1994):** matching analysis presupposes alternatives recognizable as candidates for a match, although their entry and job creation may also be part of the model. The Economics of Legibility investigates how peripheral agents and capabilities become observable candidates before matching occurs.
- **In relation to Signaling Theory (Spence, 1973):** classical signaling literature explains how senders invest in costly signals to mitigate information asymmetries concerning attributes unobservable *ex ante*, treating that unobservability as a structural assumption. The Economics of Legibility examines an endogenous dimension of this constraint: it makes explicit that, under high aggregate reading costs ($c_L$), directly inspecting contributions' texture may be unfeasible, making the signal a compression shortcut for rationing consideration. Falling $c_L$ does not invalidate signaling where relevant asymmetries persist, but changes its economic necessity as the gatekeeping filter of coordination archives.
- **In relation to peer production (Benkler, 2002):** expanding combinations of participants, resources, and projects is a direct antecedent of the present thesis. The focus proposed here concerns dependency between approval and preservation: a contribution rejected by one project may remain available to other readers, while the expectation of reconsideration may change investment in new formulations. Memory policy and the formation of the searchable universe itself are thus articulated together.
- **In relation to costly verification and forms of remuneration (Townsend, 1979):** costs of observing outcomes are part of the economic explanation of contractual relationships. The legibility of contributions, dependencies, and consequences adds the question of which participants enter consideration for receiving resources and how the continuation of support is assessed. This margin may operate through contracts but also through reciprocal relationships sustained by prospects of future cooperation.
- **In relation to legibility as a project of power (Scott, 1998):** reading costs and institutional control may jointly shape records and determine who can interpret them. This essay's economic focus adds the question of when preserving before selecting becomes viable; implementation requires examining whether participants can sustain different readings or remain dependent on a representation and access controlled by a single authority.
- **In relation to Rational Inattention and Endogenous Consideration Sets (Sims, 2003; Caplin, Dean & Leahy, 2019; Eliaz & Spiegler, 2011):** this literature endogenizes information acquisition and consideration-set formation in response to costs, beliefs, and incentives. The Economics of Legibility addresses an earlier stage: whether an alternative is formulated and preserved as an object to which attention can be directed, before attention is allocated to evaluating it for a particular demand. The terminological proximity is therefore deliberate: the difference is not denying that the consideration set can be endogenous, but shifting analysis to the structurally prior stage determining which objects can enter that set.
- **In relation to Talent Misallocation and Lost Inventors (Bell et al., 2019):** the proposed mechanism focuses on the externalization of available capabilities following a change in reading channels. This margin can interact with exposure to innovation, education, networks, and expectations of opportunity. Empirical identification must distinguish these influences rather than assume they are fully separable.
- **In relation to Archival Theory and Appraisal (Schellenberg, 1956):** selection, preservation, and subsequent uses are problems shared with archival appraisal. This essay examines their consequences for forming economic alternatives, including unfinished formulations and possible uses by other readers. This focus depends neither on restricting archives to historical memory nor on denying their administrative and decisional uses.
- **In relation to the Economics of Information Systems (Malone, Yates & Benjamin, 1987; Bakos, 1997):** the proposed focus distinguishes cheaper coordination over existing participants and offerings from changed conditions under which contributions become preservable and retrievable. These effects can coexist in electronic markets, firms, and networks. The additional question is the extent to which a formulation's availability can precede its author's hiring and survive rejection of a particular use.

### 4.4. Participation through contributions and the division of labor

By allowing a contribution to remain available before its author is hired, Regime B expands the possible unit of economic participation. A formulated problem, observation, partial solution, or contextual judgment can enter coordination without requiring the same person to execute and finance the entire project. The division of labor may also occur among those who notice a need, formulate a hypothesis, develop a solution, verify its effects, and recognize a consequence.

This possibility extends to people outside usual recruitment and research boundaries: professionals without a first opportunity, people out of employment, and experienced workers whose capabilities are poorly represented by positional filters. The proposition does not assume value in every contribution from these groups; it argues that exclusion from a position should not be taken as sufficient evidence about each formulation they can offer. Contribution records may open an additional path to participation and discovery of capabilities.

Consider a maintenance technician who identifies recurring equipment failures and, assisted by an AI agent, formulates a contextualized observation and hypothesis about their origin. A researcher may recognize the observation's relevance and propose an experiment; a manufacturer may provide equipment to test it. The observation is already a potential contribution before this team forms; its validity and use depend on subsequent stages. The mechanism also allows the participant to originate demand and its relevance criteria, rather than merely answer institutionally defined problems.

The economic consequence is the possibility of coordinating complementary contributions without requiring prior incorporation of all their authors into a single organization. This may expand relationships among people, firms, and research institutions. Informational viability does not itself resolve remuneration distribution, participation costs, or responsibility for execution; it makes possible arrangements that must answer those questions.

This separation distinguishes two institutional boundaries. The firm's boundary delimits activities under its direction; the legibility boundary delimits contributions retrievable for its decisions. The latter can expand without the former following. The organization can mobilize problem formulations, partial solutions, and capabilities from authors it did not first have to hire. The economic consequence of openness must therefore also be examined in the composition of contributions used and cooperative relationships, not merely in job counts or firm size.

### 4.5. Structural Limits of Legibility

Reversing the order of the operators acts directly on the informational frontier without abolishing physical scarcity or the need for institutional commitments. It can, however, change the conditions under which these resources are mobilized. The soundness of an economic thesis is measured by the clarity of its boundaries; five structural limits must therefore be made explicit:

1. **Material and allocative scarcity remains intact:** Focused human attention, synchronous time, financial capital, and physical resources remain strictly scarce. Reducing $c_L$ does not automatically multiply budgets or, by itself, create positions on the organizational chart. Under Regime B, what ceases to depend on a prior choice is the preservation of a contribution as a possibility for consideration. The attention and resources required for each reading remain limited; not considering a contribution on one occasion does not require discarding its record.
2. **Allocative choice and selection under uncertainty remain sovereign:** The decision to allocate capital, hire, or enter into contracts under uncertainty remains necessary and sovereign, with the fiduciary and contractual responsibilities specific to each arrangement.
3. **The firm and hierarchy retain their legitimate domain:** Decisions that are costly to reverse still require commitments and responsibilities whose organization may justify the firm and hierarchy. Opening up reading does not, by itself, determine the form of governance of execution.
4. **Representation does not exhaust capability:** Records of observable manifestations can make a competence discoverable without fully codifying its holder's practical knowledge. Mobilizing that knowledge may still require cooperation and situated presence (Polanyi, 1966).
5. **Expanding $U_L$ does not presuppose its coincidence with $U$:** Even with reduced computational costs, private information, strategic incentives to withhold, and limits to retrieval and interpretation can keep potentially externalizable contributions outside the legible horizon.

### 4.6. Legibility, Resource Mobilization, and Reciprocity

The frontier of legibility conditions not only what can be chosen, but also what can receive the means needed for its realization. Requiring advance payment may reflect liquidity needs and, at the same time, the difficulty of evaluating a contribution, following its development, and recognizing the results that might sustain continued support. When these relationships become legible at a viable cost, another possibility for coordination opens up: advancing resources in view of prospective future reciprocity, including for participants without capital or prior institutional admission.

Consider a computing infrastructure provider that makes initial credits available to develop a contribution. No advance negotiation of a percentage of revenue is presupposed. The participant who achieves results subsequently determines the value they recognize and share; the provider, in turn, evaluates whether the return received and the prospects for cooperation justify expanding, maintaining, or ending new allocations. Sovereignty over attribution and sovereignty over continuation remain with distinct agents. Past recognition does not, by itself, create a right to future resources, just as the initial allocation does not automatically determine a share in the result.

The sustainability of this circuit depends on the expected return from the relationship, considered across the set of initiatives supported, compensating for the resources mobilized, the waiting, the losses, and the costs of follow-up. The directly supported contribution may also enable complementary work, making a comparison limited to a single participant's payment insufficient. It is nevertheless necessary to distinguish the benefit produced for the network from the return that actually sustains the provider: the existence of the former does not guarantee the realization of the latter.

Legibility bears precisely on this evaluation. Records of formulation, dependency, and consequences can help distinguish a lack of results from limited recognition in the presence of significant results, without turning declarations into automatic proof of value or reciprocity. Allocating resources in stages allows exposure to be limited while evidence is produced; its time frames, however, also select, since short windows favor activities capable of rapid monetization. The absence of a track record is not equivalent to an absence of capability and can justify small-scale exploratory allocations.

This discipline can operate in both directions. Someone who recognizes little may lose future support; a provider whose expectations make cooperation disadvantageous may lose participants and the results that would help sustain its infrastructure. The legibility of contributions and conditions for continuation can favor the comparison and contestation of disproportionate distributions. It does not, however, determine a universal measure of fairness: the strength of this adjustment also depends on the alternatives available and the relationships of dependency among participants.

Ending support for execution does not require the contribution to disappear. Provided that it remains under viable custody, including in another infrastructure, its record can be reconsidered by different readers and providers. Regime B's central separation is thus preserved: selecting what will continue to receive resources need not determine what will continue to exist as a possibility for coordination. Extending this mechanism to other scarce resources depends on their costs of use, reversal, and follow-up; legibility broadens its reach without abolishing these constraints.

---

## 5. The Economics of Legibility in Dialogue with Hayek and Coase

In his seminal essay "The Use of Knowledge in Society" (1945), Friedrich A. Hayek formulated the epistemological challenge of the social sciences: society's economic problem is the effective use of knowledge fragmented and dispersed among millions of autonomous agents, which cannot be concentrated in the mind of a single central planner. The strength of the price mechanism lies in its capacity to operate as a telecommunications system, synthesizing and transmitting signals of relative scarcity without requiring the underlying knowledge to be centralized in advance.

Coase (1937) and the transaction-cost tradition (Williamson, 1979) relate the boundary of the firm to the relative costs of organizing transactions internally or through the market. The focus proposed here concerns the formation of alternatives that may precede hiring and organizational decisions. A contribution can become legible before a contract exists with its author; this expands the possibilities for choice without determining in advance which contractual arrangement will be superior.

The price mechanism coordinates dispersed knowledge brilliantly. In institutions that depend on reading contributions, however, a question precedes allocation: which formulations managed to enter the space in which they can be retrieved and interpreted? The Economics of Legibility reaches into this layer. It models the *ex ante* informational conditions that determine which externalizable contributions can cross the barrier of legibility ($U_L$) to become objects of deliberation. Within this focus, the questions can be articulated as follows:

- **Hayek (1945) asked:** How can dispersed knowledge be coordinated through prices?
- **Coase (1937) asked:** Under what relative costs should coordination take place in the firm or the market?
- **The Economics of Legibility asks, before the choice among contributions:** What institutional and technological conditions determine which portions of human knowledge and contributions come to constitute observable alternatives for coordination?

This underlying causal layer operates at two moments: first, in the conditions determining which potential contributions ($U$) come to be externalized ($U_X$) — under the hypothesis that expectations about available legibility affect the agent's own investment in formulating; second, in the conditions determining how much of what has been externalized remains legible ($U_L$) and available to enter the set of alternatives ($\mathcal{C}$) on which coordination can operate. This dynamic is confined to the domain of legibility-intensive institutions described in Section 1.1, not to the price mechanism in general, which operates through paths distinct from the direct reading of contributions.

This is a matter of structural precedence. There is a radical consequence to this inversion: an allocative mechanism may be exceptionally efficient at selecting the best alternatives among those that survived the reading funnel. Yet it may be choosing perfectly among alternatives that did not constitute the totality of economically relevant options.

When this exclusion prevents economically justified reconsideration, we are not necessarily facing a price failure, an incentive failure, or a failure of rationality. We are facing a failure of legibility. The reading infrastructure truncated the universe beforehand, conditioning the outcome through architectural blindness, even when allocation within the legible set is rational. The allocative system thus operates under an endogenous scarcity that derives not from the finitude of resources, but from the prior invisibility of alternatives that might compete for their allocation. A vast range of technical capabilities, methodological formulations, and practical insights never acquires a price or enters a deliberation, not because it is incompatible with economic logic, but because institutional barriers to reading neutralize it before it can be observed. By making preservation and contextual retrieval cheaper, under the conditions of Section 3.4, it may be possible to expand the very set of contributions on which corporate deliberation, the network of collaborators, or decisions to hire and transact can act.

The plurality of readers also matters. A broad collection subjected, in practice, to a single criterion of visibility can retain records without giving them effective possibilities for reconsideration. Availability for different demands requires viable conditions of access, retrieval, and interpretation. This openness is compatible with distributed custody and limits on access to content; it does not require a central reader to determine the value of all contributions.

At this point, it is necessary to distinguish the portability of records from the reading capacity built on them. Portable collections do not make indexes, search coverage, contextual interpretation, or the capacity to mobilize resources equivalent. A provider can develop an economic advantage in discovering and realizing productive cooperation without requiring exclusivity over its participants' records. Concentration of reading remains possible, but it is not a necessity of distributed preservation: the openness of records allows a plurality of readers, while the viability of different indexes conditions its effective exercise.

---

## 6. Empirical Implications and Research Agenda

The fruitfulness of a conceptual framework in institutional economics is assessed not only by its internal coherence, but by its capacity to generate refutable predictions. The causal topology established in this essay — the chain $U \to U_X \to U_P \to U_L \to \mathcal{C}$, the decomposition $c_L = c_P + c_R + c_I$, and the regime inversion $S \to P \to B$ versus $P \to B \to S$ — organizes five research hypotheses, distinguishing preservation, participation, attribution, the complementarity between collections and agency records, and the mobilization of resources through reciprocity.

### Thesis A (Structural): The Central Hypothesis

*The adoption of AI-accelerated Regime A ($S_{\text{IA}} \to P \to B$) may improve the speed and precision of screening, but keeps preservation subordinate to prior selection. The central hypothesis is that, under the conditions of Section 3.4, Regime B ($P \to B \to S$) increases the retrieval and use of contributions relevant to demands not anticipated at the time of screening, by retaining formulations that Regime A would have discarded.*

The proposed mechanism is the separation between rejection in one decision and the termination of informational availability. More efficient screening can improve the present choice, but what it irretrievably discards remains unavailable for other demands. Broad preservation keeps this possibility for reconsideration open, without presupposing that every particular choice set will be larger or better.

Empirically, the test can compare a prefiltered collection and a broad collection, constituted from the same initial stream of contributions, in the face of later demands unknown at screening. With search technology and the evaluation budget kept comparable, it is possible to measure the retrieval and use of contributions initially rejected, the quality of the solutions found, and the total cost of decisions.

The comparison must allow identification of which records and relationships survived in each collection. Gains produced solely by a better search model do not isolate the effect of the order of the operators. The hypothesis receives specific support when contributions unavailable in the filtered collection are retrieved and effectively used in the broad collection, including by other readers or in other domains, at costs compatible with the observed gain.

A differential prediction makes this comparison more discriminating: given the same initial stream, the advantage of reconsideration should be greater for demands that depart from the criteria of the original screening, when that stream contains relevant contributions excluded by that screening. Criteria for quality, cost, and diversity of demands must be defined before the results are observed. An absence of gains under these conditions, or a gain insufficient to compensate for the additional costs, weakens the hypothesis of economic advantage in the context examined.

### Thesis B (Behavioral): The Hypothesis of Capability Decompression

*When a more credible channel of reading and recognition increases the expected benefit of externalization relative to its costs, the hypothesis is that contributions formulated by previously peripheral agents, and proposals extending beyond the formal role of participants already admitted, will increase.*

The mechanism is a change in the participation condition described in Section 4.2, category (2b). A reduction in the effort of formulation, including through AI support, can complement the change in the channel of reception. The effect must be observed separately on the production of records, their retrieval, their evaluation, and their use: more issuance does not, by itself, demonstrate greater value.

To study the external margin, the observed universe must include people who do not yet belong to the institution. A pilot restricted to employees can identify capabilities beyond formal responsibilities, but does not estimate the participation of those left outside hiring. Research can track first contributions, the diversity of problems formulated, continuity of participation, and subsequent uses, distinguishing the effect of the new channel from changes in participant composition, compensation, or the support offered.

Experimental designs or longitudinal comparisons can examine whether the possibility of reconsideration and recognition changes behavior. The size and distribution of the effect remain empirical questions, including when openness benefits some groups or types of contribution more than others.

It is also necessary to distinguish additional externalization from simple migration between channels. Follow-up must identify new formulations, imported records, and possible reductions in participation in other environments. The question is whether the change makes credible a possibility of consideration capable of altering total investment in formulating, not merely the place where that investment becomes visible.

### Thesis C (Dynamic/Institutional): The Attribution Crisis Hypothesis

*In environments where continued collaboration depends on recognizing contributions, expanding preservation without adequate means of retrieving provenance, dependencies, and consequences can reduce incentives to participate or encourage a return to admission filters.*

The hypothesis has two pathways that must be distinguished. Through attribution, in line with the challenges of team production (Alchian & Demsetz, 1972), participants may reduce their investment when their contributions become difficult to recognize. Through congestion, issuing records can produce private visibility benefits while transferring reading costs to others, generating externalities analogous to those of spam (Rao & Reiley, 2012). Excessive issuance depends on this incentive structure; it does not follow from reducing the cost of issuance alone. Attribution instruments may help without thereby resolving every mechanism of pollution.

Empirical examination must track the time needed to reconstruct dependencies, attribution conflicts, reading costs, continuity of contributions, and any new barriers to entry. Comparisons among similar environments can assess whether these outcomes change with the introduction of records and recognition practices, taking other forms of governance and trust into account. Finding stable collaboration without such records requires examining whether other means of trust and recognition perform this function. The hypothesis loses support in contexts where attribution difficulties persist without the predicted deterioration in participation; its necessity is not presupposed for all coordination.

### Thesis D (Informational): The Hypothesis of Complementarity Between Preservation and Agency Records

*For problems in which intentions participants stand behind, declared dependencies, and recognized consequences are relevant to the decision, the hypothesis is that records of these acts add useful information to the reading of a broad collection, beyond the gain obtained from semantic retrieval alone.*

A comparison can vary two dimensions: a filtered or broadly preserved collection; and the availability or absence of participants' declarations in addition to work artifacts. Keeping reading technology comparable, the design makes it possible to distinguish the effect of preservation from the effect of agency records and investigate their interaction.

Relevant outcomes include reconstructing confirmable relationships, identifying who took responsibility for a formulation, discovering uses across domains, and subsequently using contributions. The costs of producing and verifying records must be counted. Counting declarations of value alone does not establish the economic benefit: what matters is whether the additional context improves observable decisions or cooperation. The design tests a possible extension of Regime B without making any particular protocol a condition for the validity of the structural thesis.

### Thesis E (Institutional): The Hypothesis of Reciprocity-Sustained Provision

*In environments where difficulty observing contributions and consequences restricts the upfront provision of resources, their legibility may expand the set of participants supported and sustain relationships of continued provision, even when the value of subsequent recognition remains at the beneficiary's discretion.*

The mechanism does not presuppose a previously negotiated percentage of compensation. The hypothesis is that the possibility of evaluating results, recognition, and dependencies helps the provider decide on new allocations, while the prospect of continuation becomes part of the participant's incentives. The test must compare arrangements with different access to these records, keeping the initial budget and other conditions of support comparable.

Indicators include the entry of previously unserved participants, continuity of contributions, resources actually consumed, recognition received, effects on related projects, and follow-up costs. A time horizon compatible with the maturation of activities must be observed, distinguishing an absence of results, an absence of records, and limited recognition. More allocations or more declarations do not, on their own, demonstrate sustainability: what matters is verifying whether the returns and productive relationships actually maintained justify the resources mobilized, without depending on increasing, unaccounted-for subsidies.

---

## 7. Implications of Expanding the Space of Legibility

Reversing the order of the operators allows examination of consequences extending beyond the improvement of an isolated decision. When an organization preserves contributions before selecting them, it can broaden the conditions under which its choice set is formed. When this possibility extends to a network of participants, it can also change the conditions under which problems are formulated, resources are mobilized, and activities come to acquire economic viability. Expanding the space of legibility can thus reach both the alternatives available and the objectives around which choice is organized.

The following implications are conditional developments of the proposed mechanism. Their realization depends on the viability of reading, incentives to participate, and arrangements that allow contributions to be preserved, interpreted, and recognized. Cost reductions can support this transformation; adopting Regime B also involves institutional choices about who can formulate, what remains available, and who can reconsider it.

### 7.1. The Formation of Demand and of Economic Problems Themselves

A need that lacks conditions for formulation and reception can remain economically invisible. The same applies to an objection that does not fit customer-service channels, an application the producer did not anticipate, or a condition of use that makes every available offering inadequate. Under these circumstances, the absence of a purchase tells us little about possibilities for coordination that never even came to be constituted.

Regime B allows these formulations to be preserved before a corresponding offering exists. One person describes a problem; another adds a condition; a third identifies a partial solution; other participants recognize its relevance or offer resources. Over this process, both the solution conceived and the understanding of what deserves to be solved can change. Demand itself becomes an object of distributed elaboration, developing the possibility of formulating and revising $q$ presented in Section 3.3.

An expression of interest does not, by itself, constitute solvent demand. The transition to economic viability requires reconciling willingness and ability to contribute with the resources needed for realization. The proposed gain lies in making these conditions and their possible complements legible. An activity previously unviable for each participant in isolation may become realizable when needs, capabilities, and commitments find a way to be combined. This combination can sustain market demand or make an activity possible through cooperation, without requiring a purchase. Formulating a problem, expressing interest, and constituting solvent demand are therefore distinct stages. Part of the resulting demand may come into existence because the infrastructure allowed it to form.

### 7.2. Marketing, Sales, and the Direction of Reading

This transformation reaches the economic role of marketing and sales. In arrangements where the offering is defined in advance and commercial activity concentrates on ensuring its sale, knowing the recipient may serve primarily to identify the conditions under which they will be persuaded to buy. Reading can be sophisticated and still afford the consumer little participation in defining what is offered.

Expanding the space of legibility opens another possibility: formulated needs, objections, conditions of use, and recognized consequences begin informing the design of the offering. The economic distinction lies in the direction of this influence. On one side, the aim is to adjust the recipient's willingness to the established offering; on the other, the criteria they can express and revise participate in the formation of alternatives. The same reduction in inference costs can serve both dynamics; its use does not, by itself, determine a change of regime.

Where commercial advantage depends on information asymmetries, difficulty comparing alternatives, or control of the channels of consideration, this change may reduce the effectiveness and return of persuasion. Discovering applications, communicating possibilities, and building trust remain valuable. The point of tension concerns the profitability of inducing choices that would lose their appeal under broader conditions of reading. This criterion dispenses with the figure of an authorized interpreter of "true" needs: participants themselves must be able to make explicit, confront, and revise what they consider desirable.

### 7.3. Economic Participation Beyond Available Positions

When a contribution can become legible before its author is hired, access to participation ceases to depend exclusively on previously occupying a position. As developed in Section 4.4, the unit of participation can be smaller than a job and cross more than one organization. This path may reach those who have not yet obtained an opportunity, those who have lost their employment relationship, and those whose experience is poorly represented by qualifications or job descriptions. The ability to say why something deserves to be done can enter coordination before the ability to do it.

Externalizing these contributions remains costly, and their consideration assures neither use nor compensation. For participation to be sustained, authors must find reasons to continue contributing. An arrangement that expands the extraction of formulations while keeping the consequences of their use opaque can undermine this condition. Expanding legibility makes it relevant to examine who benefits, how they recognize that benefit, and what opportunities they return to the network.

### 7.4. Advancing Resources and Retrospective Recognition

The legibility of contributions and dependencies can change the conditions under which resources are advanced, following the mechanism developed in Section 4.6. A provider can support an activity in expectation of subsequent benefits, considering its concrete formulation, the contributions already available, and the consequences recognized in previous interactions, in addition to the proponent's general characteristics.

This expectation can support predefined contracts or relationships of reciprocity in which later recognition informs the continuation of support. In the second situation, the resource provider assumes uncertainty about the return, while the beneficiary makes observable a decision that may influence future cooperation. Repeated relationships and the possibility of interpreting the history become part of the economic calculation. The risk of opportunism remains and must be weighed against the costs and benefits of this form of coordination.

There is also a complementarity effect: resources advanced to one activity may enable contributions useful to others. Reading these relationships makes it possible to assess a reach that observing an isolated transaction would miss. It also makes it possible to recognize shared components, methods, and earlier formulations whose effects are distributed among many beneficiaries. These benefits do not, however, guarantee the return sustaining whoever advanced the resources.

Genealogy supplies elements for this recognition without automatically establishing a measure of value or an obligation to distribute it. A declaration of recognition, an actual transfer, and a contractual obligation remain distinct. Money, contracts, and budgets remain instruments of coordination; the change concerns the information available for deciding when and under what conditions to mobilize them.

### 7.5. The Boundary of the Firm and Authority over Formulation

An organization can benefit from contributions it did not commission and whose authors it did not incorporate into its organizational chart. When these contributions remain retrievable and interpretable, the boundary of mobilizable knowledge can extend beyond the boundary of previously contracted relationships. Suppliers, former collaborators, users, and participants with no previous ties can become relevant through what they formulate.

In this arrangement, preservation and consideration can precede hiring, financing, and productive integration. The authority to commit resources remains wherever the institutional arrangement places it. Its continuation, however, no longer automatically justifies a monopoly over the formulation of problems and alternatives. The decision to execute can remain hierarchical while the formation of its possibilities becomes distributed. The deepest consequence is the opening of space to question the very objectives the organization presents as given.

### 7.6. Economic Power, Intermediation, and Resistance to Change

If reading infrastructure conditions the choice set, controlling it confers power over what comes to be considered. An effective expansion of legibility can weaken advantages based on exclusive access, dependence on credentials, or the ability to prevent external contributions from acquiring institutional relevance. The resulting coordination gains can coexist with losses for agents who benefited from these restrictions. Gains for participants in a network do not, by themselves, demonstrate an increase in aggregate welfare; the assessment must include costs and effects on others.

Resistance to Regime B may therefore be economically rational for certain participants, even when the change expands possibilities for the network. This does not allow every restriction to be reduced to a strategy of capture: confidentiality, security, and verification costs impose concrete requirements. It does, however, allow these requirements to be distinguished from the preservation of positions whose advantage depends on the narrowness of the space of consideration.

New intermediaries can also concentrate power. Broad preservation offers limited results when a single service controls discovery, interpretation, or conditions of access to records. Concentration alone is not enough to characterize a return to Regime A: an intermediary can offer real reading gains. What matters is examining whether control of one channel terminates availability for other readings or whether alternatives for reconsideration remain viable.

The plurality of readers, the possibility of reconsideration by other participants, and the continuity of records outside a particular relationship become economic dimensions of the architecture. A person may keep their contribution available and still depend on a dominant channel for it to produce effects. The distribution of gains depends, in part, on how contestable this control remains.

### 7.7. The Recursive Expansion of the Space of Possibilities

The effects of broader adoption of Regime B are not limited to the sum of local improvements in reading. A contribution considered can prompt a question not yet formulated; that question can mobilize another person's experience; their response can reveal an unforeseen application. The path between the potential universe and the choice set reopens with every interaction. What becomes mobilizable includes capabilities that are externalized only upon encountering a relevant formulation.

There is therefore a possibility of feedback among legibility, participation, and the formation of new alternatives. Its reach depends on the relevance and complementarity of contributions, reading costs, and the credibility of relationships. Increasing the number of records alone can produce congestion. The economically relevant result is the formation of usable possibilities that would not emerge under the previous conditions.

Investigating these implications requires observing whether new participants can formulate problems, whether these formulations change decisions, and whether the resulting benefits sustain new contributions. Retrieval of earlier records must be distinguished from the production of new formulations and from a simple increase in issuance. The question that opens is distributive and institutional: who can participate in shaping the world over which others make decisions? Expanding the space of legibility may shift this boundary and, with it, the dispute over what comes to exist as an economic possibility.

---

## 8. Conclusion: The Emancipation of Choice

Funnels, rankings, committees, curation algorithms — throughout this essay, these instruments have been revisited to make one of their economic rationales explicit: a rational response to scarce reading capacity, used by organizations and individuals faced with the need to act. Under historical reading costs, the rationality of this response generalized the simplifying perception that this coupling followed from the very nature of scarcity, thus confusing a technological contingency with a structural law.

Once the simplifications that confused the founding rationale of these instruments are overcome, an unsettling obviousness emerges: the boundary of what a decision-maker can consider is not a property of the world; it is a property of their reading infrastructure. Making this obviousness explicit has consequences with profound implications.

The first is retrospective. If the choice set has always been endogenous to the regime of legibility, then the dynamics of economic decisions contain a dimension that economic analysis has tended to leave in the background: the dimension of what never came to be considered, not for lack of merit, but for lack of reading. Potentially externalizable capabilities that may never have been formulated because no channel made their externalization economically relevant; contributions that were produced and whose traces dissipated before an external demand could reveal them; solutions that existed in relational form and were compressed into summaries preserving an appearance of completeness while discarding the texture that made them useful. The rationality of the act of choice, in these cases, may have been impeccable — but it was exercised over a fraction presented as the whole.

The second consequence is prospective. The cumulative reduction in the components of reading cost — with storage making preservation cheaper ($c_P$), semantic search expanding retrieval ($c_R$), and, decisively, advances in neural architectures reducing the cost of contextual inference ($c_I$) — offers economics more than an incremental efficiency gain within the known perimeter. By relaxing the interpretive constraint that keeps the operators limited to the historical sequence of selecting before preserving, it opens the possibility of expanding the perimeter itself — both for institutions that hire, finance, and research, and for individuals navigating knowledge ecosystems. The issue is not choosing better among the same alternatives, but coming to see alternatives that the previous regime made structurally invisible. The difference between optimizing search within a given space and changing the composition of the space over which one searches is precisely the distinction this essay seeks to make analytically visible. As developed in Section 7, this expansion may also reach the formulation of the problems and objectives around which choice is organized.

The third consequence concerns decision-making and institutions, and is also the most demanding. Separating reading from selection does not dissolve the need to select — material scarcity, the finitude of human attention, and fiduciary responsibility remain intact. What the separation requires is that the apparatus that previously operated as the sentinel of memory — whether the firm's bureaucratic routine or the overloaded individual's anticipatory filter — be redesigned to operate as an allocation mechanism over a horizon already preserved. This entails challenges that the essay deliberately does not answer: how can quality be verified without restoring the anticipatory funnel? How can provenance, dependencies, and consequences be made legible for plural judgments in distributed processes? How can the credibility of signals be sustained when prior screening no longer supplies it? Each of these questions constitutes a problem of institutional design and engineering whose resolution remains open in this essay, so that different research programs and plural experiments may emerge.

For more than a century, we have built instruments that operated with remarkable sophistication on a fraction of the potential universe of alternatives available for choice. This essay dares to suggest that the next step may not be a mere step in improving current instruments, but a leap toward instruments that change the order in which their functions operate — preserving before selecting, making legible before allocating — and thereby set in motion the expansion of the space of alternatives effectively available to economic choice.

---

## Appendix: Author's Note

This essay constitutes independent theoretical research in institutional economics; the positions expressed in it are the author's and do not constitute financial, legal, or investment advice.

---

## References

- Alchian, A. A., & Demsetz, H. (1972). Production, information costs, and economic organization. *American Economic Review*, *62*(5), 777–795. https://www.jstor.org/stable/1815199
- Arrow, K. J. (1962). Economic welfare and the allocation of resources for invention. In *The rate and direction of inventive activity: Economic and social factors* (pp. 609–626). Princeton University Press. https://doi.org/10.1515/9781400879762-024
- Bakos, J. Y. (1997). Reducing buyer search costs: Implications for electronic marketplaces. *Management Science*, *43*(12), 1676–1692. https://doi.org/10.1287/mnsc.43.12.1676
- Bell, A., Chetty, R., Jaravel, X., Petkova, N., & Van Reenen, J. (2019). Who becomes an inventor in America? The importance of exposure to innovation. *Quarterly Journal of Economics*, *134*(2), 647–713. https://doi.org/10.1093/qje/qjy028
- Benkler, Y. (2002). Coase's penguin, or, Linux and The nature of the firm. *Yale Law Journal*, *112*(3), 369–446. https://doi.org/10.2307/1562247
- Caplin, A., Dean, M., & Leahy, J. (2019). Rational inattention, optimal consideration sets, and stochastic choice. *Review of Economic Studies*, *86*(3), 1061–1094. https://doi.org/10.1093/restud/rdy037
- Coase, R. H. (1937). The nature of the firm. *Economica*, *4*(16), 386–405. https://doi.org/10.1111/j.1468-0335.1937.tb00002.x
- Coate, S., & Loury, G. C. (1993). Will affirmative-action policies eliminate negative stereotypes? *American Economic Review*, *83*(5), 1220–1240. https://www.jstor.org/stable/2117560
- Dixit, A. K., & Pindyck, R. S. (1994). *Investment under uncertainty*. Princeton University Press. https://doi.org/10.1515/9781400830176
- Echenique, F., & Li, A. (2025). Rationally inattentive statistical discrimination: Arrow meets Phelps. *Journal of the European Economic Association*, *23*(5), 1712–1742. https://doi.org/10.1093/jeea/jvaf002
- Eliaz, K., & Spiegler, R. (2011). Consideration sets and competitive marketing. *Review of Economic Studies*, *78*(1), 235–262. https://doi.org/10.1093/restud/rdq016
- Gale, D., & Shapley, L. S. (1962). College admissions and the stability of marriage. *American Mathematical Monthly*, *69*(1), 9–15. https://doi.org/10.1080/00029890.1962.11989827
- Hayek, F. A. (1945). The use of knowledge in society. *American Economic Review*, *35*(4), 519–530. https://www.jstor.org/stable/1809376
- Hirsch, F. (1976). *Social limits to growth*. Harvard University Press. https://doi.org/10.4159/harvard.9780674497900
- Keller, J. R., & Dlugos, K. (2023). Advance 'em to attract 'em: How promotions influence applications in internal talent markets. *Academy of Management Journal*, *66*(6), 1831–1859. https://doi.org/10.5465/amj.2021.1174
- Knight, F. H. (1921). *Risk, uncertainty and profit*. Houghton Mifflin.
- Levitt, B., & March, J. G. (1988). Organizational learning. *Annual Review of Sociology*, *14*, 319–340. https://doi.org/10.1146/annurev.so.14.080188.001535
- Malone, T. W., Yates, J., & Benjamin, R. I. (1987). Electronic markets and electronic hierarchies. *Communications of the ACM*, *30*(6), 484–497. https://doi.org/10.1145/214762.214766
- March, J. G. (1991). Exploration and exploitation in organizational learning. *Organization Science*, *2*(1), 71–87. https://doi.org/10.1287/orsc.2.1.71
- Mortensen, D. T., & Pissarides, C. A. (1994). Job creation and job destruction in the theory of unemployment. *Review of Economic Studies*, *61*(3), 397–415. https://doi.org/10.2307/2297896
- Nelson, R. R. (1961). Uncertainty, learning, and the economics of parallel research and development efforts. *Review of Economics and Statistics*, *43*(4), 351–364. https://doi.org/10.2307/1927475
- Polanyi, M. (1966). *The tacit dimension*. Doubleday & Company.
- Rao, J. M., & Reiley, D. H. (2012). The economics of spam. *Journal of Economic Perspectives*, *26*(3), 87–110. https://doi.org/10.1257/jep.26.3.87
- Raymond, E. S. (1999). *The cathedral and the bazaar: Musings on Linux and Open Source by an accidental revolutionary*. O'Reilly Media.
- Romer, P. M. (1990). Endogenous technological change. *Journal of Political Economy*, *98*(5, Part 2), S71–S102. https://doi.org/10.1086/261725
- Rothschild, M. (1973). Models of market organization with imperfect information: A survey. *Journal of Political Economy*, *81*(6), 1283–1308. https://doi.org/10.1086/260128
- Savage, L. J. (1951). The theory of statistical decision. *Journal of the American Statistical Association*, *46*(253), 55–67. https://doi.org/10.1080/01621459.1951.10500768
- Schellenberg, T. R. (1956). *Modern archives: Principles and techniques*. University of Chicago Press.
- Scott, J. C. (1998). *Seeing like a state: How certain schemes to improve the human condition have failed*. Yale University Press.
- Sims, C. A. (2003). Implications of rational inattention. *Journal of Monetary Economics*, *50*(3), 665–690. https://doi.org/10.1016/S0304-3932(03)00029-1
- Spence, M. (1973). Job market signaling. *Quarterly Journal of Economics*, *87*(3), 355–374. https://doi.org/10.2307/1882010
- Stigler, G. J. (1961). The economics of information. *Journal of Political Economy*, *69*(3), 213–225. https://doi.org/10.1086/258464
- Townsend, R. M. (1979). Optimal contracts and competitive markets with costly state verification. *Journal of Economic Theory*, *21*(2), 265–293. https://doi.org/10.1016/0022-0531(79)90031-0
- Williamson, O. E. (1979). Transaction-cost economics: The governance of contractual relations. *Journal of Law and Economics*, *22*(2), 233–261. https://doi.org/10.1086/466942