PRIORITY FILING - INITIAL DRAFT
A fully revised and formatted version will follow.
Gregory Caldwell Beier
Susarb LLC, Cambridge, Massachusetts
March 19, 2026
From the Prolegomena to the Unification of Economics, Computer Science, Mathematics, and Physics by Gregory Caldwell Beier (forthcoming 2026).
© 2026 Gregory Caldwell Beier. Licensed under CC BY-NC-ND 4.0.
The companion paper (Beier, 2026b, Section 3.4) identifies an improvement mechanism within the Disclosure Incentive: the act of structured disclosure improves the quality of the discloser's judgment over time. This paper develops a stronger claim. The Disclosure Incentive does not merely improve existing judgment or capture judgment that already exists in latent form. In many cases, the requirement to articulate structured reasoning about a constraint environment creates reasoning that was not previously occurring at all. A fund manager who has never assessed how a portfolio company's water usage interacts with aquifer depletion in the company's watershed is not withholding a pre-existing assessment. The assessment does not exist. When the Disclosure Incentive requires the manager to produce a Scaling field - an explicit statement of how the company's activity interacts with shared constraints at aggregate scale - the requirement forces into existence an analytical process that was absent, not hidden. The disclosure is performative in the sense Austin (1962) defined: it does not describe a pre-existing state of affairs but brings a new state of affairs into existence. Callon (1998) applied the concept of performativity to economics itself, arguing that economic models do not merely describe markets but actively shape them. The present paper applies performativity to the Disclosure Incentive specifically: the disclosure format does not merely record constraint reasoning but generates it, because the structured fields require analytical work that the absence of the format does not prompt. The economic implications are substantial. If the Disclosure Incentive only captured existing judgment, its value would be bounded by the stock of judgment currently produced. If it generates new judgment - reasoning about constraint interactions that no agent was previously performing - its value includes the entire yield of that new reasoning, which may exceed the value of the captured stock by a large margin.
Keywords: performativity, consideration generation, disclosure, improvement mechanism, constraint reasoning, Austin, Callon, deliberate practice, calibration, judgment creation
JEL Codes: D82, D83, D91, B41, Q01
The economics of information has generally treated information as something that exists and is either revealed or concealed, acquired or ignored, processed or neglected. Stiglitz's research program analyzes the consequences of information that some agents possess and others do not. The Grossman-Stiglitz paradox concerns whether agents will pay to acquire information that exists in the world but is costly to observe. The voluntary disclosure literature (Milgrom, 1981; Verrecchia, 1983; Dye, 1985) analyzes the conditions under which agents who possess information choose to reveal it.
In each of these frameworks, the information is there. The analytical question is about its flow - who has it, who wants it, what determines whether it moves from one to the other. The present paper addresses a different question: what happens when the information does not exist, and when the mechanism designed to capture it instead creates it?
The distinction is between capture and generation. The companion paper on judgment economics (Beier, 2026ar) treats judgment as an abundant resource that the price system discards - the PCR analogy, where the DNA is present in every sample but no mechanism amplifies it. That framing is correct for the judgment agents already produce in the course of their ordinary activities. A fund manager assessing a company's earnings trajectory produces judgment that the price system compresses away. The judgment exists; the architecture fails to preserve it.
But the Scaling field introduces something different. When a fund manager is required to assess how the company's water usage interacts with aquifer depletion across the watershed at aggregate scale over a ten-year horizon, the manager is not being asked to disclose a pre-existing assessment. In the vast majority of cases, no such assessment exists. The manager has never performed this analysis because nothing in the current information architecture required it, incentivized it, or provided a format for structuring it. The Disclosure Incentive does not capture this assessment. It creates it - by requiring its articulation in a specific format, scored against outcomes, accumulated into a track record.
This is the performative property. The disclosure does not describe what the manager already thinks about aquifer dynamics. It forces the manager to think about aquifer dynamics - possibly for the first time - and the resulting reasoning is a new cognitive product that would not exist without the disclosure requirement.
Section 2 develops the concept of performativity as applied to disclosure. Section 3 distinguishes generation from capture. Section 4 analyzes the economics of generated reasoning. Section 5 examines the conditions under which generation occurs. Section 6 connects to the improvement mechanism. Section 7 states limitations. Section 8 concludes.
Austin (1962) distinguished performative utterances from constative ones. A constative utterance describes a state of affairs ("the bridge is structurally sound"). A performative utterance creates a state of affairs ("I hereby commit to disclosing my assessment of this bridge's resilience under projected load conditions for the next ten years"). The performative does not report a pre-existing commitment; the act of utterance constitutes the commitment.
Callon (1998) extended performativity to economics, arguing that economic theory does not merely describe markets but actively formats them - the Black-Scholes model did not discover option prices but provided the framework through which market participants calculated and negotiated them, and the prices that resulted conformed to the model because the model shaped the behavior that produced them. MacKenzie (2006) developed this further with the concept of "an engine, not a camera" - economic models as devices that reshape the reality they purport to describe.
The Disclosure Incentive is performative in a more direct sense than either Austin's speech acts or Callon's model-market interaction. The five-field disclosure format is a template that requires the discloser to produce specific types of reasoning - Input (what evidence was consulted), Method (what analytical framework was applied), Output (what conclusion was reached), Identity (who produced the assessment), and Scaling (what happens when this activity is replicated at aggregate scale). Each field is a cognitive demand: fill this field requires doing the analytical work the field specifies. The format does not record reasoning that was already happening. For fields where the agent had not previously performed the relevant analysis - particularly Scaling, which requires constraint interaction assessment that individual-transaction analysis does not demand - the format generates the reasoning by requiring it.
The performative mechanism operates through the interaction between format and cognition. A blank field labeled "Scaling: What happens when this activity is replicated by N agents over T periods, and how do the interaction effects differ from the individual effects?" is not a passive recording device. It is a prompt that directs cognitive effort toward a specific analytical task. The prompt's effectiveness depends on the agent's domain expertise (an analyst without hydrological knowledge cannot produce a meaningful aquifer Scaling assessment), but for agents with the relevant expertise, the prompt activates analytical work that would not have occurred without it. The expertise was present; the analysis was not. The disclosure requirement converts latent expertise into active reasoning.
The distinction between generation and capture can be stated precisely. Let R_existing denote the set of constraint reasoning an agent has already performed in the course of their ordinary professional activities - the earnings models, risk assessments, and competitive analyses that the agent produces whether or not a disclosure mechanism exists. Let R_generated denote the constraint reasoning the agent produces only because the disclosure format requires it - the Scaling assessment, the explicit confidence calibration, the articulated interaction model that the agent's ordinary workflow does not demand.
The companion paper on judgment economics (Beier, 2026ar) addresses R_existing. The PCR analogy applies: this reasoning is already produced but not captured. The Disclosure Incentive preserves it.
The present paper addresses R_generated. No analogy to amplification applies, because there is nothing to amplify. R_generated is new cognitive work prompted by the disclosure format, and its economic value is the value of the decisions it informs minus the cognitive cost of producing it.
The magnitude of R_generated relative to R_existing varies by field. For the Input, Method, and Output fields, most of the relevant reasoning is already part of the agent's ordinary analytical workflow. Even in these fields, however, the transition from informal heuristic to formal, scored disclosure generates new rigor: an analyst who informally uses a model is forced to specify it precisely, and an analyst with a gut feeling about an outcome is forced to assign a hard probability. The act of formalizing informal thought is itself partially performative across all five fields, even if the Scaling field is the most purely generative. An analyst who covers a company already consults data (Input), applies analytical methods (Method), and reaches conclusions (Output). The disclosure captures rather than generates these.
For the Identity field, the disclosure requirement generates accountability rather than reasoning - the agent already knows who they are, but the public attribution creates a reputational stake that did not previously exist.
For the Scaling field, the balance shifts toward generation. Individual-transaction analysis does not require assessing interaction effects at aggregate scale. A fund manager evaluating whether to invest in a water-intensive agricultural company assesses the company's profitability at the scale of one firm. The question of what happens when five similar firms in the same watershed simultaneously increase extraction at similar rates - the Scaling question - is not part of the standard investment analysis workflow. For most analysts in most domains, the Scaling assessment is R_generated: new reasoning created by the disclosure requirement.
The companion paper on scaling (Beier, 2026ag) establishes that scaling information is the type most consequential for long-horizon coordination on shared constraints and the type most completely absent from the current information architecture. The performative property of the disclosure format is therefore most valuable where the information gap is largest - the format generates precisely the reasoning the system most needs.
Generated reasoning has different economics from captured reasoning.
Captured reasoning has zero production cost at the margin - the reasoning already exists, and the only cost is the disclosure effort (formatting, timestamping, publishing). The value of capture is the allocation improvement that results from making the reasoning visible to other agents.
Generated reasoning has positive production cost - the agent must perform analytical work they would not otherwise have done. The cost includes the time spent on the analysis, the domain expertise required (which may need to be acquired or contracted), and the cognitive effort of reasoning about constraint interactions that the agent's ordinary workflow does not address. The value of generation is the allocation improvement from making new reasoning available, plus the improvement in the agent's own judgment from having performed the analysis (the deliberate practice effect from Beier, 2026b, Section 3.4), plus the contribution to the disclosure network's combinatorial information growth (Beier, 2026ag, the factorial coordination paper).
The net value of generation exceeds the net value of capture when three conditions hold: the generated reasoning addresses a constraint interaction that was previously invisible to all market participants (high marginal information value), the agent has domain expertise sufficient to produce a well-calibrated assessment (the reasoning quality justifies the production cost), and the constraint environment is one where the missing information has large allocation consequences (the decisions the reasoning informs involve substantial capital at stake).
These conditions hold most strongly for the Scaling field in constraint-intensive sectors: energy (atmospheric carbon interactions), agriculture (aquifer interactions), infrastructure (capacity interactions), and natural resource management (depletion interactions). In these domains, the Scaling assessment is almost entirely R_generated, the marginal information value is high because no existing mechanism produces the information, and the allocation consequences are enormous because misallocation of capital in these sectors produces the reality debt documented in the companion paper (Beier, 2026af).
Not every disclosure prompt generates reasoning. Three conditions must hold for the performative mechanism to operate.
The first is domain expertise. The disclosure format can prompt an analyst with hydrological training to reason about aquifer interactions, because the analyst possesses the knowledge required to produce the reasoning but had not previously been asked to apply it in this specific way. The same format cannot prompt an analyst without hydrological training to produce meaningful aquifer reasoning - the expertise is absent, not merely latent. The performative mechanism converts latent expertise into active reasoning; it does not create expertise from nothing.
The second is format specificity. A general prompt ("disclose your views on sustainability") generates vague, unstructured responses that may or may not contain new reasoning. A specific prompt ("state your assessment of how aggregate extraction by N operators in this watershed interacts with recharge rates over T years, with confidence levels") directs cognitive effort toward a specific analytical task and is far more likely to generate reasoning the analyst had not previously performed. The five-field structure's specificity is not merely organizational convenience - it is the mechanism through which the format activates latent analytical capacity.
The third is accountability through scoring. An unscored disclosure prompt generates reasoning of uncertain quality - the analyst might produce a vague Scaling assessment to satisfy the format requirement without performing genuine analysis. Proper scoring against outcomes creates accountability: the Scaling assessment will be compared to what actually happens, and the comparison will be public. The accountability motivates genuine analytical effort rather than performative compliance, because a poorly calibrated Scaling assessment damages the track record that the agent is trying to build.
When all three conditions hold - domain expertise, format specificity, and scoring accountability - the disclosure requirement reliably generates new constraint reasoning. When any condition fails, the generation degrades: without expertise, the reasoning is empty; without specificity, the reasoning is vague; without accountability, the reasoning is shallow.
The improvement mechanism in the companion paper (Beier, 2026b, Section 3.4) operates through the cycle of articulate-publish-score-learn-repeat. Agents who receive calibration feedback improve measurably (Lichtenstein and Fischhoff, 1980). Ericsson, Krampe, and Tesch-Romer (1993) identified the conditions for deliberate practice: structured task, unambiguous feedback, sufficient iteration.
The performative property adds a dimension to this cycle that the companion paper did not develop. The first iteration of the cycle does not merely improve existing judgment - it generates the judgment for the first time. The fund manager who produces a Scaling assessment of aquifer interactions in cycle one is not improving an existing assessment. They are creating the assessment from scratch, prompted by the format. In cycle two, when the scoring feedback arrives and the manager sees how their assessment compared to outcomes, the improvement mechanism begins - but it begins from a baseline that the disclosure format created, not from a baseline that existed independently.
The implication is that the Disclosure Incentive's value in its first cycle is different in kind from its value in subsequent cycles. The first cycle generates reasoning that did not exist. Subsequent cycles improve that reasoning through calibration feedback. Generation is a one-time event for each constraint domain - once the analyst has produced the first Scaling assessment, the reasoning exists and subsequent disclosures are updates rather than creations. But as the analyst expands to new constraint domains, each new domain's first Scaling assessment is another act of generation.
Feedback improves judgment only in high-validity environments, as Kahneman and Klein (2009) argue: improves judgment only in high-validity environments. The performative property operates upstream of their condition: it generates the judgment that the high-validity environment then improves. Without the generation, there is no judgment for the feedback to act on. The Disclosure Incentive engineers both the generation (through the format requirement) and the improvement (through the scoring feedback), producing a pipeline from zero reasoning to good reasoning through the intermediate stage of imperfect-but-scored reasoning.
The paper's central claim - that the disclosure format generates reasoning rather than merely capturing it - is argued by mechanism rather than demonstrated empirically. A controlled experiment comparing the constraint reasoning of analysts who are given the five-field format with the reasoning of analysts who are not (holding domain expertise constant) would provide the empirical test. The prediction is that the format-prompted group produces Scaling assessments that the unprompted group does not - not because the unprompted group is withholding assessments but because the assessments were never produced. This experiment has not been conducted.
The boundary between generation and capture is not always sharp. An analyst who has informally thought about aquifer interactions but never formalized the assessment occupies an intermediate position - the disclosure format captures the informal reasoning while also generating a formal structure that the informal thinking lacked. Whether this is "capture" or "generation" depends on how much analytical work the formalization requires, which is a matter of degree rather than kind.
The paper focuses on the Scaling field as the primary site of generation, but the performative mechanism may also operate for the Confidence field. Many analysts have never articulated explicit probability estimates for their assessments - they hold views with informal strength ("I think this is likely" rather than "I assess a 70% probability"). The requirement to state an explicit confidence level generates a quantification that did not previously exist, and the quantification is itself a form of new reasoning that the format creates.
The claim that generated reasoning has higher marginal value than captured reasoning depends on the assumption that the constraint interactions the Scaling field addresses are important for allocation decisions. If constraint interactions are second-order relative to individual-transaction effects for most capital allocation decisions, the value of generated Scaling reasoning is correspondingly smaller. The companion papers argue that constraint interactions are first-order for long-horizon allocation in constraint-intensive sectors, but this is an empirical claim that the literature has not yet settled.
The Disclosure Incentive's most underappreciated property is that it does not merely preserve reasoning the economy already produces. For the Scaling field in particular - the assessment of how activities interact with shared constraints at aggregate scale - the disclosure format generates reasoning that no agent was previously performing, because nothing in the current information architecture required it, incentivized it, or structured it. The format is performative: it brings into existence the analytical work it appears merely to record.
The economics of this property are distinct from the economics of information capture. Capture preserves existing reasoning at low marginal cost. Generation produces new reasoning at positive cost but with potentially much higher marginal value, because the generated reasoning addresses constraint interactions that no existing mechanism was producing information about. The Scaling assessment that a fund manager produces for the first time because the disclosure format demanded it may be imperfect - it is first-cycle reasoning, unimproved by calibration feedback. But imperfect reasoning about a constraint interaction that was previously invisible to the entire market is more valuable than no reasoning at all. The scoring mechanism improves it over time. The performative mechanism had to create it first.
Dye, R. A. (1985). "Disclosure of Nonproprietary Information." Journal of Accounting Research 23(1): 123-145.
Verrecchia, R. E. (1983). "Discretionary Disclosure." Journal of Accounting and Economics 5: 179-194.
This paper presents results from the Prolegomena to the Unification of Economics, Computer Science, Mathematics, and Physics (Beier, forthcoming 2026). The performative property developed here is the generative complement to the improvement mechanism in the companion paper (Beier, 2026b): the improvement mechanism makes existing reasoning better; the performative mechanism brings new reasoning into existence. The author invites engagement from the economic methodology, science and technology studies, and behavioral economics communities.