Structured Underspecification

The architectural formula:

Principled criterion + behavioural examples + unspecified mapping = framework that warrants enforcement two ways at once.

The principled criterion warrants accusation through moral leverage. The behavioural enumeration warrants accusation through apparent specificity. The unspecified mapping lets enforcers choose between warrants per case while each application appears faithful to the text.

The operational triage test:

In contested cases, where does the mapping rule live — in the framework text, in external binding materials, in accountable institutional process, or in unreviewable enforcer judgment? If the answer is the fourth, the framework has the selective-enforcement architecture.

I. The dual-warrant formula

Three components:

Principled criterion. Borrows moral and procedural weight from a general principle — subordination, oppression, harm, fairness, safety, respect, dignity, professionalism, civility. Makes the framework morally defensible and emotionally chargeable.

Behavioural enumeration. Borrows operational specificity. Lists concrete acts as examples or instances. Makes the framework appear actionable and rule-like.

Unspecified mapping. Leaves the relationship between criterion and enumeration unstated. Are listed behaviours sufficient for the category, necessary, presumptive, heuristic, illustrative, evidentiary, rebuttable? The text does not commit, and no external binding material supplies the commitment.

The dual-warrant structure produces selective enforcement. The framework can be invoked through the enumeration (“you did the listed thing, the principle applies”) or through the principle (“the listed examples are heuristic, context permits exemption”). Enforcers move between the two warrants depending on whether condemnation or exemption is wanted. Each application looks faithful to the text because both warrants are textually available.

II.A The Mielenterveystalo specimen

The framework is useful enough to make the failure interesting. The Finnish public mental-health portal Mielenterveystalo hosts an aggression self-management programme developed by a publicly-funded specialist healthcare provider, drawing on adjacent family-services and education-agency materials. The framework is reproduced through teacher training, parenting advice, social work, and mental-health services. It separates aggression-as-feeling from violence-as-act, acknowledges aggression’s positive functions (self-defence, healthy self-esteem, demanding respectful treatment), and names operational boundaries (no harming people, no breaking things, no irreversible action). These are real distinctions, professionally drafted.

The framework’s principled criterion: violence is everything that subordinates or exploits another person. The criterion is intent/effect-based.

The framework’s behavioural enumeration: verbal berating, name-calling, insulting, invalidating, and mocking, given as examples of psychological violence. The text presents these as examples, leaving open whether they are sufficient, presumptive, or heuristic for the criterion.

The framework’s self-defence doctrine: self-defence is needed not only in physical struggle but also when someone needles, verbally berates, or otherwise oppresses another’s rights through speech. Plus the prohibition: self-defence must not use physical or psychological violence.

A charitable reading: the framework expects users to apply the principled criterion (subordination/exploitation) contextually to the behavioural examples, treating the examples as illustrative, not categorical. Under this reading, defender-role use of the listed verbs does not satisfy the criterion, and the framework is internally coherent.

The framework does not say this. It does not specify the mapping. The charitable reading is one possible interpretation; the categorical-enumeration reading is another; the framework supports both. The pathology is not that one reading is wrong — both are textually available. The pathology is that the choice between them is the enforcer’s, with no accountable institutional process recording or constraining the choice. Whether enforcer choice tracks actor status more reliably than the stated principle is the empirical signature §VII separates from the architectural diagnosis — a question for case data, not for the framework text alone.

When the framework is invoked against a teacher whose responsive verbal sharpness annoys an administrator, the enumeration reading does the work: you used a listed verb, that is psychological violence. When the framework is invoked to legitimise a student’s verbal pushback against verbal attack, the principle reading does the work: it was self-defence, no subordination. Same behaviours, opposite classifications, framework text supports both. The architectural overlap is exact, not coincidental: verbal berating appears explicitly in both the psychological-violence enumeration and the self-defence-trigger list, and the self-defence doctrine’s catch-all phrase — speech that oppresses another’s rights — restates the principled criterion (violence as subordination of another). The architecture is one concept invoked twice with opposite institutional consequences depending on which reading the enforcer selects.

The architecture is in fact triple-warranted, not merely dual. The self-defence trigger authorises response to verbal aggression; the prohibition forbids self-defence using physical or psychological violence; the principle level collapses both categories into one unmarked violence with no proportionality rule between them. The framework provides no specified method for defending against verbal aggression without committing it, and no specified mapping between the level of violence triggering self-defence and the level permitted in response. Two readings are textually supported: (a) the prohibition empties the entitlement — any actual self-defence is itself classifiable as psychological violence and therefore prohibited; or (b) the principle level’s category collapse means violence is violence, so physical response to verbal aggression is no more architecturally constrained than the verbal-response-to-verbal case the entitlement explicitly authorises. Neither reading is intended by the framework’s drafters; both are available to its enforcers. The rights-vocabulary outer armour — rights and oppression discourse deployed inside what is otherwise a clinical self-help context (see Language Has No OWASP §15 on symbols used outside their original protective context) — makes the architecture self-defending against substantive criticism: objecting to the framework reads as objecting to rights themselves.

II.B The Miller-family specimen

The same architecture operates with inverted coalitional coding in US state “harmful to minors” library and school-material statutes built on the Miller v. California (1973) three-part obscenity framework. The framework is useful enough to make the failure interesting: Miller separates protected speech from unprotected obscenity, preserves serious literary, artistic, political, and scientific work through the SLAPS prong, and defers community-specific judgments to local fact-finders rather than imposing a national standard. These are real distinctions, drafted carefully by a Supreme Court majority over a long line of prior cases.

The framework’s principled criterion: a work is obscene only if “the average person, applying contemporary community standards” would find it appeals to the prurient interest, depicts sexual conduct in a patently offensive way, and “taken as a whole, lacks serious literary, artistic, political, or scientific value” (Miller, 413 U.S. at 24). The criterion is intent/effect-based and explicitly whole-work.

The framework’s behavioural enumeration: “ultimate sexual acts, normal or perverted, actual or simulated... masturbation, excretory functions, and lewd exhibition of the genitals” (Miller, 413 U.S. at 25), elaborated by Florida F.S. 847.001(16) into thirteen enumerated sub-elements. Florida HB 1069 (2023) operationalises the enumeration at school-library level: any material a parent reports as “describing sexual conduct” must be removed within five days pending magistrate review with no statutory deadline.

A charitable reading: the statute expects reviewers to apply the principled criterion (taken-as-a-whole, SLAPS-value-preserving, community-standards-adjudicated) contextually to the enumerated content, treating the enumeration as illustrative, not categorical. Under this reading, a passage describing sexual conduct in a work of serious literary value does not satisfy the criterion, and the framework is internally coherent.

The framework does not say this. The Miller opinion does not specify the mapping rule between criterion and enumeration in contested cases beyond the bare “trier of fact applies community standards” deferral. The state statutes do not specify it. The Davis School District in Utah removed the Bible from elementary and middle schools under HB 374 in 2023 after a complainant read the statute back at it and listed the enumerated content the Bible contains; the appeal committee restored it five months later on the recorded reasoning that the Bible has “significant, serious value for minors which outweighs the violent or vulgar content it contains.” The same district under the same statute the same year removed Gender Queer on the enumeration reading. On the Road and Slaughterhouse-Five were removed from Florida school libraries under HB 1069 on the enumeration reading. In Penguin Random House v. Gibson (M.D. Fla., 13 August 2025), Judge Carlos Mendoza struck down core sections of HB 1069 and ruled that 23 named removed books are not obscene under Miller’s whole-work requirement — confirming that the same statute, applied by district committees and by a federal court, had produced incompatible mapping outcomes for the same comparator class.

Same behaviours, opposite classifications, framework text supports both. The coalitional coding is inverted from Mielenterveystalo’s — defenders here are right-aligned legislators and parental-rights groups, critics are left-aligned publishers and library associations — and the substantive content of the principled criterion (sexual content harmful to minors) is unrelated to the substantive content of the Finnish framework (psychological violence). What is invariant is the dual-warrant structure: principle borrows moral force, enumeration borrows operational specificity, mapping authority is supplied case by case by an enforcer (jury, magistrate, committee, complainant) whose contextual judgment tracks the coalitional standing of the affected work more reliably than the stated principle.

The Miller-family framework has more developed external mapping authority than Mielenterveystalo — decades of First Amendment precedent, the SLAPS prong as a partial structural anchor, federal-court review as a backstop. The presence of partial external authority moves the framework toward §III’s second category (“external binding materials supply mapping”) rather than purely the fourth (“unreviewable enforcer judgment”). The architecture is still dual-warranted; the unreviewability is partial. Appeals work, slowly and expensively. Mielenterveystalo has no comparable appeal backstop. The two specimens demonstrate the same dual-warrant architecture operating in institutional settings with substantially different appeal-backstop strength; the architecture survives that institutional variation.

III. What this is not — the dual-warrant boundary

Structured underspecification is not ordinary vagueness. All rules require interpretation. Hart’s analysis of legal open texture identifies an irreducible vagueness in legal language that must be resolved by interpretive judgment. Fuller’s Morality of Law names rule clarity as a constitutive principle of legality but does not demand infinite specification. Lipsky’s Street-Level Bureaucracy documents how front-line officials necessarily exercise discretion in routine enforcement. Davis’s Discretionary Justice examines the structural role of administrative discretion. Each adjacent literature treats discretion as unavoidable, often necessary, and sometimes desirable.

The pathology this essay names is the specific dual-warrant combination — moral/procedural force at the top, behavioural examples in the middle, unstated mapping rule between them — that allows two textually faithful but operationally opposite applications of the same framework to the same facts.

The diagnostic distinguishes structured underspecification from benign vagueness by asking where the mapping rule lives:

• If supplied by the framework text → not the pathology
• If supplied by external binding materials (precedent, agency guidance, statutory interpretation) → not the pathology, but the diagnostic must follow to the external materials
• If supplied by accountable institutional process (visible decision rules, recorded reasoning, comparator-class review, appeal paths) → not the pathology
• If supplied by unreviewable enforcer judgment → the pathology

Hart’s open texture often falls in the second category — common-law jurisdictions supply the missing mapping through case-by-case precedent. Lipsky-style street-level discretion sometimes falls in the third — when the institution has reasoned-decision and review functions. Davis’s discretionary justice sometimes falls in the fourth — when discretion is exercised without accountability infrastructure. The pathology is specifically the fourth case, in settings where enforcement consequences attach.

IV. Further candidate applications

§II.A and §II.B work two specimens at document level, coalitionally inverted, substantively unrelated, structurally identical. The architecture is plausibly substrate-general. The candidates below are additional applications worth document-level engagement; they are included because the principled-top + concrete-examples + unspecified-mapping triple appears likely, not because this essay has confirmed it document by document.

HR respectful-workplace codes. A typical large-employer code states a principled criterion (respect, dignity, civility, professional conduct), lists behavioural examples (raised voice, demeaning language, public criticism, exclusion from team events, hostile tone), and leaves unspecified when each behavioural example satisfies the principled criterion in contested cases.

Hate-speech and platform-moderation policies. Major content-moderation regimes state a principled criterion (harm reduction, safety, dignity, anti-discrimination), list behavioural examples (slurs, threats, dehumanisation, incitement), and add context exceptions (artistic merit, news reporting, criticism, satire). The mapping is often supplied internally through training rather than textually.

School discipline codes with “safe learning environment” framing. Stated criterion: all students have the right to a safe learning environment, free from harassment, bullying, and disruption. Enumeration: physical aggression, verbal harassment, social exclusion, disrespect of authority, dress-code violations, classroom disruption. Mapping unspecified in the text.

National-security classification and watchlisting frameworks. Stated criterion: protect national security; prevent harm to the United States or its allies. Enumeration: information whose disclosure could damage intelligence sources, methods, foreign relations. Mapping unspecified textually, supplied internally through agency guidance with limited external review.

Immigration discretion categories. Good moral character, public charge, national security, credible fear each combine a principled criterion with enumerated factors and leave the mapping to officer judgment with limited appeal in many cases.

The candidate list spans coalitional defenders and critics. The substrate-general claim is testable case by case — the diagnostic in §X is what does the per-case testing.

V. Why the architecture pays — the institutional function

Reward Substrate names what selects for institutional vocabulary at the discourse level. Applied to selective-enforcement architecture, the substrate pays the dual-warrant combination because the architecture preserves discretion under moralised cover:

• Discretion preservation: enforcers retain judgment authority case by case
• Coalition coherence: different audiences read different content into the same text
• Falsification resistance: any specific case can be exempted as edge case or special context
• Moral leverage: the principled criterion does the moral work without committing to specific application
• Operational appearance: the behavioural examples appear specific without committing to categorical rules
• Reform resistance: clarifying the mapping imposes enforcement costs the institution would have to bear

The dual-warrant architecture enables these rewards by making both warrants textually present, so enforcers can move between them without violating the text. Other framework architectures — purely rule-based or purely principle-based — do not produce the same rewards because they do not allow the same mid-case warrant-switching. The framework’s institutional function is precisely to remain dual-warranted; clarifying the mapping would impair that function. Structural reform faces predictable resistance from carriers who benefit from the underspecification: the reform reads as cost-imposing, because it removes the discretion-with-cover that made the framework operable.

The mechanism by which the discretion pays is the response-vector logic: institutional pressure routes through the cheapest available channel. The unspecified mapping rule keeps multiple enforcer channels equally cheap. Faced with a case where condemnation is locally costly, the enforcer routes through the principle reading; faced with a case where exemption is locally costly, the enforcer routes through the enumeration reading. The dual warrant guarantees that some reading is always the path of least friction. Removing the underspecification closes the cheap channel; the architecture pays because keeping it open keeps both channels available for case-by-case selection.

VI. Prior art and differentiation

The adjacent academic literature is well-developed. H. L. A. Hart’s open texture is the linguistic-philosophy dimension. Lon Fuller’s Morality of Law (1964) is the jurisprudential-normative dimension. Kenneth Culp Davis’s Discretionary Justice (1969) is the administrative-law-positive dimension. Michael Lipsky’s Street-Level Bureaucracy (1980) is the policy-sociology dimension. Critical Legal Studies (Kennedy, Tushnet) is the legal-realist-critical dimension.

The corpus contribution is artifact-shape, not novelty of observation. The dual-warrant architectural formula packaged as a deployment-grade diagnostic with the where-does-the-mapping-live triage test lets a reader apply the formula to any framework in 30 seconds: name the criterion, list the behavioural examples, identify whether the mapping is text-supplied, externally-supplied, accountably-supplied-internally, or unreviewably-enforcer-supplied. The first three are not the pathology; the fourth is. Hart, Fuller, Davis, and Lipsky are the source observations the corpus diagnostic operationalises.

VII. Same method, different findings

The diagnostic identifies architectural capacity for selective enforcement — text-level dual-warrant plus accountability absence plus enforcement-consequence-attachment. The empirical signature of selective enforcement in operation — same behavioural examples producing different mapping outcomes across cases, with the asymmetry tracking actor status more reliably than the stated principled criterion — is a separate claim that requires comparative case data to confirm.

A framework can have the architectural capacity for selective enforcement without that capacity being exercised asymmetrically in practice. A framework whose architecture lacks the capacity can still produce asymmetric outcomes through other mechanisms (asymmetric enforcement resources, asymmetric reporting, asymmetric political pressure). The architectural diagnostic identifies the capacity; the empirical signature confirms exercise.

The diagnostic also does not decide whether the underlying category should exist, whether a given act should be sanctioned, or which coalition is right about the value question. Harm, safety, dignity, national security, community standards, good moral character — each is a contested category with legitimate defenders and legitimate critics. The diagnostic tests where mapping authority lives within frameworks that operationalise these categories; it does not adjudicate the categories themselves. A framework that passes the diagnostic (text-supplied mapping, external binding materials, accountable institutional process) carries a contested category responsibly; a framework that fails the diagnostic carries the same category with discretionary cover.

VIII. The accountable-discretion repair

The repair is not specifying every mapping rule textually. Discretion is unavoidable and often necessary. The repair is making the discretion accountable.

For any framework with the selective-enforcement architecture, the accountable-discretion repair specifies six governance questions:

Who decides? Which institutional role exercises the mapping discretion in contested cases? Named role, named accountability chain.

What facts must they consider? What categories of evidence does the decision-maker have to engage? Not all facts; specified categories that the framework’s principled criterion implies.

What reasons must they record? What argumentation has to be written down — internally if not publicly — connecting the facts to the mapping decision? Records that survive comparator-class review.

What comparator class is checked? When the same behavioural example produces different mapping outcomes, which prior cases must be reviewed for consistency? The comparator class names the unit across which the institution accepts the discipline of comparative coherence.

What remedy range is authorised? What proportional response to what kinds of mapping outcomes? Specified ranges, not unbounded enforcer discretion.

What appeal or audit path exists? Who can challenge the mapping decision through what procedure? Internal review, external audit, judicial review — at least one accountability path must be reachable by the affected party at reasonable cost. The grade at which the appeal path bites — decorative, advisory, reputational, procedural, consequential — is what Feedback Authority classifies. An “appeal exists” that resolves at the advisory grade does not constrain the mapping; an appeal that resolves at the procedural grade (the institution must give a reasoned answer in a record another actor can review for consistency with comparator-class outcomes) does. The repair fails when it installs a lower-grade appeal under higher-grade rhetoric.

The six together convert dual-warrant unreviewable discretion into dual-warrant accountable discretion. Accountable discretion is not weaker enforcement; in many cases it is stronger enforcement. Sanctions imposed under accountable discretion are more durable against bad-faith challenge, less vulnerable to procedural reversal, less easily delegitimised as arbitrary, and more capable of carrying weight across institutional and political turnover. The framework’s principled criterion and its behavioural enumeration are preserved; what changes is the discretion-with-cover becoming discretion-with-accountability. The target is unreviewable mapping authority, not the substantive policy category.

The comparator-class element is the most institutionally rare and the most load-bearing. Selective enforcement often appears only across cases, not within one case. A framework with no comparative-application-review function cannot produce the diagnostic data that would expose the selective application. Adding the review function is the structural change most resisted because it is the change that would most expose the prior asymmetry.

IX. Category talisman as frequent amplifier

Selective-enforcement architectures frequently operate through a category talisman: a morally loaded category — violence, safety, harm, dignity, oppression, respect — placed at the principle layer. The category supplies the moral weight; the unspecified mapping supplies the enforcement flexibility. The two are independent primitives, frequently co-occurring: morally loaded categories pay better in the substrate, and frameworks built around morally loaded categories typically need flexibility to be operable across diverse cases. Structured underspecification can operate without category-talismanic content (an IT acceptable-use policy with operational criteria and unspecified mapping is still selective-enforcement-architecture), and category talismans can operate without structured underspecification (“safety” as pure veto, “violence” as rhetorical inflation in a campaign with no institutional enforcement). The architectural dependence is empirical and frequent, not definitional.

X. The diagnostic — triage test plus external-authority check

For any institutional framework that combines moral-coded principles with behavioural examples:

Step 1 — Triage. Identify the principled criterion. Identify the behavioural enumeration. Identify whether the framework text specifies the mapping rule (categorical / heuristic / presumptive / illustrative / evidentiary).

Step 2 — External-authority check. If the text does not specify the mapping, identify whether external binding materials supply it. Common-law frameworks may have precedent supplying mapping; statutory frameworks may have agency guidance; professional codes may have ethics-tribunal decisions; institutional codes may have published case-handling guidance. If external mapping authority exists and is binding, the framework’s text-level underspecification is not the pathology — the external authority is doing the work.

Step 3 — Accountability check. If the mapping is supplied by enforcer judgment without text-level or external-binding specification, identify whether accountable-discretion infrastructure exists (who decides / what facts / what reasons recorded / what comparator class / what remedy range / what appeal). If the accountability infrastructure is present, the discretion is accountable rather than the pathology.

Step 4 — Pathology identification. If steps 1-3 yield “the framework text does not specify, external binding materials do not specify, and accountable-discretion infrastructure is absent in the institutional setting where enforcement consequences attach,” the framework has the structured-underspecification pathology.

Follow-on empirical test (separate from the 30-60 second diagnostic). Once architectural pathology is identified, the predictive hypothesis is: same behavioural examples will produce different mapping outcomes across cases, with the asymmetry tracking actor status (coalition alignment, demographic profile, institutional standing) more reliably than the stated principled criterion. Confirming this requires comparative case data — comparable cases with comparable behavioural facts and different mapping outcomes correlated with actor-status variables. The architectural pathology creates the capacity and incentive for the asymmetric pattern; the empirical pattern requires its own evidence.

The four-step test is operationalisable by a corpus reader on most framework documents in 30-60 seconds, with the qualifier that step 2 (external-authority check) may require domain knowledge the reader can defer to a domain expert. The follow-on empirical test requires case data and is properly a separate empirical project.

XI. Cluster note

Structured Underspecification is a diagnostic at the enforcement-architecture layer. Causal Talisman names the parallel move at the causal-claim layer (morally protected distal cause-name substituting for contribution accounting). Slot-Function Fallacy names the move at the institutional-architecture layer (slot existence claimed as function performance) — covered as small additions to What Bureaucracy Is, The Procedural Object, and Inside the Closure Machine rather than standalone. Category Talisman is a frequent amplifier (§IX), folded into Compression Paradox. See Related.

XII. Close

Frameworks survive by being dual-warranted. The architecture is what good-looking institutional design defaults to when the institution needs flexibility under moralised cover, and what enforcers value when they want to retain discretion that the text appears to constrain. The cure is making the discretion accountable — recorded reasoning, comparator-class review, appeal paths. Institutions that depend on the dual-warrant architecture for operational flexibility resist accountability infrastructure because the accountability removes the flexibility-with-cover the architecture was supplying. The resistance is itself informative: it indicates the architecture was doing institutional work the institution does not want to give up.

Accountable discretion is not weaker enforcement; in many cases it is stronger enforcement, more durable against challenge and more capable of carrying weight across turnover. The diagnostic gives the diagnosis; the repair carries the cost the institution will resist; the resistance carries the diagnostic information that confirms the analysis.

Related:

• The Causal Talisman — sibling primitive at causal-claim layer.
• The Statistic Was Still Known Internally — worked-case essay where Mielenterveystalo appears as discourse-vocabulary backdrop.
• The Refusal to Compute — institutional failure mode against which the diagnostic operates.
• The Compression Paradox — cognitive form of single-axis compression; Category Talisman fold lives here as the institutional-vocabulary version.
• What Bureaucracy Is — institutional-claim layer (Slot-Function Fallacy folded here); also in The Procedural Object and Inside the Closure Machine.
• Reward Substrate — selection pressure on institutional vocabulary; §V applies it to the dual-warrant architecture.
• Hardening Devices — substrate-native cures; accountable-discretion infrastructure (§VIII) is the institutional-substrate hardening device for the discretion-with-cover pathology.
• The Response Vector — institutional pressure routes through the cheapest channel; the unspecified mapping rule keeps multiple enforcer channels cheap.
• Feedback Authority — the five profile-grades that classify what the §VIII appeal path actually compels.