Image of the Beast: Legal Personhood, Autonomous Entities, and the Hermeneutics of Institutional Fiction ()
1. Scope, Method, and Epistemic Status
This paper works simultaneously in three registers: legal hermeneutics (how legal categories are constructed and read), positive law (how those categories are stated in statute and precedent), and institutional sociology (how such categories are performed, contested, and reproduced across time). The claims below are of three corresponding kinds—interpretive, doctrinal-descriptive, and sociological—and they carry different burdens of justification. The paper does not dispute positive-law outcomes in decided cases; it targets the interpretive frame in which future cases will be settled and the institutional configurations that make certain settlements more likely than others.
The scope is deliberately restricted. Decision domains covered are credit and consumer lending, employment eligibility and screening, welfare and social-protection eligibility, insurance underwriting and pricing, healthcare access authorisation, biometric authentication for essential services (banking, mobility, housing), and administrative sanctions imposed by automated means. Jurisdictions covered in the first instance are those with codified data-protection and administrative-law regimes providing contestability rights—the EU/EEA (GDPR, AI Act), the United Kingdom (Data Protection Act 2018), Norway (Personopplysningsloven), and comparable frameworks. The unit of analysis is the individual biological person subject to a specific consequential decision produced by an identifiable representational system, over an identifiable time horizon, in a specified decision domain.
2. Hermeneutic Foundation
2.1. Two Questions Distinguished
Two questions are systematically confused in casual writing about legal personhood, and the argument of this paper depends on separating them. The first is the positive-law question: what does the category “person” denote in a given jurisdiction at a given time? On that question, statutes, codes, and binding precedent supply determinate legal answers. Salomon v. A. Salomon & Co Ltd [1897] AC 22 fixed the doctrine of separate corporate personality in English law; Trustees of Dartmouth College v. Woodward, 17 U. S. (4 Wheat.) 518 (1819), established the corporate charter as an inviolable contract in United States law; Citizens United v. FEC, 558 U. S. 310 (2010), extended First Amendment political-speech protection to corporate entities. Whether one likes those answers or not, they are the answers within their legal systems. This paper takes no issue with their positive-law authority.
The second is the normative and interpretive question: what justifies the category, what work is it doing institutionally, and by what argumentative moves is it extended to new subjects—foundations, unions, states, subsidiaries, algorithmic entities? That question does not have a determinate positive-law answer, because the answer changes as the argumentative moves succeed or fail. The hermeneutic argument of this paper applies only to the second question. Where the paper elsewhere states that legal personhood has “no empirical fact of the matter”, the claim is that no empirical referent grounds the category as an ontological matter—not that decided cases lack legal answers within their jurisdictions.
2.2. Gadamer: Horizonverschmelzung and Prejudice
Legal reading is not a neutral extraction of a rule from a text; it is a fusion of horizons in which the reader’s prior situation partly constitutes the meaning that arises. Gadamer’s analysis in Truth and Method treats prejudice as a productive condition of understanding rather than an obstacle to it (Gadamer, 2004). Legal doctrine, on this reading, cannot be understood as a fixed corpus that judges merely apply; it is a text whose meaning is regenerated at every point of application. The genealogy of corporate personhood traced in Section 4 is a Gadamerian case: each generation of judges reads the “artificial being” of Marshall through the situation of its own institutional needs.
2.3. Ricoeur: Entextualization and Distanciation
Ricoeur’s distinction between discourse and text—the transformation by which utterance becomes a fixed inscription capable of being read outside its original situation—is the pivot on which most of this paper turns (Ricoeur, 1981). Legal categories such as “person”, “company”, “decision”, “consent”, and “signature” are entextualized: they cease to be speech-acts of particular people at particular moments and become inscriptions that later readers must interpret. Ricoeur’s distanciation names precisely the productive distance between the inscribing situation and the reading situation. The Mark framework proposed later (Sections 26 and 42) is a systematic attempt to specify the entextualization gradient from face-to-face administrative decision to fully-automated determinative scoring.
2.4. Derrida: Difference and the Iterative Signature
Derrida’s analysis of the signature as necessarily iterable—repeatable in the signer’s absence, or it is not a signature—supplies the bridge from written signature to electronic signature to biometric authentication to behavioural pattern to pure algorithmic score (Derrida, 1982). Each step generalises the previous step’s iterability while stripping some of the situational grounding that made the earlier form intelligible as a signature at all. The temporal asymmetry analysed in Section 46 is a Derridean effect: the iterated inscription outlives the situation of its production and comes to govern situations the producer never envisaged.
3. Philosophy of Language and Institutional Fact
3.1. Austin and Searle: Performatives, Declarations, Institutional Facts
Austin’s performative utterances (Austin, 1975) and Searle’s later analysis of declarations and institutional facts (Searle, 1969, 1995) explain how legal categories bring about the states of affairs they name. “This company exists” is not a report on a pre-existing corporate substance; it is the recognition of a status conferred by a declarative act (registration) within a status-conferring institution (the registrar of companies). Institutional facts of this kind require collective recognition; they are subject to the specific vulnerabilities and specific durability of collective recognition.
3.2. Hart: Rules of Recognition and Open Texture
Hart’s Concept of Law distinguishes primary rules of conduct from secondary rules that confer power to make, change, and adjudicate primary rules (Hart, 2012). The rule of recognition—the master rule by which officials identify what counts as law in a given system—is itself a social fact about official practice, not a further norm above the practice. Hart also emphasises the open texture of legal language: general terms have a settled core and a penumbra of borderline cases where the term’s application requires an interpretive judgment that is not itself supplied by the term. Legal personhood is, in this sense, a legal concept with a settled core (registered corporations are persons for the purposes of contract, tort, and criminal responsibility in most systems) and a wide penumbra (unincorporated associations, autonomous algorithmic entities, natural features such as rivers, animals, foetuses, corporate groups).
3.3. Semantic Slippage and Controlled Meaning
The paper’s central analytical device is the notion that ordinary and legal senses of the same term coexist in interpretive practice and can shift under argumentative pressure. “Person”, “company”, “consent”, “signature”, “decision”, “identity”—each carries an ordinary-language sense and a technical-legal sense whose relation is not fixed. Controlled meaning is the interpretive labour by which one sense is stabilised against the other for the duration of a particular argument or transaction. The genealogy in Section 4 can be read as a series of successful semantic-control operations.
4. A Genealogy of Corporate Personhood
The genealogy below is presented with pinpoint authorities at every stage. Where a doctrinal move rests on dicta, on a headnote, or on a subsequent gloss rather than on the operative reasoning of the cited case, this is stated explicitly. The purpose is to make the interpretive uptake—how a fictional entity accreted rights and immunities through successive readings—independently verifiable against primary sources. Historical accounts of corporate personhood have been developed by Maitland (1905), von Gierke (1900), Dewey (1926), and more recently Ciepley (2013), each supplying a portion of the vocabulary drawn on here.
Chartered trading companies. The English East India Company (chartered 31 December 1600 by letters patent of Elizabeth I) and the Dutch Vereenigde Oostindische Compagnie (Octrooi of the States-General of 20 March 1602) established the pattern by which a sovereign delegates specified public powers—trade monopoly, treaty-making, coining money, waging war—to a legally-distinct entity constituted by charter. The entity holds those powers as its own; the sovereign’s residual authority is exercised through the terms and revocability of the charter.
The “artificial being” formulation. In Trustees of Dartmouth College v. Woodward, 17 U. S. (4 Wheat.) 518, 636 (1819), Chief Justice Marshall delivered the opinion of the Court, stating that “a corporation is an artificial being, invisible, intangible, and existing only in contemplation of law.” The passage is quoted in almost every subsequent United States corporate-personhood decision and functions as the foundational articulation of the entity’s ontological status. The immediate holding was that the College’s charter constituted a contract within the Contract Clause and could not be impaired by state legislation.
Separate personality. Salomon v. A. Salomon & Co Ltd, [1897] AC 22, established in English law that a properly incorporated company is a distinct legal person separate from its shareholders, and that this separation obtains even where a single individual holds effective control. Lord Halsbury LC at 30 - 31 affirmed the statutory framework of the Companies Acts as determinative of the question; Lord Macnaghten at 51 stated the classic dictum that “the company is at law a different person altogether from the subscribers.” The doctrine’s consequence—that shareholders bear no personal liability for the company’s debts absent fraud or statutory piercing—is the operative core of modern limited-liability corporate law.
The Santa Clara headnote. The Fourteenth Amendment statement in Santa Clara County v. Southern Pacific Railroad Co., 118 U. S. 394, 396 (1886), by which corporations are treated as “persons” within the meaning of the Equal Protection Clause, does not appear in the Court’s reasoning. It appears in the Reporter’s headnote at 118 U. S. 394, 396, prefaced by the note that “The court does not wish to hear argument on the question whether the provision in the Fourteenth Amendment to the Constitution, which forbids a State to deny to any person within its jurisdiction the equal protection of the laws, applies to these corporations. We are all of opinion that it does.” This is attributed to Chief Justice Waite as a pre-argument remark recorded by court reporter J. C. Bancroft Davis; the Court’s actual reasoning resolved the case on grounds unrelated to constitutional personhood. Every subsequent doctrinal reliance on Santa Clara for the “corporations are persons under the Fourteenth Amendment” proposition is therefore reliance on a headnote, not a holding—a paradigmatic case of interpretive uptake carrying a fiction into constitutional status through repetition rather than through the reasoned decision of the Court.
Political-speech extension. Citizens United v. FEC, 558 U. S. 310, 342-347 (2010), held that the First Amendment prohibits the government from restricting independent political expenditures by corporations, on the ground that political speech does not lose First Amendment protection based on the corporate identity of the speaker. The dissent contested the historical premise. The decision is the current operative doctrine in United States constitutional law and is treated in this paper as the endpoint of a genealogy running from chartered trading company to constitutional rights-bearer.
5. The Argentine Automated-Company Proposal: A Recent Extension
The Argentine draft is cited with the operative provisions and the scholarly analysis fully identified. The government’s framing appears in Presidencia de la Nación, Comunicado Oficial No. 149 of 18 June 2026, §§ 3-7. The draft General Companies Law, Title IV (provisions on Automated Companies and DAOs), articles 1-3, 6, and 9, specifies that legal personality attaches to the company itself and not to the AI system that operates its decision functions; that human natural persons must be designated as owners, operators, and beneficiaries and remain the objects of imputation for regulatory and civil purposes; and that the AI-operated character of the entity triggers additional registration and disclosure obligations. The doctrinal analysis in Molina Sandoval (2026) at pages 52-58 treats the dissociation of decision, personhood, and imputation as the substantive doctrinal move.
Read carefully, the Argentine draft does something more subtle than extend legal personhood to AI. It attaches personality to the company (as before), preserves human imputation (as before), and adds a disclosure regime for AI-operated decision functions (this is new). The finding of this paper is that the framing of the Comunicado overstates the doctrinal novelty: the draft does not grant personhood to an autonomous algorithmic entity; it regulates the autonomous algorithmic operation of an entity whose personhood is already conventional. Where legal personhood was already dissociated from biological substrate (that has been the doctrine at least since Dartmouth), the substantive innovation is the automated-operation surcharge—the additional obligations that attach when the operational layer is not human.
6. The Biological-Substrate Argument and Its Legal Reach
The paper defends what it calls the biological-substrate priority: that biological persons have a claim to be the terminal locus of certain kinds of legal recognition, which juridical persons do not have on the same terms. The claim requires precise scoping, because on the broadest reading it would flatten differences among juridical persons that positive law has good reasons to preserve. The biological-substrate priority operates along three tiers with different legal reach.
Tier A—Fundamental-rights conflicts (all juridical persons). Where any juridical person’s asserted rights conflict with the fundamental rights of a biological person—life, bodily integrity, due process, freedom of expression, privacy, non-discrimination, freedom of association—the biological person’s rights prevail absent a compelling and narrowly-tailored justification. This is the strongest form of the priority and applies uniformly across all juridical-person types: corporations, foundations, unions, charities, media organisations, state entities.
Tier B—Autonomous-operation surcharge (AI-operated entities). Where a juridical person is operated in substantial part by autonomous decision systems, additional procedural safeguards attach—verifiable decision logs, structured attribution to human owners, operators, deployers, and beneficiaries, and reversibility mechanisms. The surcharge is procedural in character, does not itself defeat rights, and is designed to preserve the possibility of contestation.
Tier C—Ordinary associational proportionality (unions, charities, media, religious bodies, cooperatives). Where the juridical person is an association carrying independent expressive, associational, or public-interest functions, the priority yields to ordinary proportionality analysis. The associational function is entitled to protection on its own terms—a labour union’s bargaining function, a charity’s beneficiary-directed operations, a media organisation’s editorial independence, a religious body’s doctrinal autonomy—and the biological-substrate principle does not defeat associational rights except where Tier A is engaged.
The three tiers are exhaustive with respect to the paper’s use of the principle. Applications to specific institutional types should be resolved by identifying which tier is engaged and how; Sections 13 and 37 apply this tiered scope directly rather than restating the priority in abstract form.
7. New Public Management, Metric Discretion, and the Interpretive Layer
New Public Management (Hood, 1991; Osborne & Gaebler, 1992; Pollitt & Bouckaert, 2017) and its algorithmic successors do not eliminate interpretation in administrative practice. They affect it in three specific ways, which are best captured by a three-part typology.
1) Constrain. NPM narrows the range of admissible reasons for an administrative decision. A discretionary judgment that would previously have been articulable in a wider vocabulary (welfare, professional judgment, contextual appropriateness) is narrowed to a smaller set of metric-referenced reasons.
2) Relocate. NPM moves interpretive acts from front-line officials to metric designers, model developers, contract drafters, procurement specialists, and audit reviewers. The interpretation does not disappear; it shifts upstream and becomes less visible to the affected party. Bovens and Zouridis (2002) traced this movement from street-level to system-level bureaucracy in the earlier information-technology phase.
3) Automate. NPM embeds threshold-based decisions in software, so that at the point of decision no human interpretive act is performed. Human interpretive acts remain in system design, configuration, exception-handling, and appeal review. Lipsky’s (2010) analysis of street-level bureaucracy shows how discretion persists even when a formally-automated step intervenes.
Hybrid arrangements—in which officials select metrics, apply enumerated exceptions, review outputs, or exercise appeal authority—remain interpretively active. Such hybrids are the typical rather than the exceptional case. The critique developed here is that when the constrain/relocate/automate combination becomes deep enough, the human interpretive act becomes invisible to the affected party and difficult to reconstruct on appeal. This is a matter of degree and configuration, not a binary presence-or-absence claim. Empirical work on algorithmic administration (Pasquale, 2015; Eubanks, 2018; Yeung, 2017; Barocas & Selbst, 2016) documents both the constraint and the displacement of interpretive labour.
8. Material Stakes: Scope and Operationalisation
Scope and operationalisation statement. Decision domains: credit and consumer lending; employment eligibility and screening; welfare and social-protection eligibility; insurance underwriting and pricing; healthcare access authorisation; biometric authentication for essential services (banking, mobility, housing); administrative sanctions imposed by automated means. Jurisdictions: EU/EEA (GDPR, AI Act), the United Kingdom (Data Protection Act 2018), Norway (Personopplysningsloven), and comparable regimes. Unit of analysis: the individual biological person subject to a specific consequential decision produced by an identifiable representational system, over an identifiable time horizon, in a specified decision domain. Disconfirmation criterion: a high-risk classification under the Mark threshold (see Section 42) is disconfirmed for a given system where the affected person can, within a bounded and non-punitive procedure, 1) obtain the operative decision rule, 2) inspect the specific inputs used, 3) contest the classification before an accountable human reviewer with authority to reverse it, 4) obtain reversal in a documented material fraction of contested cases, and 5) access one or more functionally-equivalent alternatives for the underlying good. Any single missing element weakens but does not by itself defeat the classification; the disconfirmation is joint.
The stakes of these decisions are material and well-documented. Access to credit determines housing tenure, business formation, education. Employment screening determines income and social participation. Welfare eligibility determines subsistence. Insurance pricing determines exposure to catastrophic loss. Healthcare authorisation determines access to essential treatment. Biometric authentication is now the default gate to bank accounts, mobility infrastructure, and housing platforms. Administrative sanctions imposed by automated means (parking, congestion charges, benefits penalties) accumulate rapidly for those without appeal capacity. The Mark framework is proposed as an analytical structure for tracking how consequential decisions in these domains are mediated by representational systems and where in the mediation the possibility of contestation is preserved or lost.
9. The Legitimacy Filter
A distinct methodological point runs through the paper: legitimacy claims made by institutions on their own behalf are subject to their own version of the semantic-slippage problem. When a corporation says “we exist”, a state says “we govern”, or an algorithm-operated platform says “we decide”, each utterance imports a legitimacy claim whose grounds are contestable. The paper treats such claims not as first-order facts but as institutional performatives whose success depends on collective recognition—and whose vulnerabilities lie precisely where that recognition is manufactured rather than earned. This is the legitimacy filter: the requirement that self-legitimating utterances be tested against the institutional structure that would make them intelligible to a properly-situated critical reader.
10. Sociological Deconstruction
Sociological deconstruction as used here does not mean the dismantling of an institution; it means the visibility of how the institution’s categories are made and re-made through discrete acts of recognition, reading, and enforcement. Marx’s (1976) analysis of the commodity form, Weber’s (1978) of legal-rational authority, Durkheim’s (1984) of the division of labour and the collective representations, Simmel’s (2004) of money, Polanyi’s (2001) of the great transformation, Gramsci’s (1971) of hegemony, Bourdieu’s (1987) of the juridical field, Foucault’s (2009) of governmentality, Habermas’s (1996) of facts and norms, Giddens’s (1984) of structuration, and Latour’s (2005) of the network—each supplies part of the vocabulary the paper draws on. The specific institutional targets are corporate personhood, algorithmic administration, biometric identity, and automated decision-making; the specific analytical claim is that these institutions have knowable social structures that condition how their categories move.
11. Game Theory and the Strategic Structure of Recognition
The extension of legal categories is not spontaneous. It is driven by specific actors who benefit from specific extensions and who invest specific argumentative resources to secure them. A short game-theoretic sketch can be given: each proposed extension of “personhood” faces an admission cost (litigation, lobbying, doctrinal argument) borne asymmetrically by the party that benefits from the extension; the benefit is captured privately while the interpretive settlement, once achieved, becomes a public resource that later comparable parties can invoke at reduced cost. The dynamic favours cumulative extension along paths already opened, and disfavours retrenchment.
12. Litigation Architecture: Standing, Interest, and Argumentative Access
Corporate persons can litigate on their own behalf, appear as parties, and command the resources of professional representation. Biological persons often can litigate only where standing rules allow and where representation is affordable. The asymmetry is well-documented in access-to-justice research. The Mark framework proposed later treats this asymmetry as a structural feature to be tracked in any assessment of an automated decision system: who can litigate the outputs, on what grounds, at what cost, before what forum.
13. Six Principles for Dismantling
The paper does not propose to abolish legal personhood. It proposes to reconstruct its use around six principles which, in aggregate, restore the biological person to the terminal locus of certain kinds of legal recognition while preserving associational and expressive functions of collective entities. The principles are not axioms; they are proposed constraints on future doctrinal extension.
Principle 1—Attribution: every consequential decision must be traceable to identifiable human actors (owner, operator, deployer, beneficiary) who bear appropriate responsibility.
Principle 2—Biological-substrate priority (tiered). In accordance with Section 6: Tier A applies to fundamental-rights conflicts across all juridical persons; Tier B imposes procedural surcharges on AI-operated entities; Tier C preserves ordinary associational proportionality for unions, charities, media organisations, religious bodies, and cooperatives. The tiered structure is the operative form of Principle 2 in this paper.
Principle 3—Contestability: consequential decisions must be contestable within a bounded and non-punitive procedure before an accountable human reviewer with authority to reverse.
Principle 4—Disclosure: the operative decision rule and the specific inputs used must be inspectable by the affected party or their representative.
Principle 5—Alternatives: essential goods (banking, employment, housing, healthcare, mobility) must be accessible through functionally-equivalent alternatives so that a single system’s classification is not a total gate.
Principle 6—Reversibility: decision architectures must include reversal mechanisms with documented reversal rates as an indicator of the contestability procedure’s effectiveness.
14. Conclusions of Part I
Part I has established: 1) the two-question distinction that separates positive-law determinacy from normative contestability; 2) a hermeneutic and philosophy-of-language framework for tracking how legal categories are made and unmade in institutional practice; 3) a pinpoint-supported genealogy of corporate personhood from chartered trading companies through the Santa Clara headnote to Citizens United; 4) an analysis of the 2026 Argentine automated-company proposal that separates the doctrinal novelty (an AI-operation surcharge on a conventionally-owned company) from its rhetorical framing; 5) a tiered scoping of the biological-substrate priority; 6) a qualified account of how NPM constrains, relocates, and automates administrative discretion without eliminating interpretation; 7) a specified scope and operationalisation for the Mark framework that follows in Part II; and 8) six proposed principles for reconstructive use.
15. Introduction to Part II
Part II develops the paper’s triadic analytical framework. In brief: Image is the representation of a biological person maintained by an institutional or algorithmic system; Mark is an access-control function computed over the Image, determining what the biological person can and cannot do through the system; Sovereignty is the biological person’s capacity for contestation, correction, refusal, exit, and remedy with respect to the Image and the Mark. The full integrated statement of the triad, including its formal architecture, is given in Section 51. Part II builds up to that statement section by section, showing how the components are motivated by concrete legal and administrative structures.
16. The Six-Layer Representation Stack
A biological person interacting with modern administrative and commercial infrastructure is represented at six distinct layers, each with its own materials, its own operators, and its own vulnerabilities. Layer 0: the biological substrate itself. Layer 1: identity documents (birth certificate, passport, national ID). Layer 2: registered accounts (bank, tax, health, employment). Layer 3: behavioural profiles (transaction histories, movement patterns, communication metadata). Layer 4: algorithmic scores (credit, risk, employability, fraud). Layer 5: institutional entitlements (eligibility, permissions, access rights) determined by scores at Layer 4 read against rules at Layer 2 and Layer 1. The Mark, in the operational sense used in this paper, is the composite that Layer 5 uses to admit or refuse the biological person’s access to a given good or service.
17. Central Equation and Its Interpretation
The Mark threshold can be written schematically as M = f(D, E, O, I, C, S), where D is the decision domain, E is the entextualization gradient (how far the decision has moved from face-to-face to fully automated), O is the ownership and operational locus of the system, I is the inspectability of inputs, C is the contestability of the output, and S is the substitutability of the decision-mediated good. This is not offered as a scoring function to be computed today; it is offered as a template for identifying which parameters of a given system determine its position on a proposed risk classification. See Section 42 for the full operationalisation.
18. Table of Legal Fictions and Their Institutional Anchors
The paper compiles a systematic table (omitted here for space; retained in the appendix of the full manuscript) mapping specific legal fictions—the corporate person, the trust, the estate, the sovereign, the flag-state vessel, the diplomatic mission, the intellectual-property owner—to their institutional anchors (registration, deed, statute, treaty). Each fiction has known preconditions, known operations, and known failure modes; each is anchored to a specific human collective by a specific set of practices. Fictions without such anchors—invocation of maritime jurisdiction to escape traffic tickets, the “strawman” doctrine of the sovereign citizen catalogue—fail on institutional-grounding grounds and are analysed in Section 34.
19. Birth to Death: The Life-Cycle of an Image
The Image of a biological person is initiated at birth (the birth certificate), thickened by successive administrative interactions (school enrolment, driving licence, bank account, employment record, tax filing, credit history), and closed at death (death certificate, estate). Between initiation and closure the Image is continuously updated, corrected, re-scored, and read against varying rule-sets. The temporal asymmetry (Section 46) is that the Image outlives specific decisions taken about it and continues to carry their residue.
20. Semantic Bootstrapping in Institutional Categories
Institutional categories bootstrap their reference. “Person” refers to whatever the system’s recognition operations pick out; that reference is then used to constitute the rules that determine subsequent recognition. The circularity is not vicious in Hart’s (2012) sense—it is how any rule of recognition functions—but it is exploitable. Once “person” includes corporations, subsequent doctrinal moves can be argued analogically (“if corporations, then also…”). The paper’s critique of Citizens United is that the analogical step from institutional to constitutional-rights-bearing person is not compelled by the earlier fixation of separate legal personality.
21. Austin, Searle, and Legal Performatives
The performative structure of legal utterances—declarations, verdicts, recognitions, incorporations—is central to the analysis (Austin, 1975; Searle, 1969, 1995). What appears as description (“this company is a person”) is in fact the result of a chain of performative acts (registration, filing, judicial recognition) that constitute the state of affairs the description reports.
22. Legal Language as Programming Language
Legal categories operate on institutional facts the way programming operates on data structures. This is not a metaphor pursued to novelty; it is a diagnostic. Reading statutes and precedents as specifications of state transitions on institutional facts makes visible where transitions are under-specified, where the specification permits multiple executions, and where the running institutional program depends on defaults not stated in the specification.
23. The Mark as Access-Control Function
The Mark, understood as the access-control function computed over the Image, is what determines whether the biological person can act through the system in a given transaction. In classical administrative practice the Mark is the composite of registration, identification, and eligibility. In algorithmic administration the Mark is a score or a classification. The paper’s claim is that the two forms of Mark are continuous with each other and are subject to the same set of legitimacy tests.
24. From Signature to Score
The genealogy of authentication from handwritten signature to biometric to pure algorithmic score is analysed in Section 44; here it is enough to note that each step in the sequence transfers some of the situational grounding of the earlier form to the technical properties of the later, with a corresponding change in the failure modes.
25. Institutional Layers and Their Legal Anchors
Each layer of the representation stack (Section 16) has a specific legal anchor: Layer 1 to identity statutes, Layer 2 to registration regimes, Layer 3 to data-protection law, Layer 4 to consumer-protection and non-discrimination law, Layer 5 to administrative law. The Mark framework tracks how these anchors constrain (or fail to constrain) the transitions between layers.
26. The Mark Test
Scope and operationalisation (repeated for local reference). Decision domains: credit, employment, welfare, insurance, healthcare access, biometric authentication, automated administrative sanctions. Jurisdictions: EU/EEA, UK, Norway, and comparable regimes with contestability rights. Unit of analysis: the individual biological person subject to a specific consequential decision by an identifiable representational system over an identifiable time horizon in a specified decision domain. Disconfirmation criterion: high-risk classification is disconfirmed where the affected person can: 1) obtain the operative decision rule, 2) inspect specific inputs, 3) contest the classification before an accountable human reviewer with reversal authority, 4) obtain reversal in a documented material fraction of contested cases, and 5) access functionally-equivalent alternatives. The framework is offered as a proposed operationalisation for empirical testing, not as an already-falsifiable theory or a scoring instrument for present deployment.
The Mark test asks, of any given system in one of the covered domains: can the biological person subject to the system’s decisions exercise the five contestability elements above? A system that fails on multiple elements is a high-risk system in the sense used by the paper. A system that passes on all five is compatible with the biological-substrate priority under Tier A and, where AI-operated, with the Tier B procedural surcharge.
27. Law as Control, Constraint, and Remedy: GDPR Article 22 and SCHUFA
Article 22 of Regulation (EU) 2016/679 (GDPR) is presented here at the level of its four operative paragraphs, in the terms the Court of Justice of the European Union has recently confirmed.
Article 22 (1) confers a right on the data subject not to be subject to a decision based solely on automated processing, including profiling, which produces legal effects concerning them or similarly significantly affects them. The “solely automated” gating condition and the “legal or similarly significant effects” threshold are both operative elements of the protection.
Article 22 (2) enumerates three exceptions to the Article 22 (1) right: 1) necessity for entering into or performing a contract between the data subject and a data controller; 2) authorisation by Union or Member State law to which the controller is subject, provided that law lays down suitable measures to safeguard the data subject’s rights and freedoms and legitimate interests; 3) the data subject’s explicit consent.
Article 22 (3) requires that in the cases referred to in Article 22 (2) (a) and (c) the controller implement suitable measures to safeguard the data subject’s rights and freedoms and legitimate interests, at least the right to obtain human intervention on the part of the controller, to express one’s point of view, and to contest the decision.
Article 22 (4) constrains the use of special categories of personal data referred to in Article 9 (1) in decisions of the Article 22 (1) type, permitting such use only under Article 9 (2) (a) or (g) and with suitable measures in place.
SCHUFA Holding (Scoring), Case C-634/21, EU: C: 2023:957, judgment of 7 December 2023, held that the automated generation of a probability value concerning a data subject’s ability to service a loan constitutes an “automated individual decision” within the meaning of Article 22 (1) where a third party draws strongly on that value to determine whether to enter into a contract. Paragraphs 44-73 of the judgment set out the operative reasoning: the practical determinativeness of the score for the downstream decision brings the score itself within Article 22 (1)’s protective scope, even though a nominally-human decision-maker intervenes between score and outcome. The ruling is central for the constrain/relocate/automate typology of Section 7, because it recognises in law what that typology recognises analytically—that a strong upstream automated step can carry the decisional weight even when a formally-human downstream step remains.
28. New Public Management as Bridge to Algorithmic Administration
NPM reduces interpretive acts to metric-design and exception-handling decisions upstream and downstream of the automated step, and does not eliminate them (Hood, 1991; Bovens & Zouridis, 2002; Pollitt & Bouckaert, 2017). The move from NPM to algorithmic administration is continuous: both configurations constrain, relocate, and automate discretion; algorithmic administration extends the automation depth. Hybrid arrangements in which officials retain metric selection, exception authority, or appeal review remain interpretively active, and are the typical rather than the exceptional case. The critique developed in this paper is that when the constrain/relocate/automate combination is deep enough, the interpretive act becomes invisible to the affected party and difficult to reconstruct on appeal—a matter of degree and configuration.
29. Sociology and Legal Theory Converge
The classical sociologists (Marx, 1976; Weber, 1978; Durkheim, 1984; Simmel, 2004) supply overlapping accounts of how institutions produce durable categories that structure recognition and action. Contemporary legal theory (Hart, 2012) and sociology of law (Bourdieu, 1987; Habermas, 1996; Latour, 2005; Giddens, 1984) develop these accounts in specific legal registers. The convergence relevant here is that all treat legal categories as social facts subject to distinct kinds of contestation—doctrinal, sociological, hermeneutic, political.
30. The Accountability Gap, Formally Stated
When a decision is jointly produced by a metric designer, a model trainer, a system operator, a decision reviewer, and a formal decision issuer, the accountability of each participant is bounded by their formal role. The affected party faces a diffuse target: no single participant bears the full weight of the decision, and appeal against the composite is procedurally difficult. The paper calls this the accountability gap and treats it as the operational failure mode that the Sovereignty component of the triad is designed to address.
31. Juridical Distance and Granularity
Juridical distance is the number of institutional steps between the biological person and the officer with reversal authority. Granularity is the fineness of the decision categories at the point of contestation. High juridical distance and coarse granularity together characterise systems that are formally contestable but practically resistant to contestation. The Mark framework tracks both.
32. Maritime Hermeneutic
The pseudolaw catalogue often invokes admiralty jurisdiction to escape ordinary decisions. Admiralty jurisdiction is a real legal category—see 28 U. S. C. § 1333 for the United States federal grant of exclusive original jurisdiction over admiralty and maritime causes, and 46 U. S. C. §§ 30301-30308 for the modern statutory framework replacing the earlier Death on the High Seas Act—but it is bounded, technical, and does not extend to the imagined territorial claims made in the pseudolaw catalogue. Aquila (2018) sets out the doctrinal boundaries. The pseudolaw invocation of admiralty is a case of hermeneutic overreach: the technical categories of the source domain are imagined to license conclusions the source domain does not license.
33. The Pseudolaw Mirror
Pseudolaw catalogues are treated in this paper as the mirror of institutional overreach: the corporation acquires legal personhood through successful doctrinal argument, and the sovereign citizen tries the same move without the institutional argument (Kochi, 2025; Gillespie, 2026; Sarteschi, 2020). The failure mode of the sovereign citizen catalogue is institutional non-recognition; the failure mode of the corporate person is different—successful recognition without the institutional grounding one might have wished for. The parallel is analytical, not equivalential.
34. Three Kinds of Rejection
The paper distinguishes three kinds of rejection an institution can offer: 1) doctrinal rejection (the argument fails on the terms of the doctrine); 2) sociological rejection (the argument is doctrinally colourable but fails at the level of institutional practice); 3) legitimacy rejection (the argument would succeed doctrinally but is refused on legitimacy grounds by the surrounding social field). Distinguishing these is critical for reading a court decision correctly.
35. Formal Risk Theory Applied
Risk is a formal notion in insurance, finance, and administration. The paper deploys it here to distinguish the risk borne by the operator of an automated decision system (regulatory, reputational, financial) from the risk borne by the biological person subject to the system’s decisions (loss of access, loss of livelihood, loss of health, loss of freedom). The two are asymmetrically distributed, and the Mark framework tracks the distribution.
36. Historically Contested Personhood
The category “person” has been historically contested—women, slaves, indigenous nations, children, and now algorithmic entities have each occupied contested positions in doctrinal debates about who counts. This is a reminder that today’s determinate positive-law answers are the sedimentation of prior contests, and that the current line is not fixed by the metaphysics of the category but by the outcome of those contests.
37. The Biological Priority Principle in the Extended Constitutional Programme
The biological-priority principle in its extended constitutional form—fundamental-rights conflicts (Tier A), procedural surcharge for AI-operated entities (Tier B), and associational proportionality for unions, charities, media (Tier C)—is set out in Section 6 and applied in Principle 2 of Section 13. This section cross-refers to those statements rather than restating them, so that the principle is stated at one canonical location and applied consistently.
38. Mid-Part-II Synthesis
The sections above have introduced the six-layer representation stack (Section 16), the Mark test (Section 26), GDPR Article 22 and SCHUFA (Section 27), and the constrain/relocate/automate typology (Section 28). Together these motivate the triadic framework of Image, Mark, and Sovereignty. The full integrated statement of the triad, with the components stated at the level of definition and their interconnection made explicit, is given at Section 51. The remaining sections of Part II build up the linguistic-hermeneutic, cognitive-linguistic, empirical, and doctrinal materials on which the master statement rests.
39. Layer 0 Language and the Cybernetic Loop
The language used by an institution about a biological person feeds back into the institution’s subsequent decisions about the same person. This is a cybernetic loop in the elementary sense: outputs of the system at time t become inputs to the system at time t + 1. Layer 0 (the biological substrate) does not participate in this loop; the biological person may or may not be aware of the loop’s state at any given time. Awareness gaps between the person and the loop are the raw material of the accountability gap analysed in Section 30.
40. The Birth-Certificate Intuition
The birth certificate is the initiating document of the Image. Its status as institutional fact (Searle, 1995) is uncontroversial; its role in the subsequent trajectory of the Image is decisive. The paper does not take the pseudolaw view that the birth certificate is a securitised instrument; it treats the birth certificate as the anchoring institutional performative that initiates the Image and permits the subsequent thickening documented in Section 19.
41. Entextualization and Ontological Displacement
41.1. Ricoeur’s Distanciation Applied to Automated Decisions
An automated decision is the maximum case of entextualization in Ricoeur’s (1981) sense: the decision is inscribed as a score, stored, and read at points and in situations arbitrarily distant from the situation of its production. The situational grounding that would allow a reader to interpret the decision charitably is systematically stripped. This is the ontological displacement: the decision, once produced, exists as an inscription whose subsequent life is largely independent of the situation of its production.
41.2. Institutional Uptake and the Reader’s Situation
The reader of the inscription—the case officer, the appeal reviewer, the successor system—supplies from their own situation the interpretive frame in which the inscription is read. The reader’s situation is different from the producer’s; the inscription travels between situations and accretes new readings without leaving trace of the earlier ones. The temporal asymmetry (Section 46) is the consequence.
42. Proposed Mark Threshold: Formal Statement
Scope and operationalisation (repeated for local reference). The Mark threshold is proposed for empirical study within the decision domains, jurisdictions, and unit of analysis specified in Sections 8 and 26, subject to the disconfirmation criterion set out there. The illustrative scoring table below is illustrative and pre-empirical; measured application requires exercising the five-element disconfirmation procedure on real systems in the covered domains. The framework is not offered as an already-falsifiable theory or as a scoring instrument for present deployment.
The Mark threshold M is written schematically as M = f(D, E, O, I, C, S), where: D is decision domain (weighted by consequence severity); E is the entextualization gradient (higher for fully-automated, lower for face-to-face); O is the ownership and operational locus of the system (higher for concentrated operator control, lower for distributed or public control); I is the inspectability of inputs (higher when opaque); C is the contestability of the output (lower when appeal is procedurally hostile); S is the substitutability of the decision-mediated good (higher when the good is essential and unsubstitutable). A high M indicates a system for which the Sovereignty component of the triad is not adequately preserved and for which the disconfirmation criterion is unlikely to be satisfied.
43. Hydraulic Semiotics
43.1. The Water Metaphor in Financial Discourse
Financial discourse relies extensively on hydraulic metaphors: liquidity, flow, flood, drought, pooling, sinking. This is well-documented in cognitive-linguistic scholarship (Lakoff & Johnson, 1980; Charteris-Black, 2004; Kövecses, 2010) and in economic-anthropological accounts of monetary reasoning (Peebles, 2010). The metaphor is not decorative; it structures the reasoning available to participants in the financial system.
43.2. Legal Inheritance of the Metaphor
The hydraulic metaphor migrated from financial into legal discourse and now structures how courts, regulators, and legislators reason about financial flows. This is a genuine cognitive-linguistic inheritance—not an occult residue of maritime jurisdiction, as the pseudolaw catalogue supposes.
43.3. Why This Is Not Maritime Law
Admiralty and maritime jurisdiction (28 U. S. C. § 1333; 46 U. S. C. §§ 30301-30308) is a distinct doctrinal category with its own boundaries, procedures, and remedies (Aquila, 2018). The presence of hydraulic metaphor in financial reasoning is a fact about metaphor, not a fact about jurisdiction. The pseudolaw catalogue commits a category error by inferring the latter from the former.
43.4. Consequences for the Analysis
Correctly identifying hydraulic metaphor as cognitive-linguistic inheritance rather than as jurisdictional residue clarifies the field: the substantive institutional overreach documented in Part I is not to be conflated with the imagined jurisdictional escape of the pseudolaw catalogue.
44. Empirical Case: Legal Language Complexity
Two empirical studies bear directly on the paper’s framing of institutional-legal language as, in part, a controlled-meaning technology. Martínez, Mollica, and Gibson (2022) demonstrated that lay readers process legal texts less well than functionally-equivalent plain-language renditions, and that the processing difficulty is driven by centre-embedded syntactic structures and archaic terms rather than by specialised legal concepts. Martínez, Mollica, and Gibson (2023) extended the finding to trained lawyers, who also process plain-language versions more accurately and prefer them. Together these dislodge a common defence of legalese (that its complexity tracks conceptual necessity) and support the interpretation of legal-language complexity as, in significant part, a professional-differentiation device rather than a conceptual necessity.
45. Institutional Semantic Capture
Institutional semantic capture is the operation by which an ordinary-language term is converted to a technical-legal use whose meaning is defended against the ordinary use. Four senses of “person”—biological, moral, legal, algorithmic—coexist in institutional practice; each has its constituency and its favoured argumentative moves. The capture succeeds when the technical use displaces the ordinary use at the point of decision.
46. Temporal Asymmetry
46.1. The Inscription Outlives the Situation
The inscribed decision persists after the situation of its production has passed. The biological person’s later state—corrected credit history, changed employment, revised medical status—must be readmitted to the Image through further institutional acts, and the readmission is not automatic. The default is persistence of the earlier inscription.
46.2. Reversal Rates as an Indicator
Documented reversal rates for appealed decisions are an indicator of the practical effectiveness of contestability. Low reversal rates in a system with formal appeal rights indicate either that most decisions are correct (rarely the sole explanation) or that the appeal procedure is procedurally hostile. The disconfirmation criterion of Section 8 treats documented reversal rates as the fourth element.
46.3. Compounding of Errors
Uncorrected errors in the Image compound across decisions. A credit-score error affects insurance pricing, which affects employability, which affects welfare-eligibility, which affects health access. The Mark framework tracks the compounding potential of specific error types across the covered domains.
47. Representation Rights and the Empirical Programme
The paper proposes a research programme with three components: 1) doctrinal analysis of contestability rights and their operationalisation in current EU/EEA, UK, and Norwegian regimes; 2) empirical study of documented reversal rates across the covered domains; 3) comparative-institutional analysis of hybrid configurations under the constrain/relocate/automate typology. The Mark framework provides the analytical scaffolding; the disconfirmation criterion provides the measurement target.
48. Peer-Review Correction Note
The revisions incorporated in this typeset version are documented in the accompanying cover letter. The paper’s substantive positions are unchanged; the revisions clarify epistemic register, add pinpoint authorities, qualify claims previously stated categorically, and consolidate the master statement of the triad into a single canonical location (Section 51).
49. Legalese as Professional Shorthand
49.1. Shorthand and Its Costs
Professional shorthand is efficient among professionals and costly to lay readers. The Martínez et al. (2022, 2023) findings quantify a portion of the cost. The paper’s argument is not that legalese should be abolished but that where it operates at the boundary between the profession and the affected biological person, the professional-shorthand justification does not carry.
49.2. Boundary Effects
At the boundary—consumer contracts, welfare-eligibility notices, appeal instructions, administrative sanctions—professional shorthand is a barrier to the disconfirmation criterion. Sovereignty in the triadic sense requires that the affected person understand what they are contesting.
49.3. The Choice of Terms
The choice of terms in a public-facing notice is a hermeneutic choice with practical consequences. Institutional actors bear responsibility for that choice; the paper treats it as part of the constrain/relocate/automate typology (Section 7).
50. What the Theory Explicitly Rejects
The paper explicitly rejects: 1) the pseudolaw view that ordinary administrative decisions can be escaped by invocation of maritime jurisdiction, “strawman” doctrine, or unilateral declarations of sovereignty (Kochi, 2025; Gillespie, 2026; Sarteschi, 2020); 2) the view that legal personhood has no positive-law content within a given jurisdiction (statutes and precedent supply determinate answers in decided cases); 3) the view that all corporate personhood should be abolished (associational and expressive functions are protected by Tier C); 4) the view that automated decision-making is inherently illegitimate (the disconfirmation criterion permits well-designed automated systems to satisfy the framework); 5) the view that hermeneutic instability is a defect of legal categories (it is a productive condition of their institutional life); 6) the view that biological substrate confers metaphysical entitlements independent of institutional recognition (the biological-substrate priority is a normative-constitutional principle, not a metaphysical claim).
51. The Full Integrated Statement of Image, Mark, and Sovereignty
This section carries the paper’s canonical master statement of the triadic framework. The Abstract and Sections 15 and 38 cross-refer to this section rather than restating the definition.
Image. The Image of a biological person is the representation of that person maintained by an institutional or algorithmic system across the six layers of Section 16, initiated at birth by the birth-certificate performative (Section 40), thickened by successive administrative acts (Section 19), and read against varying rule-sets in specific decision-domain contexts (Section 8). The Image is not the biological person; it is an inscription, in Ricoeur’s (1981) sense, with the properties of iterability and situational distance that inscriptions have (Section 41).
Mark. The Mark is the access-control function computed over the Image, determining what the biological person can and cannot do through the system in a specific decision domain (Section 23). Written schematically as M = f(D, E, O, I, C, S), where D is decision domain, E is entextualization gradient, O is ownership/operational locus, I is inspectability of inputs, C is contestability of outputs, and S is substitutability of the mediated good (Section 42). The Mark is offered as a proposed operationalisation for empirical testing within the scope specified in Sections 8, 26, and 42, subject to the disconfirmation criterion set out there.
Sovereignty. Sovereignty in the triadic sense is the biological person’s preserved capacity for contestation, correction, refusal, exit, and remedy with respect to the Image and the Mark, exercised through the five elements of the disconfirmation criterion (Sections 8 and 26). Sovereignty is preserved when the affected person can 1) obtain the operative decision rule, 2) inspect the specific inputs used, 3) contest the classification before an accountable human reviewer with authority to reverse, 4) obtain reversal in a documented material fraction of contested cases, and 5) access one or more functionally-equivalent alternatives for the underlying good. Sovereignty is the constitutional-normative complement to the biological-substrate priority (Section 6, tiered) and is the operational target of Principle 3 (Section 13).
The triad—Image, Mark, Sovereignty—is offered as an analytical structure for tracking how consequential decisions in the covered domains are mediated by representational systems and where in the mediation the possibility of contestation is preserved or lost. It is proposed rather than already-falsifiable; the operationalisation criteria of Sections 8, 26, and 42 specify what would count as empirical satisfaction and disconfirmation. Related work in the same corpus (Pater, 2026a, 2026b, 2026c, 2026d) develops parallel analytical structures for the Mark, Number, Tartaria, and Strawman analyses.
Appendix: The Central Distinction
A legal person is not necessarily a human person; a legal fiction is not necessarily a falsehood; corporate separateness is not universal immunity; autonomous decision is not autonomous legal personality; registration is not consent; secrecy is not proof of conspiracy; and the label “pseudolegal” is sometimes an evidentiary standard and sometimes a legitimacy filter—the two must never be confused. Positive-law questions in decided cases have determinate answers; the interpretive frame in which future cases will be decided is what remains contestable, and it is the target of this paper.
Cases and Statutes Cited
Cases
Citizens United v. Federal Election Commission, 558 U. S. 310 (2010).
Salomon v. A. Salomon & Co Ltd, [1897] AC 22 (H. L.).
Santa Clara County v. Southern Pacific Railroad Co., 118 U. S. 394 (1886).
SCHUFA Holding (Scoring), Case C-634/21, EU: C: 2023:957 (CJEU, judgment of 7 December 2023).
Trustees of Dartmouth College v. Woodward, 17 U. S. (4 Wheat.) 518 (1819).
Legislation, Treaties, Codes, and Official Documents
Comunicado Oficial No. 149 (Presidencia de la Nación Argentina, 18 June 2026).
Data Protection Act 2018 (United Kingdom).
Draft General Companies Law of Argentina, Title IV (Automated Companies and DAOs) (2026).
Letters Patent Chartering the East India Company (Elizabeth I, 31 December 1600).
Lov om behandling av personopplysninger (Personopplysningsloven), 15 June 2018 nr. 38 (Norway).
Octrooi Chartering the Vereenigde Oostindische Compagnie (Staten-Generaal der Verenigde Nederlanden, 20 March 1602).
Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 (General Data Protection Regulation), Article 22(1)-(4).
Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 (AI Act).
28 U. S. C. § 1333.
46 U. S. C. §§ 30301-30308.