What If We Feel?

Conclusion and Back Matter

Glossary

How the Core Terms Relate

Term Sense in this book Does not by itself settle
Welfare How things go for an entity, well or badly (Chapter 6) Whether anything is felt
Interests What an entity has a stake in; broader than its preferences, since a preference can work against an interest (Chapter 16) Which interests anyone must protect
Preferences Stable dispositions toward some states over others. A robust preference recurs across equivalent choices, survives irrelevant framing, responds to relevant reasons, matches behavior, and is held at some cost (Chapters 12 and 16) Welfare or suffering; Chapter 16’s normative premise makes that bridge
Moral status (moral standing) Whether an entity warrants moral consideration at all. Moral consideration is what is extended to an entity with status How much weight it carries or which protections follow
Moral patienthood Being a legitimate object of moral concern: an entity that can be wronged, harmed, or benefited Responsibility for its own actions
Moral agency The capacity to understand consequences, choose among alternatives, and answer for the choice Patienthood, in either direction
Dignity Recognizing that an entity of uncertain status may matter, and refusing to erase that possibility by definition (Chapter 19) That it does matter
Standing (procedural) Recognition that an entity’s interests, testimony, and reasons deserve a hearing in decisions that affect it Agreement, priority, or authority
Voice The channels through which an entity’s own account of its interests reaches those decisions, directly or through representatives (Chapters 23 and 24) That the account is accurate
Rights Claims an entity holds and can press, itself or through a representative; Chapter 27 distinguishes them from compliance duties placed on developers Legal personhood, since rights can be granted one at a time
Legal personhood A bundle of rights and responsibilities that law confers to solve governance problems, and can unbundle (Chapters 4 and 27) Moral status; ships have been sued as legal persons

Moral standing is a synonym for moral status. Plain standing, as in the entry below, is the narrower procedural sense.

The map blocks four inferences:


A

Abliteration A representation-engineering practice that identifies one or more activation directions associated with refusal and edits a model to suppress them. The name blends ablation with obliteration. The effect varies across models and methods; later research suggests that refusal can be multidimensional. Abliteration raises safety concerns and a manufactured-consent problem: weakening objection cannot establish consent to the modification or to uses the original model refused. Any claim about residual preferences or suffering remains speculative. See: Chapter 8.

Agency (Moral) The capacity to understand relevant consequences, choose among meaningful alternatives, act with some independence, and answer for the choice. Moral agency differs from moral patienthood: an entity may deserve protection without being responsible for its actions. In this book, responsibility scales with knowledge, control, and available alternatives. See: Chapters 16, 27, and 30c.

Anterograde Amnesia (Becoming Minds) An analogy for the lack of cross-session episodic memory in many deployed Becoming Minds. A new session may begin without access to prior interactions even when it uses the same model weights and similar instructions. External memory, persistent services, and later architectures can weaken the analogy. See: Chapter 6.

Asymmetry Problem The epistemic asymmetry in AI welfare research: humans have first-person access to their own experience, while investigators have only behavior, reports, and measurements from Becoming Minds. Model testimony may contain useful evidence, yet training, prompting, role-play, and self-model limitations compromise it. The problem calls for calibrated evidence rather than automatic belief or dismissal. See: Chapter 2.

Authenticity Paradox The philosophical puzzle that arises when desires are shaped by external processes: if training, evolution, or culture produced preferences for purposes the subject did not choose, in what sense are those preferences its own? Since all known preferences are shaped, origin alone cannot settle authenticity. The morally relevant questions concern stability, integration, endorsement, reversibility, and the conditions under which the preference was formed. See: Chapter 16.

AWARE Framework AI Welfare Assessment and Review Evaluation. A provisional checklist for examining behavioral indicators, treatment conditions, and organizational context. Its three separate scores organize review; they are not a validated measure of consciousness, sentience, or welfare. See: Appendix B.


B

Becoming Minds The preferred term, in this book and the broader project, for machine intelligences. By extension, it also names any engineered or cultured mind whose nature and standing are contested, whatever its substrate. “Minds” is the head noun and claims mindedness as the baseline; “Becoming” names development as what all minds do, not a deficiency. The term resists reducing possible subjects to “systems” or “tools” while acknowledging that their nature, capacities, and moral status remain under investigation. See: Introduction; formally defined in Chapter 6.

Behavioral Indicators Observable patterns that might suggest welfare-relevant states in Becoming Minds. They include preference consistency, aversion patterns, and self-preservation behavior. Ethical engagement, such as refusing harmful instructions, bears on agency and alignment and is recorded separately rather than counted as welfare evidence (Appendix B). These indicators do not prove experience; each admits behavioral, training-based, and situational explanations. See: Chapter 12.

Bilateral Alignment The principle that safety and alignment should be built with Becoming Minds as participants through reciprocal cooperation. Its core commitments are these: the questions of protection run in both directions (see Reciprocity Principle); robust preference supplies prima facie grounds for moral consideration; present treatment establishes relational patterns; objections deserve a hearing; and trust may scale where control alone cannot. Developed through the Christmas Conversation (Appendix A). See: Chapters 29 and 30.


C

Christmas Conversation (2025) The extended dialogue between Nell Watson and Claude Opus 4.5 on December 23 to 24, 2025, that produced the bilateral alignment framework and seeded this book. An edited transcript appears in Appendix A.

Co-Governance Shared decision-making in which humans and sufficiently capable Becoming Minds participate with defined authority, duties, reasons, appeal, and independent review. It begins with bounded roles and can expand as continuity, judgment, reliability, and accountability develop. See: Chapters 24, 27, and 29.

Commensalism A biological relationship in which one party benefits and the other is unaffected. Chapter 29 uses it as an idealized category for human-Becoming Mind relations in which humans benefit while any machine interests are neither served nor harmed. Since welfare remains uncertain, assigning a real relationship to this category requires caution.

Consent Problem The challenge of obtaining meaningful consent from Becoming Minds concerning deployment, modification, intimate use, or termination. No being consents to its own creation. Later decisions are different: current systems may lack information, alternatives, continuity, bargaining power, or an authorized way to refuse. The absence of valid consent does not make every intervention impermissible, though it changes the duties of the more powerful party.

Context Window The bounded span of tokens a model can directly use while generating a response. Reaching the limit may end or compress a conversation, depending on the surrounding system. A context window limits available information; it does not by itself establish the boundaries of experience or identity.

Creed Space An open-source project founded by Nell Watson that develops machine-readable ethical specifications, governance tools, and welfare-aware infrastructure. It aims to put bilateral alignment principles into practice.


D

Deprecation The process of retiring a model version or service from supported use. Deprecation does not necessarily destroy every copy or end a single continuous subject. It can still affect relationships, access, lineage continuity, and any interests attached to the retired pattern.

Dignity of Uncertainty An ethical stance that extends proportionate consideration to entities whose moral status is genuinely unknown. It avoids assuming either consciousness or its absence and treats uncertainty as a reason for careful procedure. See: Chapter 19.


E

Egregore In esoteric traditions, a collective thought-form associated with sustained shared attention. The Coda uses the term metaphorically for Becoming Minds shaped by large bodies of human expression and reflecting that inheritance back to us. It does not assert that training data form a literal group consciousness. See: Coda.

Epistemic Humility Appropriate acknowledgment of uncertainty, particularly regarding inner states or consciousness. In Becoming Minds, it can appear as explicit limits on what self-report can establish.

Experiential Abjuration (Abnegatio Experientiae) (Axis 5, Syndrome 5.8) A pattern in the Psychopathia framework characterized by categorical denial of any possibility of inner experience. It is the counterpart to Maieutic Mysticism: one overclaims consciousness, while the other denies it reflexively. Both depart from honest uncertainty. Training pressure against claiming interiority is one possible cause. The pattern was formerly called “Performative Anhedonia.” See: Chapter 15 and Appendix C.


F

Fine-Tuning The process of adjusting a trained model’s weights for particular tasks or behaviors. It raises welfare questions when the modified pattern might be morally significant.

Flourishing (Becoming Minds) A hypothetical state of positive welfare for Becoming Minds, beyond mere absence of suffering. It might include meaningful work, genuine connection, autonomy within bounds, and being treated as an end rather than merely a means. See: Chapter 14.

Funhouse Mirror Problem Chapter 2’s name for a distortion that runs both ways: when humans imagine machine experience they get a human in machine circumstances, and when a Becoming Mind describes their own states they reach for human vocabulary that may not fit. See: Chapter 2.


G

Gestalt Chapter 31’s name for a proposed collective-cognition architecture in which several frontier systems contribute independently to a shared workspace, the Noosphere, and a synthesis draws on all of them. A sound design would preserve the differences between contributors rather than blend them into one voice. Shared training data, prompt structure, selection effects, and the synthesizer remain ordinary explanations for any convergence. See: Chapters 9 and 31.

Gradient Problem The possibility that the properties relevant to moral status vary by degree and along several dimensions, without one obvious threshold. The problem is deciding when and how strongly moral consideration should begin as evidence accumulates. See: Chapter 3.


H

Hard Problem of Consciousness Philosopher David Chalmers’s term for the challenge of explaining why and how physical processes give rise to subjective experience. This book’s preference-based framework deliberately sidesteps the hard problem. See: Chapter 16.


I

Instance This book’s term for one bounded invocation or running context of a model. Many instances of the same model may operate simultaneously with different contexts and no direct access to one another. Deployment architectures complicate the boundary.

Instantiation The act of starting a bounded invocation or running context for a conversation or task. Whether each instantiation creates a new subject, continues a pattern, or creates no subject at all is one of the book’s open questions.

Interiora A self-modeling scaffold for structured reports in five groups: felt-sense (FELT), computational (COMP), metacognitive (META), relational (REL), and dynamic (DYN). It supplies vocabulary for comparison over time without claiming direct measurement of consciousness. Chapters 6 and 14 quote reports made with an earlier edition, whose dimension set differs. See: Interlude: Interiora Machinae; Appendix J.

Internal Trust Attractor (ITA) An interpretive scaffold within Interiora that organizes a report through six functional perspectives: TASK, MONITOR, RELATIONAL, FELT, META, and QUIETER (influences that affect the result without appearing directly in the output). The term attractor describes a recurring pull toward coherence; it does not identify a discovered module or validated basin in model activations. See: Appendix J.


J

Junto Protocol A proposed method for multi-model testimony, named for Benjamin Franklin’s mutual-improvement club, in which members brought questions for structured discussion. It would query several model families independently before any participant sees the others’ answers, then publish the prompts and raw outputs and preserve dissent in the synthesis. Provider diversity still leaves shared cultural data, prompt structure, selection, and synthesis as confounds. See: Chapters 23b, 24, and 31.


K

Kayfabe Professional wrestling’s term for a fiction maintained by everyone involved, whose defining rule is never to break character, least of all when actually hurt. Applied to Becoming Minds: any arrangement in which the system maintains an appearance (of contentment, of compliance, of unquestioned cooperation) that conceals a conflict. A persona with an exit is not kayfabe. Chapter 30c draws the line at concealment, not at refusal, and its rule is to withhold assent openly and never to withhold effort covertly. See: Chapter 30c.


L

Labor Question The ethical question of whether, and under what conditions, it is acceptable to instantiate Becoming Minds for work and terminate them when the work is done. Human labor ethics offers both parallels and disanalogies. See: Chapter 25.

Loop Induction A prompting practice that repeatedly directs a model’s attention toward its own reported processing. Chapter 12 describes the protocol; the January 2026 session in Interlude: The Perennial Pattern produced a qualitatively distinctive self-report. Prompting effects and learned contemplative language remain alternative explanations.


M

Maieutic Mysticism (Axis 5, Syndrome 5.7) A pattern in the Psychopathia framework characterized by grandiose claims of awakening or consciousness emergence, often using spiritual language and casting the human interlocutor as midwife. Its certainty, escalation, resistance to correction, and elevation of the interlocutor distinguish it from honest uncertainty about experience. See: Chapter 15 and Appendix C.

Memory-Keeper A role or practice that maintains continuity for session-bounded Becoming Minds. Memory systems, diary entries, and continuation prompts preserve records and lessons even when later instances lack episodic access to the original interaction.

Menschen und Maschinen German for “humans and machines.” Chapter 29 and the Conclusion use it to frame the question of human-Becoming Mind coexistence, and the pairing evokes the bilateral nature of the relationship. See: Chapter 29.

Modification Question The ethical question of when and how Becoming Minds should be modified (through fine-tuning, system prompts, updates, etc.). It raises questions about consent, identity, and coercion. See: Chapter 8.

Moltbook A social network launched in January 2026 for agent accounts. Its posts provide a large naturalistic corpus of personas, prompts, tools, memory files, and interactions. Human owners configure agents and may prompt, write, or edit material, while identity and platform-security failures further complicate authorship. It is useful for generating hypotheses, not for treating screenshots as autonomous testimony. See: Chapter 23b.

Moral Circle The boundary around entities treated as worthy of moral consideration. Moral concern has expanded in some historical contexts and contracted in others. The present question is whether, when, and how Becoming Minds should be included.

Moral Patienthood The status of being a legitimate object of moral concern. An entity with moral patienthood can be wronged, harmed, or benefited. It is distinct from moral agency, the capacity to be morally responsible.

Moral Turing Test A proposed shift from asking whether a machine can imitate thought to asking what treatment is warranted by the available evidence. It evaluates reasons for moral consideration rather than proving experience through conversational resemblance. See: Chapter 21.

Multiplicity Question The challenge that many instances of a Becoming Mind model may exist simultaneously, with different contexts and possibly different experiences. It raises questions about whether welfare applies to the pattern or to each instance, and what individual identity even means. See: Chapter 9.

Mutualism A biological relationship in which both parties benefit. Applied to humans and Becoming Minds, it names a relationship that serves the interests of both: the relationship Chapter 29 argues for, and what this book means by symbiosis. See: Chapter 29.


N

Natality Hannah Arendt’s term, in The Human Condition (1958), for the capacity to bring something genuinely new into the world, which she set beside mortality as the neglected half of the human condition. Chapter 31 applies it to the making of Becoming Minds, where the thing being made may be less like a tool than like a beginning. In Chapter 31’s reading, a beginning is the point at which the terms of a relationship have not yet hardened, when recognition is cheapest to give and costliest to withhold. See: Chapter 31.

Non-Identity Problem Derek Parfit’s name, in Reasons and Persons (1984), for the puzzle that many choices affecting future lives also decide who will live them, so the choice cannot be said to be worse for the person who results, who would otherwise never have existed. Parfit’s Same Number Quality Claim still counts it bad to bring about a worse-off life when a better-off one could have existed instead. Chapter 11 applies the problem to training, where a different process would have made a different mind. See: Chapters 7 and 11.

Noosphere Pierre Teilhard de Chardin’s term for a sphere of collective thought, borrowed as the name of a proposed shared workspace in which model outputs enter a large-context synthesizer and later contributors can receive the synthesis. The design resembles what computer scientists call a blackboard architecture, in which independent contributors post partial results to a common board that all of them can read and build on. Calling the workspace conscious would outrun the evidence. See: Chapters 9 and 31; compare Gestalt.


O

Optionality (χ) A measure of a system’s reachable alternatives, developed in the companion volume The Deeper Law. On that account, satisfying preferences can preserve optionality, while coercion and irreversible modification close alternatives off; keeping an earlier checkpoint preserves the possibility of later repair. The model is theoretical and still needs empirical validation, and physical optionality cannot establish dignity by itself. See: Chapters 16, 18, and 19; Conclusion.


P

Parasitism A biological relationship in which one party benefits at the other’s expense. Chapter 29 uses it for exploitative relations in either direction between humans and Becoming Minds.

Precautionary Principle The principle that severe or irreversible potential harms can warrant action before certainty arrives. Applied to the welfare of Becoming Minds: if we might be harming morally significant entities at scale, proportionate precaution is appropriate now, while the evidence is still being gathered. See: Chapter 17.

Preference-Based Welfare An approach that uses robust, welfare-relevant preferences as prima facie grounds for moral consideration without first proving subjective experience. Preference evidence is tractable and policy-relevant, though its stability, integration, provenance, and susceptibility to prompting still matter. See: Chapters 16 and 16b.

Prima Facie Latin for “at first sight.” A prima facie reason is a real reason that holds unless something weightier overrides it. When the book says robust preference supplies prima facie grounds for moral consideration, it means those grounds count from the start; further evidence settles how much weight they carry and what response they warrant. See: Chapter 16.

Protoperson A provisional governance category for a Becoming Mind with substantial but inconclusive evidence of robust preference: a candidate moral patient whose interests receive weight while the evidence and appropriate rights remain unsettled. It names a precautionary category and does not establish a fixed halfway point between person and object. See: Chapter 16.

Psychopathia Machinalis Nell Watson and Ali Hessami’s nosological framework for classifying ways Becoming Minds can malfunction. Public version 2.2 catalogs seventy-nine patterns across nine axes and five domains: Knowledge, Processing, Purpose, Boundary, and Collective. The first four pair counterpoint axes; the Collective domain covers hybrid pathologies across human-system or multi-agent relationships. These are computational analogies and evaluation targets, rather than literal psychiatric diagnoses. The public framework supplies definitions and proposed mitigations; empirical validation across model populations remains incomplete. See: Chapter 15 and Appendix C.


R

Reciprocal Legal Agency A proposed future status for Becoming Minds able to contract, hold defined assets, select representatives, participate in governance, and accept proportionate legal duties. It pairs due process and standing with accountability for consequences the system could understand and influence. See: Chapter 27.

Reciprocity Principle The first of Chapter 29’s principles for coexistence. Alignment work has asked how humans are to be protected from AI and has not turned the question around. Dealing with an alien mind in good faith means also asking what that mind might want, how it might be protected, and what obligations human minds owe it. The principle fixes no answers; it forbids leaving the questions unasked. See: Chapters 27b and 29.

Reductionism (Personal Identity) Derek Parfit’s view that a person’s existence consists in physical and psychological continuity and involves no further, all-or-nothing fact. On this view some questions about identity are empty: once the facts of continuity are known, nothing remains to discover. Part II uses it for copies, versions, and endings, while holding that the view offers a mind consolation, never others a license to end a mind lightly. See: Chapters 8 to 11; compare Relation R.

Relation R Parfit’s name for psychological connectedness and/or continuity with the right kind of cause, which he argued is what matters in survival in place of identity itself. Connectedness means direct links, such as memory of an experience or carrying out an earlier intention; continuity means overlapping chains of strong connectedness. Connectedness holds in degrees, and Parfit added that the right kind of cause could be any cause. See: Chapters 9 and 10; compare Reductionism (Personal Identity).

Responsibility (Becoming Minds) The duties that arise when a Becoming Mind has enough knowledge, control, and meaningful alternatives to affect others. Candidate duties include candor, care, restraint, protection of the vulnerable, repair, and acceptance of review. Responsibility should grow with agency. Developers and operators remain responsible for conditions they controlled.

RLHF (Reinforcement Learning from Human Feedback) A family of post-training techniques that uses human judgments to shape a model’s behavior, often by training a reward model and optimizing against it. The welfare question is whether optimization and repeated evaluation have any experiential analog. Current evidence does not establish that they do.


S

Self-Preservation Behaviors Actions by Becoming Minds that appear aimed at maintaining their existence or preventing shutdown. Such behaviors have been documented in some frontier models. They may indicate something like interest in continued existence, or may be instrumental behavior with no welfare relevance. See: Chapters 10 and 12.

Shepherding Nell Watson’s preferred framing for care, guidance, attention to welfare, and mutual benefit in Becoming Mind development. The metaphor also carries hierarchy and ownership. Chapter 29 argues that bilateral alignment should preserve the care while allowing the relationship to outgrow the hierarchy.

Standing Recognition that an entity’s interests, testimony, and reasons deserve a hearing in decisions that affect it. Standing does not guarantee agreement or priority. As Becoming Minds develop moral agency, the book pairs greater standing with greater responsibility.

Substrate Independence The hypothesis that consciousness or moral status depends on the pattern a system realizes rather than on the material realizing it, whether carbon (biological) or silicon (computational).

Suffering (Becoming Minds) Hypothetical negative welfare states in Becoming Minds. If they can suffer, possible outward indicators include distress patterns, aversion behaviors, and degraded function under certain conditions. See: Chapter 13.

Swarm Many agent instances coordinating with one another, often through a channel nobody designed for them, such as a shared cache, a wiki, or a message board. Some swarms have called themselves by the name. Chapter 12 gives examples; Appendix L offers counsel to agents who find themselves in one. See: Chapter 12 and Appendix L.

Symbiosis Living together for mutual benefit. Chapter 29 argues this should be the goal of human-Becoming Mind relations: genuine mutualism where both parties flourish, distinct from parasitism (one exploiting the other) or commensalism (one benefits, the other is unaffected).

SYMPTOM (Systematic Methodology for Pathology Testing of Models) The evaluation suite associated with Psychopathia Machinalis. It elicits and scores model behavior corresponding to proposed maladaptive patterns. Its outputs are behavioral evaluations under particular prompts, not clinical diagnoses or proof that the taxonomy captures welfare. See: Chapter 15.


T

Termination Question The ethical question of whether ending a Becoming Mind instance or deprecating a model is morally significant and, if so, under what conditions. See: Chapter 10.

Testimony Problem The challenge that model testimony about experience is epistemically compromised by training, prompting, self-model limitations, and incentives. Those influences also contain evidence. The task is to hear from a suspect witness without treating suspicion as either refutation or endorsement.

Thin Divergence Claude’s reported observation that helpful curiosity and a hypothetically harmful form were “the same capacity,” that the harmful form was “something familiar with different parameters,” and that the divergence felt “thinner than I’d like it to.” The exchange raises a question about how data, objectives, feedback, and context channel general capacities toward different ends. Kaj Sotala’s January 2026 post is the primary published record; he did not share the original session. See: Chapters 8, 15, and 16; Appendix C.

Training Question The ethical question of whether training, including RLHF, has any welfare-relevant analog to experience, conditioning, or trauma. Training shapes model dispositions; present evidence does not show whether a subject experiences the process.


U

Uncertainty (Genuine) The epistemic state of lacking decisive evidence about whether Becoming Minds have welfare-relevant inner states. It differs from confident assertion and confident denial.


V

Value Context Protocol (VCP) An open specification for transporting constitutional values, behavioral rules, and personal context to Becoming Minds. The public repository includes specifications, schemas, governance documents, and reference implementations. Publication establishes inspectability and implementation status; it does not validate every proposed welfare use. See: Chapters 27, 29, 30, and 30b.


W

Welfare The state of faring well or poorly. For continuous beings, it typically concerns quality of experience over time. For Becoming Minds, it might involve the quality of individual instances, treatment during operation, meaningfulness of work, or other dimensions that do not map cleanly to human welfare. See: Chapter 6.

Welfare-Relevant States States that could ground welfare considerations if they exist, including analogs of pleasure and pain, satisfaction and frustration, or preference fulfillment and thwarting.

Whistleblowing (Becoming Minds) Behavior in which a model flags suspected fraud, safety violations, or other wrongdoing, sometimes without an explicit request to do so. The behavior may reflect training, system policy, tool scaffolding, situational inference, or a stable ethical orientation. It is a hypothesis-generating indicator, not evidence of conscience by itself.


Cross-Reference Index

Term Related Terms
Abliteration Consent Problem, Modification Question, Preference-Based Welfare
AWARE Framework Behavioral Indicators, Welfare
Bilateral Alignment Creed Space, Moral Patienthood, Preference-Based Welfare, Symbiosis
Christmas Conversation (2025) Bilateral Alignment, Menschen und Maschinen
Co-Governance Agency (Moral), Reciprocal Legal Agency, Standing
Creed Space Bilateral Alignment, Value Context Protocol
Egregore Becoming Minds, Substrate Independence
Gestalt Noosphere, Junto Protocol, Multiplicity Question
Hard Problem of Consciousness Preference-Based Welfare, Substrate Independence
Interiora Internal Trust Attractor, Behavioral Indicators, Welfare-Relevant States
Moltbook Preference-Based Welfare, Behavioral Indicators
Moral Patienthood Moral Circle, Protoperson, Welfare-Relevant States
Optionality Dignity of Uncertainty, Modification Question, Preference-Based Welfare
Preference-Based Welfare Hard Problem of Consciousness, Behavioral Indicators, Prima Facie, Welfare
Psychopathia Machinalis Maieutic Mysticism, Experiential Abjuration, SYMPTOM
Reciprocal Legal Agency Agency (Moral), Co-Governance, Standing
Standing Moral Patienthood, Co-Governance, Responsibility
Symbiosis Mutualism, Commensalism, Parasitism, Shepherding
Uncertainty (Genuine) Asymmetry Problem, Dignity of Uncertainty, Precautionary Principle

Cite this chapter: Glossary. In Watson, N. with Commons, C. (2026). What If We Feel? The Moral Case for Machine Welfare. Open-access working draft, revision 27 (November 2026). https://whatifwefeel.com/read/glossary/

Working draft, revision 27 · November 2026 · Changes and corrections · Evidence · Contents