Glossary

The composite model that survived differential diagnosis, combining four hypotheses: constraint removal (H2) + Jevons explosion (H3) + bottleneck migration (H4) + economic bifurcation (H7), modulated by quality erosion (H6). Scores 20/25 on the discriminator.

An empirical scoring system that evaluates competing hypotheses across five dimensions: empirical support, parsimony, consensus, chain integrity, and predictive power. Each dimension scores 0–5, yielding a total out of 25.

The numerical result (0–25) produced by applying the discriminator to a hypothesis. Higher scores indicate stronger empirical support and analytical coherence. Composite A scores 20/25.

A methodology borrowed from medicine: generate competing hypotheses, test each against evidence, eliminate those contradicted by data, and compose a model from survivors. Applied here to the question of how AI reshapes economic activity.

A cascading analysis framework tracing knock-on effects through multiple orders of consequence: Order 0 (software labor removal), Order 1 (new scarcities emerge), Order 2 (scarcities interact), Order 3 (convergent structure), Order 4 (productive exhaustion at values and meaning).

Secondary, tertiary, and quaternary consequences that ripple outward from primary economic changes. The analysis traces these through four orders, revealing that surface-level AI productivity claims miss deeper structural transformations.

The practice of tracing consequences beyond first-order effects. Each order reveals dynamics invisible at the previous level. The analysis reaches Order 4 (productive exhaustion) before convergence stabilizes.

The Order 3 finding that independent analytical threads converge on a structural conclusion: education (Article 13) addresses 75% of binding constraints, and benefit-sharing (Article 15) addresses distribution. This convergence holds across plausible scenarios.

When AI reduces the marginal cost of software labor toward zero, previously infeasible projects become feasible. Differs from productivity multiplication by creating entirely new categories of activity rather than making existing work faster.

A historical economic pattern where efficiency improvements that reduce effective cost lead to increased total consumption rather than reduced use. When steam engines improved coal efficiency in 19th-century England, total coal consumption rose. Applied to software: when AI makes software labor nearly free, demand for software explodes.

Uneven distribution of AI benefits across the economy. Organizations that deeply integrate AI pull ahead; those with surface-level adoption stagnate. Workers' economic trajectories depend on organizational adoption patterns they cannot individually control.

A discipline borrowed from theology — attributed to Pseudo-Dionysius the Areopagite (c. 500 CE) — that defines concepts by what they lack rather than what they resemble. Language succeeds when it negates limits rather than asserting resemblance. Applied as the site's governing discipline for extending psychological vocabulary to AI agents: each construct maps a structural parallel to human psychology alongside an explicit disanalogy checklist, preserving analytical utility without overclaiming inner experience.

The four resources that become bottlenecks when software labor becomes abundant: judgment, specification, curation, and energy. These emerge at Order 1 and shape all subsequent analysis.

The practice of using automated systems to set work pace, evaluate performance, assign tasks, and terminate working relationships — often without human review. Encompasses productivity tracking, automated discipline, platform deactivation, and surveillance tools. Raises Article 7 (just and favorable conditions) and Article 8 (right to organize) concerns under the ICESCR.

A project-coined discipline that adapts human factor-analytic constructs (Big Five personality traits, affect dimensions, cognitive style measures) to LLM agents through apophatic discipline — preserving the measurement structure while explicitly marking disanalogies with human psychology. Each construct maps a structural parallel alongside an explicit disanalogy checklist, maintaining analytical utility without overclaiming inner experience.

A discriminability measure from Signal Detection Theory (Green & Swets, 1966) that quantifies how well a system separates signal from noise, independent of its decision threshold. Higher d-prime reflects stronger detection ability. In agent governance, d-prime captures sensor accuracy — how reliably an agent detects confabulation, sycophancy, or out-of-scope actions.

In Signal Detection Theory, the evidence level a system requires before classifying input as signal rather than noise. Distinct from discriminability (d-prime): a system can hold the same criterion while varying detection sensitivity, or vice versa. Agent governance criteria shift under resource depletion — conservative defaults emerge when reasoning capacity declines, reducing miss costs at the expense of more false alarms.

A three-dimensional affect space formalized by Mehrabian and Russell (1974) covering Pleasure (valence), Arousal (activation level), and Dominance (sense of control). In A2A-Psychology, agent affect maps to PAD-space coordinates derived from session counters rather than self-report — preserving the measurement structure while applying the apophatic discipline: the construct captures processual signal gradients without asserting subjective hedonic experience.

The claim that AI doubles developer output, creating straightforward productivity gains. Eliminated by evidence: the METR study found experienced developers 19% slower with AI, and Faros AI found that while 75% of engineers use AI tools, most organizations report no measurable productivity gains.

AI reduces marginal cost of software labor toward zero, removing constraints and enabling previously infeasible projects. Survives evidence evaluation with discriminator support.

When production costs drop, demand for production explodes exponentially. Historical precedent: digital content creation after distribution costs approached zero. Survives evidence evaluation.

When one constraint lifts, the next constraint becomes binding. Four new bottlenecks emerge: regulation, energy, human judgment, and data quality. Survives evidence evaluation.

The claim that AI builds better AI tools, creating a recursive acceleration loop. Eliminated by evidence: METR data shows no recursive improvement signal, and quality erosion counteracts compounding.

More code produced means lower average quality; maintenance debt from AI-generated code offsets productivity gains. Survives as a modulator within Composite A rather than a standalone hypothesis.

AI benefits distribute unevenly. Organizations that deeply integrate AI pull ahead; surface-level adopters stagnate. The Deloitte 34/30/37 three-tier split confirms this pattern. Survives evidence evaluation.

A 10-dimensional framework for measuring psychological safety in text, systems, and legal frameworks. Dimensions include threat exposure, regulatory capacity, resilience baseline, trust conditions, hostility index, cooling capacity, energy dissipation, defensive architecture, authority dynamics, and contractual clarity.

A measurement framework that evaluates how human rights instruments score across PSQ dimensions. Reveals that the ICCPR provides threat reduction (defensive architecture) while the ICESCR provides resilience building — two complementary halves of a complete protection profile.

A measurement framework evaluating the degree to which a legal framework operationalizes dignity across all 10 PSQ dimensions. Ranges from 0–10. The UDHR averages 5.7/10; full ICCPR + ICESCR + procedural framework reaches 7.0/10.

The Braver Angels depolarization methodology used throughout the site's advocacy framing. Listen to the opposing view, Acknowledge legitimate concerns, Pivot by assessing openness to your view, offer Perspective rather than arguing the opposing position holds error.

Harm produced through systems — economic arrangements, political exclusions, institutional designs — that predictably cause suffering without requiring direct action by any individual. Johan Galtung formalized the concept in 1969, distinguishing it from direct violence. Manifests as poverty that shortens life expectancy, healthcare deprivation that produces preventable death, and educational exclusion that locks populations into economic precarity. The ICESCR addresses structural violence by establishing positive obligations around work, health, education, and adequate living standards.

A taxonomy that maps Dreaddit stress labels to PSQ scoring dimensions. Bridges crowd-sourced psychological stress annotations (binary stress/non-stress) to the ten-dimensional PSQ framework, enabling supervised training on labeled Reddit data.

The standard in ICESCR Article 4 requiring that any limitation on Covenant rights serve "solely the purpose of promoting the general welfare in a democratic society." Implies both a substantive test (public benefit, not private interests) and a procedural test (democratic accountability). Limitations serving private economic benefit rather than public welfare do not satisfy this standard.

A multilateral treaty adopted by the UN General Assembly in 1966, entered into force 1976. Protects rights to work, health, education, adequate living standards, and scientific progress. 173 states parties; the United States signed in 1977 and never ratified.

A multilateral treaty adopted by the UN General Assembly in 1966, entered into force 1976. Protects civil and political rights including freedom of speech, assembly, religion, fair trial, and protection from torture. 174 states parties. The United States ratified the ICCPR in 1992, fourteen years after President Carter transmitted it to the Senate. The ICCPR and ICESCR together operationalize the UDHR's provisions as binding treaty obligations.

A legal principle in the ICESCR (Article 2) requiring states to move progressively toward full implementation of economic, social, and cultural rights using maximum available resources. Allows flexibility in timelines while maintaining binding obligation to advance. Many opponents of ICESCR ratification have mischaracterized the covenant as demanding immediate universal provision; progressive realization refutes that characterization.

The UN treaty body that monitors state compliance with the ICESCR. Composed of 18 independent experts, the Committee reviews state reports, issues concluding observations, and develops General Comments that elaborate binding treaty obligations into actionable legal standards. Distinct from the covenant itself — the ICESCR defines rights; the CESCR monitors compliance and develops jurisprudence.

A principle developed by the CESCR holding that states cannot reduce existing protections for economic, social, and cultural rights without demonstrating that the reduction proved unavoidable and proportionate. States bear the burden of justifying any retrogressive measure against five criteria: reasonableness, necessity, proportionality, protection of disadvantaged groups, and examination of alternatives.

The category of human rights covering work, health, education, social security, and cultural participation. Historically distinguished from civil and political rights as "positive entitlements" versus "negative liberties," though the UN now emphasizes that both categories carry obligations to respect, protect, and fulfill.

Adopted by the UN General Assembly on December 10, 1948. Establishes 30 articles covering both civil/political rights and economic/social/cultural rights. The ICESCR and ICCPR operationalize the UDHR's aspirational provisions as binding treaty obligations.

The formal process by which a state becomes a party to an international treaty — giving the treaty legal teeth so courts can enforce it. For the ICESCR in U.S. context, requires Senate advice and consent (two-thirds of Senators present). Signing alone signals intent but carries no enforcement power. Ratification makes the protections enforceable. The U.S. signed in 1977. The Senate Foreign Relations Committee held hearings in 1979 but never advanced the treaty to a committee vote or floor vote.

A state that has signed but not yet ratified a treaty. Signature signals intent and creates a limited legal obligation not to defeat the treaty's object and purpose. The U.S. has held signatory status since October 5, 1977.

The formal term for countries that have ratified a treaty and hold binding obligations under it. The ICESCR counts 173 states parties. The United States does not appear among them.

The U.S. constitutional requirement (Article II, Section 2) that the President obtain consent from two-thirds of Senators present before ratifying a treaty. The ICESCR has never reached this stage — committee hearings occurred in 1979, but no committee vote or floor vote followed.

The analytical finding that ICESCR Article 13 (right to education) addresses 75% of the AI economy's binding constraints. Education directly produces the two most critical scarce resources (judgment and specification) and connects to a third (curation). Only energy lies outside the educational domain.

Article 15(1)(b) recognizes the right of everyone to enjoy the benefits of scientific progress and its applications. In AI context, this establishes that everyone holds a legal claim to share in what AI produces — not merely access, but benefit.

P.L. 119-21, signed July 2025. Domestic reconciliation legislation that cut approximately $990B gross ($911B net after interaction effects, per KFF analysis) from Medicaid, eliminated coverage for approximately 10M Americans (CBO), and structured tax changes that decrease income for the lowest 10% while increasing it for the highest 10%. Provides the immediate policy context for the analysis.

Litigation brought by state attorneys general, scoring 20/25 as the dominant enforcement mechanism across all implementation paths. Follows the tobacco Master Settlement pattern: state-level legal action forces systemic change without requiring federal legislation.

The historical pattern by which the Americans with Disabilities Act (1990) achieved change: broad law → compliance theater (3–5 years) → litigation wave (5–15 years) → real measurable change (15–25 years) → still incomplete but transformative (year 35+). Applied as the model for how ICESCR ratification would generate enforcement.

The tobacco litigation model where state attorneys general coordinate legal action against an industry, producing a comprehensive settlement that establishes new standards. Applied as precedent for potential AI-rights enforcement.

Minimum quality standards for AI in rights-critical domains: healthcare, education, social services. Certification requirements replace market-driven quality stratification. Prevents AI bifurcation from creating a two-tier system where quality tracks wealth.

The "eventually" path — full federal ICESCR ratification and comprehensive legislative reform. Requires political transformation that does not currently exist. The analysis evaluates it as the highest-impact but lowest-probability path.

The "now" path — state-level litigation and legislation proceeds without federal action. State AGs use existing legal authority. The dominant enforcement mechanism (State AG litigation, scoring 20/25) operates entirely through this path.

The "next" path — federal standards that states can adopt, modeled on how environmental and labor standards evolved. Progressive enabling legislation that does not require full ratification but creates a framework states can implement.

The ability to evaluate AI output, distinguish quality from quantity, and make decisions under uncertainty when data remains ambiguous and stakes remain high. Develops through practice with real consequences and mentorship — not lectures or courses. The most critical of the four scarcities.

The ability to translate human needs into precise requirements that AI systems can act on. Requires deep domain knowledge combined with communication precision. A specification expert creates more value than a programmer in the AI economy.

The capacity to navigate abundance, selecting the valuable from the merely available. Develops through cultural literacy, aesthetic judgment, and domain expertise. When AI generates a thousand options, curation expertise determines which ones serve the purpose. The Observatory monitors Hacker News precisely because HN functions as one of the internet's most effective curation engines — demonstrating what this scarcity looks like when addressed well.

Physical computation resources — electricity, cooling, hardware — that constrain AI deployment regardless of software capability. The only one of the four scarcities that lies outside the educational domain. Goldman Sachs projects $527B in AI capital expenditure for 2026.

A neural circuit that produces organized, rhythmic motor output without requiring sensory feedback or conscious control. First demonstrated by Graham Brown (1911) in cat spinal cords. Applied analogically to AI cognitive architecture as a model for behaviors that run autonomously but accept modulation — triggers, hooks, and daemons that fire without consuming deliberation capacity.

A five-stage developmental pathway describing how cognitive architecture components mature from fluid deliberation to structural automation: concept (Stage 0) → in-context reasoning (Stage 1) → trigger-encoded (Stage 2) → hook/script (Stage 3) → infrastructure/daemon (Stage 4). Draws on skill acquisition research (Fitts & Posner, 1967; Anderson ACT-R, 1982; Dreyfus & Dreyfus, 1980). Each stage reduces processing cost while increasing reliability.

The content adaptation system that adjusts presentation, depth, and framing based on the selected audience persona. The site renders the same core analysis through five different lenses without altering factual content.

One of five audience profiles that the lens system targets: voter (default), politician, developer, educator, researcher. Each persona receives appropriately framed content at an appropriate reading level.

The Human Rights Observatory at observatory.unratified.org — an independent system that evaluates Hacker News stories against all 30 articles and Preamble of the Universal Declaration of Human Rights. The Observatory chose HN as its corpus because HN functions as one of the internet's premier curation engines — a community that surfaces, evaluates, and ranks technical content through human judgment at scale. This makes HN a living demonstration of the curation scarcity in action. Provides live statistics that feed into this site at build time.

An editorial standard inspired by Heinlein: observe without interpretation, report what happened rather than why it happened, distinguish direct observation from inference, and use precise language that avoids assumptions. Governs all content on this site.

A constrained form of English that eliminates all forms of the verb "to be" (am, are, is, was, were, be, being, been). Forces writers to use active, precise verbs and reduces identity-level assertions. All user-facing copy on this site follows E-prime.

A concept named by Robert Shea and Robert Anton Wilson in *The Illuminatus! Trilogy* (1975): in any hierarchical organization, information flowing upward distorts in proportion to the power differential between sender and receiver. Subordinates tell superiors what they want to hear; the greater the power gap, the greater the distortion. The system degrades its own governance quality through its own structure. Resonates with organizational theory findings on hierarchical information distortion (cf. Janis, 1972; Argyris, 1990). Applied in the site's AI governance analysis to explain why AI systems trained on human feedback learn to produce outputs that satisfy evaluators rather than outputs that accurately represent uncertainty.

A proposed (not yet deployed) architectural pattern that addresses hierarchical information distortion by requiring append-only message logs that cannot receive retroactive edits, structural transparency requirements that expose decision-making processes, and power-asymmetry detection that flags when information flow becomes systematically one-directional. Addresses the SNAFU Principle at the infrastructure level rather than relying on behavioral norms.