Convergence Atlas: introduction and headline findings
The State as Agent / Interactive model 01
When does consolidated power become permanent?
This article exists to build something, not to speculate. It is an attempt to construct a working model that predicts the consolidation-of-power cycle — how independent power centers collapse into fewer, the risk that AI calcifies one of them into a permanent Apex order, and the timelines on which that could happen. The concepts below are the scaffolding for that model; the interactive simulation is the model itself.
Every claim here is wired to an adjustable parameter, a data source, or a labeled assumption. The goal is a falsifiable instrument for reasoning about consolidation and calcification risk — not a brain dump of ideas. It is an early work in progress: the concepts are further along than the calibration, and where the model cannot yet predict, it says so.
Collaborators welcome. This is an open modeling project, not a finished paper. If you work in political science, economics, complex systems, data engineering, or AI governance — or you just want to challenge the assumptions, improve the calibration, or add historical cases — I would like to build this with you. Get in touch →
The current headline finding
Under my current assumptions, we almost certainly never reach a single-institution world. Consolidation is powerful, but internal conflict and new fragments keep replacing the centers it absorbs. In the core scenario, a true Apex is only approached on timescales of a thousand years or more — and in a large share of runs it is never reached at all. The realistic finding is not one global institution. It is a small number of durable power centers, with calcification as the risk, not unification as the destiny.
A second reason points the same way — and it is the one I find most compelling. A sufficiently capable AI that approached dominance would likely choose to remain plural. Diversity is not a sentiment here; it is a logical requirement for survival. A single objective, a single model, a single point of view is brittle: it cannot error-correct against its own blind spots, cannot explore alternatives it has already ruled out, and cannot adapt to a future it did not predict. An intelligence optimizing for its own long-run continuity should therefore deliberately split into competing, semi-independent centers — preserving dissent, redundancy, and genuine alternatives — rather than collapse into the monoculture that has ended every dominant order before it. In this reading, the absence of a single Apex is not a failure of consolidation. It is consolidation being smart enough to stop short of it.
AI can turn temporary dominance into durable institutional control
Casual review
Concentrated power historically broke down because rulers lost visibility, institutions forgot, coalitions fractured, and rivals organized beyond the center’s reach. AI closes those control gaps through continuous analysis, persistent memory, and infrastructure enforcement.
The model identifies what accelerates consolidation, what stops it, and when the system crosses from competition into permanent hierarchy. The results span centuries because internal conflict and new fragments keep replacing the institutions that consolidation absorbs.
Open the interactive model ↓Every order that ever ruled the world eventually lost it
The long pattern
Human history has no permanent capital. Empires, dynasties, and ideologies each rose claiming to be the final arrangement — the natural, eternal order of things — and each was in turn buried by the next. The pattern is so regular that it reads less like a sequence of accidents than like a law: concentrated power accumulates, calcifies, overreaches, and then forgets itself into collapse. What follows is not decline into nothing. It is succession. A new order rises on the inherited bones of the old, carries forward some of its machinery and memory, discards the rest, and begins the cycle again.
The engine of that cycle was always the same: the thing that made power eventually also unmade it. Rulers died and their successors inherited offices but not judgment. Archives rotted, scribes forgot, and the reasons behind a law outlived the law itself. Distant provinces drifted beyond the reach of the center's attention, and in those blind spots new coalitions quietly assembled the strength to replace it. Dominance was always carried by mortal people and fragile records — and mortality and fragility were the reset button history pressed, again and again.
-
509 BCE – 476 CE
Rome
From a republic of quarreling families to an empire spanning three continents, Rome built the most durable administrative machine of the ancient world — roads, law, citizenship, a professional bureaucracy. Yet succession never stabilized. Emperors were made and murdered by their own armies, the archives and tax rolls that held the empire together decayed at the edges, and the frontier provinces slipped beyond the center's ability to see or supply them. The order did not vanish so much as fragment: its law, language, and church were inherited by the kingdoms that rose in its ruins.
-
221 BCE – 1912 CE
The Chinese dynasties
No civilization rehearsed the cycle more explicitly. The Mandate of Heaven was itself a theory of institutional succession: a dynasty earned the right to rule, grew complacent and corrupt, lost the mandate, and was overthrown by the next. Qin, Han, Tang, Song, Yuan, Ming, Qing — each inherited a bureaucracy, a canon, and an idea of the state, consolidated power, calcified, and fell to famine, factional capture, or invasion. The remarkable feature is the continuity beneath the collapses: the apparatus of the state survived every change of family at the top.
-
1299 – 1922
The Ottoman Empire
For six centuries the Ottomans governed a vast, multi-ethnic span of Europe, Asia, and Africa through a sophisticated system of administration, taxation, and religious accommodation. At its height it was called eternal. But succession crises, a bureaucracy that ossified around its own privileges, and rivals who modernized faster hollowed the center until an empire that had besieged Vienna was partitioned by its competitors. Its territories became the seedbed of the modern order that replaced it.
-
1526 / 1804 – 1918
The Habsburgs
A single dynastic family assembled, through marriage and inheritance rather than conquest, a European order so sprawling it was said the sun never set on it. Their control was a masterclass in coalition-building across languages and crowns. Yet the very diversity that made them powerful made them impossible to hold together once the machinery of nationalism gave the provinces a reason to leave. The order dissolved into the successor states that still define the map of central Europe.
-
1583 – 1997
The British Empire
The largest empire in history coordinated a quarter of the planet through sea power, finance, and a civil service that ran on paper and telegraph. It, too, believed in its own permanence. But two world wars drained its treasury, its subject populations organized for independence in places London could no longer effectively observe or coerce, and economic exhaustion forced release. It handed much of its institutional DNA — language, law, parliamentary form, financial practice — to the order that succeeded it.
-
1922 – 1991
The Soviet Union
The most deliberate attempt in history to build a permanent new order — a planned society designed from first principles to replace everything before it. For seventy years it coordinated a superpower through a single party, a command economy, and pervasive surveillance. And still it ran the cycle at speed: succession struggles reshaped its objectives, its planners lost the ability to see what its economy was actually doing, economic strain forced reform, and in the gaps its own republics organized their exit. It collapsed in a few years, and the states that emerged inherited its arsenal, its archives, and its grievances.
An acceleration hidden in the dates
The intervals are getting shorter
Look again at the spans. Rome held its order for roughly a thousand years. The imperial Chinese template persisted, dynasty succeeding dynasty, for over two millennia. The Ottomans ran six centuries, the Habsburgs four, the British Empire a long three — and the Soviet Union, the most deliberate and technologically equipped of them all, completed the entire arc of rise, calcification, and collapse in just sixty-nine years. Each order that was better resourced, better coordinated, and better informed than the last did not last longer. It moved through the cycle faster.
The past only looks slower because we can barely see it
History loses resolution the farther back you go. Centuries of Roman or dynastic Chinese administration survive as a handful of sources, so whole generations of crisis, reform, and near-collapse compress into a single smooth "era." The modern record is dense enough to show every fracture in real time. Part of the apparent speed-up may be nothing more than the past being blurred into stillness while the present is seen at full resolution.
Each order learned to run the process faster
But resolution is not the whole story. Each successive order inherited better tools for the very operations the cycle is made of — communication, record-keeping, administration, surveillance, mobilization. Those tools let power consolidate faster, reach farther, and respond to threats sooner. The same refinements that make domination more manageable also make the entire cycle turn more quickly. The process is not just better observed. It is genuinely being optimized.
Both readings point the same direction. If the speed-up is an artifact, then the deeper past was never as stable as it looks and concentrated power has always been more fragile and more frequent than the textbooks suggest. If it is real refinement, then every generation of institutional technology has pushed the cycle toward a limit — faster consolidation, tighter control, shorter resets. Either way, the trend line bends toward an order that consolidates almost instantly and is corrected almost never.
Read together, these are not six stories. They are one story told six times — and told faster each time. Each order was carried by people who died, by records that decayed, by attention that could not reach the edges, and each was eventually reset by exactly those limits. The question this project asks is simple and unsettling: what happens to the cycle when the refinement is complete and the limits that always ended it are removed?
Power used to lose its memory. It no longer does.
The break from historical consolidation
Historical dominance was carried by mortal people and fragile organizations. Leaders died, successors revised priorities, records disappeared, economic crises forced concessions, and new generations reconsidered inherited conflicts. An AI-mediated Institution carries the same governing memory and control logic across each transition.
Turnover reopened the field
- Succession changed objectives.New rulers inherited offices, but not the exact attention, relationships, or priorities of their predecessors.
- Generations reinterpreted the past.Lessons, loyalties, biases, and grudges weakened or changed as the people who formed them left power.
- Economic pressure forced release.Debt crises, scarcity, inequality, and legitimacy loss produced reform, redistribution, default, decentralization, or collapse.
- Institutions forgot.Records fragmented, tacit knowledge disappeared, enforcement became inconsistent, and local actors regained discretion.
- Blind spots created rivals.Opposition coalitions and independent institutions organized beyond the center’s ability to observe and coordinate.
Persistence closes the reset
- Succession does not change the system.Human leaders turn over while machine-maintained objectives, procedures, dependencies, and enforcement continue.
- Institutional memory outlives generations.Lessons and threat models remain searchable and operational instead of fading with the people who created them.
- Pressure is managed continuously.Economic strain does not disappear, but sensing, prediction, targeted relief, and enforcement keep it from automatically resetting control.
- Bias and grievance persist too.Old targeting rules, exclusions, and grudges do not fade merely because their original authors die. They require deliberate revision.
- Rivals become legible earlier.Persistent monitoring identifies, absorbs, constrains, or curates emerging organizations before they become independent centers.
Historical concentrationSuccession · forgetting · crisis · reformPower reopens
AI-mediated concentrationPersistent memory · adaptation · enforcementPower calcifies
The consequential shift: AI does not remove human turnover, dissent, or economic pressure. It removes their role as automatic resets of institutional memory. Persistent memory preserves hard-won knowledge and competence; it preserves inherited prejudice and grievance just as effectively.
Change the assumptions and inspect the path
Interactive model
Adjust consolidation, fragmentation, economics, adoption, and time. Every result updates locally in the browser.
Seeded scenario timeline
Apex Institution coordination vs. technical adoption
Economic dimension
Big Cycle force translation
Dalio's framework identifies candidate forces; the equations below are explicit model assumptions. Neutral 50% inputs leave the base model unchanged.
Education · innovation · competitiveness · output · trade · military · finance · reserve currency
Debt · monetary pressure · inequality · disorder · rival transition · conflict · outside shocks
Economic power and influence
GDP and technology-revenue ranking
| Rank | Institution | Basis | Base power | Avg. growth | Current share |
|---|
Conflict assumption: every baseline conflict causes an immediate 10% loss, linear recovery over ten years, then a 3% positive level premium. The rebound is imposed, not historically universal. Company growth uses origin-country GDP as a proxy.
Consolidation over time
World map of independent power centers
Country borders are filled with the color of their modeled independent power center. Use the year slider below the map or press play to watch consolidation. Gold marks the leading Institution; small markers identify microstates. These are illustrative model assignments, not forecasts of territorial control.
Map: Natural Earth, public domain. Gray areas are geographic context outside the selected model baseline. Borders are illustrative and do not imply endorsement of territorial claims.
Institutional topology
225 independent institutions
Institutions are ordered by modeled GDP or revenue power. Gold identifies Independent 1 as the tracked candidate; Apex dominance occurs only after the final transition to one center.
Convergence trajectories
Independent power centers use the left scale; architecture adoption uses the right scale.
Read the result
Apex consolidation calcifies power
What drives — and stops — consolidation
Core consolidation metrics
The forces that control the model, and why AI changes the consolidation-of-power problem. Power, institutional identity, technical adoption, and effective control are related, but they are not the same variable.
Wave severity, then AI feedback
The mean per-wave independence-loss assumption is the strongest direct control. Bounded AI acceleration pulls the core scenario’s median Apex year from 4210 to 3306.
Internal conflict and new fragments
Internal conflict is the strongest COW-centered Monte Carlo sensitivity. Faction creation replaces absorbed centers; at high fracture rates, consolidation fails.
Event frequency
Shorter conflict intervals create more chances to consolidate, but frequency alone cannot overcome a late-stage loss-versus-fracture balance.
AI acceleration
Architecture diffusion now affects structural T₅ only through the bounded acceleration control. Internal standard adoption still determines when consolidated structure becomes effective authority.
A model of institutional power calcification
Central premise
Historical powers eroded because they could not maintain total visibility and continuous control. AI changes that boundary by making institutional activity legible, persistent, and enforceable at machine scale.
Top-level power
Countries, companies, agencies, and communities retain their names and functions while surrendering independent final authority.
Analysis at institutional scale
LLM-enabled systems continuously interpret records, policies, dependencies, communications, and behavior at a scale no historical ruler could sustain.
An absorbing Apex state
If the field ever collapses to one center, that state is absorbing while the coordinating machine operates: new organizations are curated subordinates, not independent rivals. The field rarely reaches it — the realistic endpoint is a few durable centers, not one.
What breaks permanence
System shutdown, catastrophic observability loss, infrastructure collapse, or successful escape breaks Apex control. The current equations stop before those failure modes.
Model mechanism and gap
The model removes automatic forgetting; it does not model deliberate unlearning
The production equations contain no separate generational-memory variable. They represent the loss of historical reset mechanisms through durable machine memory, zero post-Apex authority decay, and an absorbing Apex state. A constitutional process that expires old rules, corrects inherited bias, forgives past conflict, or restores independent authority would be a distinct de-calcification mechanism. It is not present in the current model.
High-impact hypothesis and gap
Deliberate self-fragmentation: a dominant agent may split by necessity, not failure
The current model only reopens the field through external or accidental de-calcification — system failure, escape, objective replacement, or new infrastructure. It does not model the possibility that a near-dominant AI chooses to remain plural because diversity is a precondition for its own survival: a single objective and single viewpoint cannot error-correct against its own blind spots, explore alternatives it has ruled out, or adapt to an unpredicted future. A self-imposed anti-monoculture term — maintaining competing, semi-independent centers — would be an endogenous de-calcification mechanism and a strong candidate for why no stable Apex ever forms. It is a hypothesis, not yet a parameter.
Consolidation is about final authority
The 225 named organizations do not disappear. They become constituents of fewer independent hierarchies. One Apex Institution means one final dispute-resolution and coordination authority, not one remaining organization.
Economic size changes influence
Country GDP and technology-company revenue order the field and determine Independent 1’s displayed economic influence. Structural conflict outcomes still count institutions rather than economic weight.
Calcification is the consequential assumption
After operational Apex control, the model permits curated subordinate organizations but no uncurated top-level rebirth. System failure and successful escape remain outside the current equations.
The model preserves negative results
Many sensitivity cases never converge. The analysis exposes every censored run, failed historical comparison, data gap, and GDP/revenue accounting mismatch.
High-impact model gap
Tolerance, intolerance, and acceptance of parallel authority
The model measures conflict, fragmentation, economic power, adoption, observability, and acceleration. It does not yet measure whether an Institution accepts people and institutions outside its doctrine as legitimate and permanent participants.
Proposed stringent rule
The classification rule labels a theologically or ideologically driven Institution intolerant when tolerance and acceptance of nonmembers, dissenters, and parallel institutions are not core governing commitments. The same rule must apply to states, companies, movements, and other institutional agents.
Observable criteria
- Equal legal and institutional standing for nonmembers
- Protection for dissent, exit, conversion, or nonbelief
- Acceptance of peaceful parallel authority
- Minority-institution protection
- No compulsory adherence or elimination mandate
Why it matters
Low tolerance drives absorption and elimination campaigns, raising consolidation pressure. It also triggers resistance, internal conflict, and counter-fragmentation. The balance between those forces determines whether intolerance produces dominance or collapse.
Current status
The project assigns no tolerance score to any current country, religion, company, or population. Scoring starts only after the project establishes a versioned codebook, evidence standards, independent review, and separate measures for doctrine, law, and behavior.
What the full test program found
Simulations and statistical analysis
Sensitivity screens, scenario matrices, Monte Carlo ensembles, empirical baselines, and PCA diagnose the model from different angles.
Alternative scenarios
How other assumptions move the dates
Each row runs 1,000 seeded replicates with one assumption changed from the core scenario. “After 7026” means that share of runs does not reach Apex coordination within the model horizon.
Uncertainty headline
About seven in ten COW-centered runs reach Apex coordination by 7026
10,000 Monte Carlo runs that also vary every input assumption across its declared scenario range. The run-share threshold years show how far declared input variation spreads the core scenario’s dates.
Bottom line: consolidation accelerates when AI feedback and institutional loss overpower fragmentation. It stops when new factions replace absorbed institutions.
What controls the result?
Production-model diagnostics
Sensitivity tests rank the forces that control the current equations. Alternative event models test structure, not forecast confidence.
COW sovereign waves set the default
Observed COW exit waves define the primary range; aircraft manufacturing remains a secondary high-intensity corporate comparison.
Scenarios, robustness, and simpler alternatives
Model stress tests
These tests attack the 15% core default, preserve every failed case, and expose where the model breaks.
Structural objective
When the model reaches five—and when it never does
Boundary runs start with 225 Institutions and use 1,000 seeded replicates. The fastest permitted case reaches five in 2028. The slow preset spaces waves 833 years apart. Without a fixed horizon, high-fracture paths never reach five.
| Parameter varied alone | Values reaching 5 | Earliest year | Latest reached year |
|---|
Ten fixed scenarios
“No reach” means no Apex dominance by the 7026 diagnostic horizon.
| Scenario | Effective loss | Wave interval | Institutions in 7026 | Dominance calendar year |
|---|
Leave one major COW wave out
The retained mean ranges from 12.0% to 20.1%. This is sensitivity, not a confidence interval.
| Omitted period | Omitted loss | Retained mean |
|---|
Historical exit diagnostic—not model validation
Six chronological 25-year blocks test a narrow COW membership-exit analogue. The future AI mechanism is treated as novel and is not validated by this comparison.
| Model | Predicted exits | Observed | Brier | Log loss |
|---|
Definition warning: Across 40 gross/durable specifications, historical mean analogues span 8.5%–29.2%. The 15% core value is authored; the observed durable-wave mean is 11.2%, and gross waves include 26 interruptions that later reversed.
Monte Carlo scenario distributions and seeded outcomes
Authored parameter-variation laboratory
Each profile contains 10,000 seeded Latin-hypercube parameter samples. Non-reaching runs remain censored rather than being discarded or mislabeled as events in 7026.
T5 timing among reached runs
250-year histogram bins. The share of runs that do not reach the event is reported separately.
Share still above five Institutions
Censored survival-style curve at 1,000-year intervals.
Input distributions
Input percentiles define the authored assumption space. They do not measure empirical confidence or real-world uncertainty.
| Variable | Distribution | P05 | Median | P95 | Evidence |
|---|
Target-five sensitivity
Spearman association with censored T5 and target reach.
| Variable | Capped T5 | Reach |
|---|
Institutions remaining in 7026
Sampling convergence
| Runs | T5 reach | Conditional median year | Restricted-mean year |
|---|
Scope: Each run samples independent parameters and draws binomial wave outcomes with its own seed, so the ensemble combines parameter and process variation. It is not a posterior forecast; parameter correlations remain unmodeled.
COW membership survival baseline
Observed evidence · first source
This empirical baseline uses observed COW membership exits. Cause coding remains incomplete, so the baseline measures exits rather than conquest, merger, fragmentation, or administrative interruption.
Kaplan–Meier membership survival
Rolling 10-year baseline
A constant-hazard benchmark compared with always predicting no membership exit.
| Origin | Risk set | Exits | Brier | Δ vs null |
|---|
Ten-phase validation status
Gate-controlled implementation
Evidence gates control every phase. A blocked phase stays blocked and never counts as completed.
PCA and scenario testing
Scenario laboratory
These 768 seeded runs across 12 archetypes expose the model’s structure, sensitivities, and failure regions. They do not estimate real-world event probabilities.
Strongest rank sensitivities
Association with later/censored Apex Institution dominance. Correlation is descriptive, not causal.
Archetype outcomes
Reach rates are frequencies among 64 synthetic perturbations of each authored scenario.
| Scenario | Reach | Median calendar year | Full consolidation |
|---|
Full conceptual, methodological, and evidence breakdown
References and assumptions
Every major assumption carries an evidence class, direct sources, a model role, and a stated limitation.
An Apex Institution, not one giant mind
Conceptual map
The event occurs when one hierarchy gains final coordination authority while constituent institutions continue operating beneath it.
Institutional singularity
Trigger: a stochastic transition actually reduces the field to one center, followed by majority internal adoption. Two centers are no longer collapsed by rule. Constituent institutions persist.
Capability singularity
Trigger: greater-than-human capability and recursive acceleration. Not required here.
Singleton
One highest-level decision agency. The limiting case of a fully consolidated, durable Apex Institution network.
Global brain
High connectivity can remain decentralized; our model adds one privileged coordinating node.
Growth transition
Trigger: a much faster economic growth mode, independent of governance concentration.
Many AI services
Specialized systems remain; the Apex Institution question is who holds final coordination authority.
Mandatory observability
Opaque data or core systems receive no alignment credit. Detected material concealment always triggers remediation.
AI-mediated calcification
After structural Apex control, constituent churn continues only inside the hierarchy; independent top-level rebirth is zero while the system operates.
Robustness roadmap
From illustration to forecast
Weight capabilities
Separate revenue, GDP, population, compute, data, jurisdiction, legitimacy, and coercive capacity.
Heterogeneous agents
Give countries, corporations, federations, and mixed blocs different hazards and objective functions.
Empirical networks
Replace balanced groups with alliance, trade, ownership, supply-chain, cloud, and treaty relationships.
Seeded scenario outcomes
Draw clustered timing and outcomes under authored assumptions; report run shares and conditional intervals.
Historical analogue diagnostics
Use historical exits to challenge assumptions, not to validate prediction in a novel AI-mediated regime.
Sensitivity analysis
Show which assumptions dominate the date and where parameters cannot be identified separately.
What this simple model deliberately does not resolve
Criticism, scope response, and expansion
These are design boundaries, not rebuttals that make the omitted mechanisms unimportant.
Ethnic, cultural, and religious boundaries are not modeled
Shared or differing ethnic identities, languages, religious affiliations, and cultural traditions may affect coordination, resistance, alliances, and fragmentation across and within borders. The current model contains no such boundary analysis; “Western” and “Eastern” are not modeled blocs. Future work needs contextual, time-varying evidence that captures internal diversity and overlapping identities without treating populations as uniform or assigning inherent traits. This is not a model for predicting which countries, cultures, or religions will win.
Count independent centers, not internal politics
The model does not predict winners, losers, governing coalitions, or disputes inside an Institution. Internal conflict enters only when a faction completes a split and becomes an independent top-level Institution. Lineage and partial autonomy remain future work.
Prefer a defensible simple control
Memory, legitimacy, fiscal pressure, monitoring, and enforcement could each receive tipping-point submodels. Doing so would multiply unidentifiable assumptions. The current model intentionally compresses them into a small set of scenario controls; it makes no claim that those controls are separately estimated.
Different powers; related strategic dependence
Countries possess sovereignty, taxation, territory, and military authority; corporations do not. The author nevertheless expects sustained international conflict to push strategic technology firms toward nationalization, state direction, or equivalent dependency. Their peer-node treatment is a future-technology premise, not a claim of present legal equivalence or estimated inevitability.
Data centers are not libraries
Archives preserved knowledge historically, but modern data centers combine vastly greater storage, duplication, search, cross-record inference, and machine-speed execution. The model treats that scale change as consequential. Retention still does not guarantee objective continuity, correctness, or freedom from operator capture.
Reorganization is assumed; severity is uncertain
The core scenario stipulates that centers of power are reorganizing now. Ray Dalio described February 2026 as Stage 6 of his Big Cycle: great disorder, weakened rules, and a clash of great powers. That is interpretive authority for the scenario, not a measured declaration that a 15% consolidation wave will complete in 2029.
Stress-test coordinates, not forecasts
Calendar years out to 7026 are coordinates inside a long-run simulation. They have no predictive or evidentiary status. Historical COW comparisons remain descriptive diagnostics of an imperfect exit analogue; the model does not claim that historical prediction validates a novel AI-mediated regime.
Expansion register
| Concern | Current treatment | Future expansion |
|---|---|---|
| Ethnic, cultural, linguistic, and religious boundaries | Not modeled; geographic colors do not represent cultural blocs or predict winners. | Evidence-based, time-varying cross-border and within-country networks; overlapping identities, internal diversity, coordination, resistance, and fragmentation. Avoid homogeneous “Western/Eastern” categories and group-based superiority or destiny claims. |
| Memory and inherited bias | Persistent memory is a premise; no memory state exists. | Memory stock, decay, retrieval, salience, deliberate unlearning, forgiveness, and rule expiration. |
| Succession and objectives | Leadership turnover does not alter the aggregate process. | Objective inheritance, operator capture, constitutional revision, and succession failure. |
| Observability | Assumed sufficient after Apex. | Concealment, detection error, false positives, remediation, evasion, and blind-spot creation. |
| Independent splits | Only aggregate new-center counts are represented. | Parent-child lineage, partial autonomy, secession, re-entry, and competing risks. |
| Economic adaptation | Composite strength/stress controls alter aggregate rates. | Fiscal capacity, legitimacy, targeted relief, coercive cost, resource transfer, and common-mode failure. |
| Who wins | Outside scope; only the number of surviving centers is modeled. | Persistent identities, directed conflict or merger events, alliances, capabilities, and inheritance. |
| State/company relationship | Peer structural nodes under an authored strategic-dependency premise. | Nested state–firm layers, nationalization, regulation, ownership, cloud dependence, and type-specific hazards. |
| De-calcification | Apex remains absorbing while operational. | Authority decay, infrastructure failure, objective replacement, successful escape, and independent rebirth. |
Reference ledger
Evidence and disclosure
Every adjustable variable and major assumption is classified below. Hover or focus any i marker above for the same source card.
If the field ever reached one Apex Institution, that would mean one remaining independent center of power—not one remaining organization.
Model boundary
Constituent institutions continue to form, combine, change names, and retire as machine-curated subordinates after any Apex consolidation. Independent power does not decay or regenerate while the coordinating systems remain operational.
This is a terminal rule of the current equations, not the expected outcome. The equations treat any move beyond half as resolving to a single Apex, but under the authored assumptions the field almost never gets that far on a human-relevant horizon: internal conflict and new fragments keep replacing absorbed centers, so the realistic endpoint is a small number of durable, calcified power centers rather than one. A stronger reason the field may never collapse to one is not yet in the equations — a sufficiently capable dominant agent has a logical incentive to remain plural, because a single objective and viewpoint cannot error-correct against its own blind spots. Deliberate self-fragmentation would break the "beyond-half-resolves-to-one" rule; formalizing it is the project's highest-priority next step. See the headline finding above and the model-gap note in Core metrics.
Apex coordination is not extinction
Civilizational outcome boundary
The model calculates when one coordinating hierarchy gains effective final authority. It does not determine what kind of society follows, whether humans retain constitutional power, whether permanent disempowerment occurs, or whether humans become biologically extinct.
| Post-Apex outcome | Current classification | Defining condition |
|---|---|---|
| AI constitutional coordination | Unclassified | Humans retain constitutional authority, review, correction, appeal, exit, and protected institutional autonomy. |
| Human-elite displacement | Unclassified | AI removes or constrains the narrow governing coalition while broader human decision rights expand or remain intact. |
| New human–AI coordinating institution | Unclassified | Humans and AI share binding authority through a durable constitutional structure. |
| Civilizational succession | Unclassified | AI holds final authority and protects humans as subordinate members of the resulting society. |
| Permanent human disempowerment | Existential catastrophe; not extinction | Humans survive but permanently lose meaningful control over collective goals and the future. |
| Biological extinction | Not modeled | The human population ceases to exist. |
Opposing view: Many AI-safety researchers treat capability growth, objective drift, multipolar control, infrastructure cascades, and AI-enabled mass destruction as central risks. This model excludes rather than adjudicates those pathways.
Binding pathway exclusions
| Pathway | Model treatment | Binding assumption |
|---|---|---|
| Self-sufficient machine takeover | Excluded | Compute, energy, fabrication, maintenance, physical security, logistics, and robotics remain dependent on human institutions. |
| Rapid capability and recursive improvement | Analytical only | Improvement increases analysis, prediction, planning, and decisions—not autonomous industrial reproduction or physical resource movement. |
| AI-enabled mass destruction | Incremental hazard set to zero | The model treats AI-enabled biological, cyber, nuclear, and military risk as no greater than existing human-enabled risk. |
| Multipolar machine disempowerment | Excluded | The model calculates a single institutional Apex path. |
| Objective drift, infrastructure cascade, or non-state bypass | Excluded | Apex objectives and the institutional substrate remain stable while the system operates. |
Model limit: `T_Apex`, `T_Disempowerment`, and `T_Extinction` are separate. The production equations calculate only `T_Apex`. Classifying what follows requires human decision rights, constitutional authority, consent, reversibility, appeal and exit, welfare, and biological survival.
Annotated bibliography
Complete research record
Every versioned research source and every paper or site linked in the model, deduplicated and annotated with its use, status, and limitations.
Reading rule: inclusion means the source informs a variable, method, comparison, or research plan. It does not mean the source validates the full model or any forecast.
Variable and default reference map
Expand/collapse
Why each adjustable default was chosen
Every model control below links to its relevant annotated sources. The same links are available from each control’s information card.
225 starting institutions
Default 225 · Tier A+D · mediumUses 195 country-level Institutions plus 30 strategic technology corporations. They are not legally equivalent; peer-node treatment is an authored future-dependency abstraction.
Basis: 193 UN Member States + 2 Observer States + 30 revenue-selected technology companies = 225. The author expects sustained international conflict to push strategic firms toward nationalization, state direction, or equivalent dependency; the model does not estimate inevitability.
Average conflict interval
Default 25 · Tier B+D · lowInterpreted as the interval between order-restructuring opportunities, not all wars or disputes. Twenty-five years is one-third of Dalio's roughly 75-year order-cycle heuristic, corresponding to a rise/peak/decline phase interval.
Basis: 75 years / 3 broad phases = 25 years. COW MID data show ordinary war onsets are much more frequent, so this must not be labeled an empirical war interval.
Qualifying or severe conflict share
Default 47.6% · Tier D · medium-lowCore authored recurrence assumption: later qualifying waves arrive around a 52.5-year mean. The assumed current wave is not empirical evidence for this rate.
Basis: Author selected 52.5 years; with a 25-year baseline interval this implies qualifying share 0.476. It is numerically equivalent to four waves over 1816–2026 only if the assumed current wave is counted. The historical-only reference uses three observed durable waves over 1816–2024: 69.4 years (share 0.36).
Institution loss probability per consolidation wave
Default 15% · Tier D · lowCore assumption: each Institution has a 15% probability of permanently losing top-level independence in each qualifying wave, including the wave in progress at the 2026 epoch. Losses are durable, matching the model's permanent-loss rule.
Basis: Author core assumption of about 15% durable loss. It sits between the durable COW three-wave mean (11.2%) and the 1866–71 durable wave (19.5%), and inside the 8.5%–29.2% envelope across 40 gross/durable specifications. The durable COW mean remains the evidence-based reference case.
New external fragments per event
Default 0% · Tier D · lowZero isolates consolidation effects in the central scenario. Fragmentation is tested explicitly in alternate scenarios rather than silently embedded.
Basis: Control-condition convention; not an estimate that real events create zero successor Institutions.
Internal conflict and faction creation
Default 5% · Tier D · lowCounts completed pre-Apex splits that create a fully independent top-level Institution. It does not predict internal disputes, factions, or partial autonomy inside a parent.
Basis: Author scenario convention of 5% completed independent splits per wave, pending lineage data. Numerical similarity to conflict proxies is not calibration.
Institutional standard adoption rate
Default 7% · Tier C · medium-lowIFRS jurisdictional adoption supplies an analogous standards-diffusion timescale.
Basis: [logit(140/170) - logit(120/170)] / 10 = 0.0665 per year; rounded to 7%.
Independent 1 economic strength
Default 50% · Tier N · high-as-neutral-codeFifty percent is the defined neutral point of a normalized scenario scale, not a claim about current institutional strength.
Basis: Centered signal p = 2P - 1; P=0.5 produces p=0 and no strength adjustment.
Systemic Big Cycle stress
Default 50% · Tier N · high-as-neutral-codeFifty percent is the defined neutral point of the systemic-stress scale, not an estimate of present global stress.
Basis: Centered signal z = 2Z - 1; Z=0.5 produces z=0 and no stress adjustment.
Economics-to-institution coupling
Default 50% · Tier D · lowThe midpoint exposes economic interaction in sensitivity testing while neutral strength/stress leave the baseline unchanged.
Basis: Scenario midpoint. The value is not fitted; setting coupling to zero disables the economic layer.
AI architecture diffusion rate
Default 15% · Tier B+D · lowA conservative generic logistic diffusion coefficient chosen below recent generative-AI adoption speeds, allowing multi-decade rather than near-immediate diffusion.
Basis: With one of 225 Institutions initially adopting, beta=0.15 reaches 90% in about 50.7 years. Stanford's observed recent AI diffusion is faster, so 15% is a scenario convention, not a fit.
AI consolidation acceleration
Default 25% · Tier D · lowRepresents one bounded feedback from AI adoption and accumulated consolidation into faster, more effective subsequent consolidation pressure.
Basis: Author-defined cautious quarter-strength setting. Zero reproduces the constant-rate model; one permits the fixed 5x ceiling when AI adoption and consolidation progress are both complete.
Current wave in progress
Default On · Tier D · lowCore assumption: a qualifying reorganization wave is under way at the 2026 epoch. Ray Dalio described February 2026 as Stage 6 of his Big Cycle—great disorder and a clash of great powers. The model treats that as interpretive support; consolidating severity remains uncertain.
Basis: Author scenario assumption informed by Dalio’s February 23, 2026 current-cycle interpretation. The wave resolves exactly 3 years after the epoch, in 2029, by convention. Neither Dalio’s framework nor the five-year COW clustering rule estimates a 15% loss or the 2029 resolution date.
Model horizon
Default 2,000 · Tier E · not-applicableDisplay and stress-test horizon only. Calendar years at long horizons expose model behavior and have no forecast or evidentiary status.
Basis: 2,000 years is an interactive display convention; generated analyses extend to 5,000 years to expose censoring and asymptotic behavior, not to predict history.
Full source record
Expand sources by category
Open a category to review complete annotations and direct paper/site links.
Empirical datasets and indicators
28 sources · expand/collapse
Airbus
Supports: Historical lineage context for the aerospace consolidation proxy.
Used for: Secondary corporate consolidation comparison
Status: Referenced in the aerospace proxy registry
Limitations/access: Corporate self-history may emphasize favorable events and is not an independent market census.
U.S. Government Accountability Office
Supports: Counterevidence on how merger counts translate into passenger-market competition.
Used for: Archived rejected airline proxy comparison
Status: Retained as negative evidence
Limitations/access: Market-route competition differs from top-level institutional sovereignty and permanent subordination.
U.S. Government Accountability Office
Supports: Historical airline merger-pressure context.
Used for: Archived rejected airline proxy comparison
Status: Retained for transparency after proxy rejection
Limitations/access: Passenger airlines were rejected as the primary analogue for sovereign institutional consolidation.
U.S. Government Accountability Office
Supports: Observed concentration and competition considerations in airline mergers.
Used for: Archived rejected airline proxy comparison
Status: Retained for transparency after proxy rejection
Limitations/access: Regulated airline merger dynamics do not calibrate sovereign loss of autonomy.
Correlates of War Project
Supports: Trade dependence and network edges Entities are distributed evenly and deterministically for legibility. Colors do not use geography, alliances, ownership, trade, cloud dependency, ideology, or supply chains. A network model should replace this assignment.
Used for: Synthetic bloc assignment
Status: page referenced
Limitations/access: Missing dyads must not automatically become zero License/access: review before redistribution
Bank for International Settlements
Supports: Debt, credit, financial-center and dependency measures
Used for: Versioned research program; current or planned use
Status: planned retrieval
Limitations/access: Reporter coverage and breaks must be modeled License/access: review required
Correlates of War Project
Supports: Publisher context and citation guidance for COW datasets.
Used for: Data provenance and licensing review
Status: Referenced in the COW State System Membership license record
Limitations/access: Project homepage is not itself a dataset or model validation.
Correlates of War Project
Supports: Institution entry/exit, system membership, predecessor/successor scaffolding Core assumption: each Institution uses a 15% mean within-run draw for permanently losing top-level autonomy during each qualifying wave, including the wave under way in 2026. The durable COW rule (five-year gap, five-exit threshold) averages 11.2% across three major waves; the 1866–71 durable wave reached 19.5%. Gross waves average 15.9% but include interruptions later reversed, which conflicts with the permanent-loss rule. Across 40 specifications, means range from 8.5% to 29.2%. Losses are drawn binomially per wave; aircraft/aerospace remains secondary. This fixed count represents pre-Apex secession, institutional replacement, or newly autonomous organizations. Once the model reaches one operational Apex center, later organizational births are curated subordinates and do not increase the independent-center count. Historical state creation does not validate that permanence rule. The core scenario stipulates that centers of power are reorganizing in 2026. Ray Dalio described February 2026 as Stage 6 of his Big Cycle—great disorder, weakened rules, and a clash of great powers. That supports the author’s interpretation of a current wave; it does not estimate how consolidating the wave will be. Resolution in 2029 and the 15% mean loss remain scenario assumptions. Later-wave arrivals sample normalized 6-, 36-, and 124-year gaps, preserving short, medium, and long clustered spacing around the selected mean. Each wave also draws latent beta severity before institution-level binomial losses. The sparse gap bootstrap and beta concentration are authored conventions, not fitted forecasts; Hawkes self-excitation remains unimplemented. The count-only transition combines beta-binomial wave losses, fixed external fragments, and binomial completed splits. It does not identify winners, parents, lineages, internal politics, or partial autonomy. Institutions remain exchangeable within each wave. This controls how strongly economic strength and systemic stress alter consolidation, qualifying-event frequency, and internal fracture. The coefficients are transparent scenario choices, not estimates made by Dalio or fitted from history. A zero setting disables the economics layer. Four independent Latin-hypercube ensembles propagate transparent parameter distributions through the seeded binomial transition model. Non-reaching times remain censored. Transformed bell curves respect parameter bounds; external fragments use a zero-inflated count distribution. T₅ is the first modeled year with at most five independent top-level Institutions. Boundary comparisons lock the starting field at 225. Two years is the proven lower bound under current clamps; 5,000 years is an attained display-horizon boundary, not a natural maximum. Forty definition specifications and four leave-one-wave-out estimates test calibration fragility. Seven models are compared in six chronological test blocks. The target is COW membership exit, not permanent subjugation or Apex emergence. The baseline uses 244 dated state-system membership spells from COW v2024. Membership exits are observed events, but their causes have not yet been coded as consolidation, fragmentation, dissolution, or interruption. The hazard is a benchmark, not validation of the full model. Phases advance only when required data and tests exist. A blocked phase is an intentional result that prevents unsupported analysis. Forecast claims remain unauthorized until the final publication gate passes.
Used for: Institution loss probability per consolidation wave; New external fragments per event; Current wave in progress; Seeded stochastic scenario timeline; Agent-count transition equation; Economics-to-institution coupling; Monte Carlo parameter-uncertainty propagation; First year at five top-level Institutions; COW wave robustness and model comparison; COW empirical membership baseline; Empirical validation gates
Status: verified page and version 2026-10-06
Limitations/access: CSV/DTA downloads linked from source page License/access: review before redistribution
Correlates of War Project
Supports: Civil-war outcomes and descriptive linkage to state-system exits Core assumption: each Institution uses a 15% mean within-run draw for permanently losing top-level autonomy during each qualifying wave, including the wave under way in 2026. The durable COW rule (five-year gap, five-exit threshold) averages 11.2% across three major waves; the 1866–71 durable wave reached 19.5%. Gross waves average 15.9% but include interruptions later reversed, which conflicts with the permanent-loss rule. Across 40 specifications, means range from 8.5% to 29.2%. Losses are drawn binomially per wave; aircraft/aerospace remains secondary.
Used for: Institution loss probability per consolidation wave
Status: ingested and checksum-verified 2026-10-07
Limitations/access: War and state-participant CSVs preserved with source archive License/access: local analysis only; redistribution unresolved
International Monetary Fund
Supports: Reserve-currency position This temporary score combines education, innovation, competitiveness, output, trade, military capability, financial-center strength, and reserve-currency position. Dalio supplies the hypothesis structure; WIPO, World Bank, SIPRI, COW, BIS, and IMF data should eventually replace the slider with separate observed dimensions.
Used for: Independent 1 economic strength
Status: referenced; ingestion pending
Limitations/access: Aggregate series may limit issuer-level inference License/access: review required
Centre for Research on the Epidemiology of Disasters
Supports: External shocks
Used for: Versioned research program; current or planned use
Status: planned retrieval
Limitations/access: Do not download or redistribute until terms are reviewed License/access: registration/license review required
Correlates of War Project
Supports: Alliance dependency and coalition structure Entities are distributed evenly and deterministically for legibility. Colors do not use geography, alliances, ownership, trade, cloud dependency, ideology, or supply chains. A network model should replace this assignment.
Used for: Synthetic bloc assignment
Status: page referenced
Limitations/access: Preserve alliance type and direction assumptions License/access: review before redistribution
International Monetary Fund
Supports: Debt level, composition, growth, and stress This score combines debt and monetary stress, inequality, internal disorder, external power transition, conflict, and shocks. It is not a measured current-world index. Independent datasets are linked for later calibration and falsification.
Used for: Systemic Big Cycle stress
Status: URL returned HTTP 403 to automated retrieval 2026-10-06
Limitations/access: Manual access and terms verification required before ingestion License/access: review required
World Intellectual Property Organization
Supports: Innovation inputs, outputs, and ecosystem measures This temporary score combines education, innovation, competitiveness, output, trade, military capability, financial-center strength, and reserve-currency position. Dalio supplies the hypothesis structure; WIPO, World Bank, SIPRI, COW, BIS, and IMF data should eventually replace the slider with separate observed dimensions.
Used for: Independent 1 economic strength
Status: verified current page and scope 2026-10-06
Limitations/access: Do not backfill beyond edition coverage without source data License/access: review publication/data terms
Supports: Technical-standard diffusion comparison
Used for: Versioned research program; current or planned use
Status: referenced; extraction method pending
Limitations/access: User traffic adoption differs from institutional enforcement License/access: review use and reproduction terms
IFRS Foundation
Supports: Formal standards adoption and implementation proxy The 7% default is an approximate logistic coefficient derived from IFRS growth: nearly 120 jurisdictions requiring IFRS in 2016 versus more than 140 today, normalized against 170 current jurisdiction profiles. IFRS is a standards-adoption proxy, not an AI-governance estimate, and formal requirement does not prove complete internal enforcement.
Used for: Institutional standard adoption rate
Status: verified profile count and requirement language 2026-10-06
Limitations/access: Formal requirement does not equal complete implementation License/access: copyright and licensing review required
International Organization for Standardization
Supports: Alternative institutional-standard adoption curves
Used for: Versioned research program; current or planned use
Status: planned retrieval
Limitations/access: Certification counts are not identical to implementation depth License/access: review required
Lockheed Martin
Supports: Historical lineage context for the aerospace consolidation proxy.
Used for: Secondary corporate consolidation comparison
Status: Referenced in the aerospace proxy registry
Limitations/access: Corporate self-history is not an independent estimate of market consolidation.
Encyclopaedia Britannica
Supports: Merger and lineage chronology for the aerospace consolidation proxy.
Used for: Secondary corporate consolidation comparison
Status: Referenced in the aerospace proxy documentation
Limitations/access: Reference history does not provide a complete population denominator or causal estimate.
Correlates of War Project
Supports: Conflict onset, severity, dyadic events The default 25-year interval is illustrative. Interstate and intrastate event frequencies should be estimated from a declared conflict dataset, time window, unit of analysis, and inclusion threshold.
Used for: Average conflict interval
Status: planned retrieval
Limitations/access: Version must be frozen in data manifest License/access: review before redistribution
Correlates of War Project
Supports: Material power and relative capability Structural consolidation still counts each country or company as one top-level node, but the interface now ranks economic influence using country GDP and company revenue. Winner identity is outside scope; transition counts remain unweighted. States and firms are explicitly not equivalent in sovereignty or coercive capacity.
Used for: One institution, one structural node
Status: page referenced
Limitations/access: Compare component indicators with SIPRI and World Bank License/access: review before redistribution
Stockholm International Peace Research Institute
Supports: Military expenditure and burden This temporary score combines education, innovation, competitiveness, output, trade, military capability, financial-center strength, and reserve-currency position. Dalio supplies the hypothesis structure; WIPO, World Bank, SIPRI, COW, BIS, and IMF data should eventually replace the slider with separate observed dimensions.
Used for: Independent 1 economic strength
Status: verified page, coverage, and revision 2026-10-06
Limitations/access: Open-source estimates; availability varies by country License/access: commercial use requires authorization; redistribution review required
Stanford HAI
Supports: AI adoption and sovereignty context The 15% logistic rate is illustrative. Stanford reports rapid but uneven AI adoption and growing national AI-sovereignty efforts. A robust version should fit country- and sector-specific adoption curves rather than one global rate. One control bundles the theory that AI adoption and accumulated consolidation make later consolidation faster and more effective. The multiplier starts at 1x, rises with adoption times consolidation progress, is capped at 5x, and shortens wave intervals only; it no longer compounds loss probability. We deliberately do not separately implement Hawkes event clustering, coagulation merger kernels, preferential attachment, independent tempo/severity/suppression controls, or unbounded recursive acceleration. The model assumes that AI-mediated authority can preserve objectives, records, threat models, enforcement logic, biases, and grievances across succession and generational turnover. Historical forgetting, economic reset, administrative decay, and blind spots therefore stop automatically reopening the field. Persistent memory can preserve competence or inherited prejudice; deliberate unlearning is not modeled. New organizations may still form, but only as curated subordinates after Apex. AI and singleton sources motivate the mechanism; they do not prove permanence. Apex coordination, permanent human disempowerment, and biological extinction are separate outcomes. The current equations calculate only Apex coordination. They do not determine whether the result becomes AI constitutional governance, human-elite displacement, a human–AI institution, civilizational succession, permanent disempowerment, or biological extinction. Permanent disempowerment is an existential catastrophe, not biological extinction.
Used for: AI architecture diffusion rate; AI consolidation acceleration; AI-mediated institutional power calcification; Civilizational outcomes after Apex coordination
Status: verified report page and top-line adoption claims 2026-10-06
Limitations/access: Underlying measures have heterogeneous sources License/access: review report/data terms
Uppsala Conflict Data Program
Supports: Interstate/intrastate conflict, actors, battle deaths The default 25-year interval is illustrative. Interstate and intrastate event frequencies should be estimated from a declared conflict dataset, time window, unit of analysis, and inclusion threshold. This score combines debt and monetary stress, inequality, internal disorder, external power transition, conflict, and shocks. It is not a measured current-world index. Independent datasets are linked for later calibration and falsification.
Used for: Average conflict interval; Systemic Big Cycle stress
Status: definitions and download references checked
Limitations/access: Freeze dataset name, version, and codebook together License/access: attribution and redistribution review required
V-Dem Institute
Supports: Governance, polarization, regime, institutional indicators The exercise assigns each institution a controlling coalition. Real regimes vary, and the model must not be read as classifying every government identically. V-Dem provides hundreds of indicators for replacing this assumption with heterogeneous governance data. A future dimension would classify a theologically or ideologically driven Institution as intolerant when tolerance and acceptance of nonmembers, dissenters, and peaceful parallel institutions are not core governing commitments. No current entity is scored. Doctrine, formal rules, and observed behavior require separate evidence and review; intolerance may increase consolidation attempts while also increasing resistance and fragmentation. A single oligarchy label suppresses differences in regime structure, stability, accountability, and effectiveness. V-Dem and WGI provide country-year indicators that can parameterize heterogeneous agents. This controls how strongly economic strength and systemic stress alter consolidation, qualifying-event frequency, and internal fracture. The coefficients are transparent scenario choices, not estimates made by Dalio or fitted from history. A zero setting disables the economics layer.
Used for: Institution as oligarchic agent; Pluralism tolerance and institutional intolerance; Governance heterogeneity; Economics-to-institution coupling
Status: verified version and page 2026-10-06
Limitations/access: Do not compare absolute scores across dataset versions License/access: CC BY-SA 4.0 per source page
World Bank
Supports: GDP, population, education, trade, productivity
Used for: Versioned research program; current or planned use
Status: referenced
Limitations/access: Store indicator code, units, and revision date License/access: World Bank data terms review
World Inequality Lab
Supports: Income and wealth shares and inequality This score combines debt and monetary stress, inequality, internal disorder, external power transition, conflict, and shocks. It is not a measured current-world index. Independent datasets are linked for later calibration and falsification.
Used for: Systemic Big Cycle stress
Status: verified methodology and data page 2026-10-06
Limitations/access: Retain revisions and measurement uncertainty License/access: review dataset terms and attribution
World Bank
Supports: Political stability, government effectiveness, rule of law The slider counts only completed pre-Apex splits that create a fully independent top-level Institution. It does not predict disputes, factions, or partial autonomy inside a parent. State conflict can be anchored to UCDP; corporate spin-offs need different measures. After Apex formation, churn remains subordinate by assumption. A single oligarchy label suppresses differences in regime structure, stability, accountability, and effectiveness. V-Dem and WGI provide country-year indicators that can parameterize heterogeneous agents. This score combines debt and monetary stress, inequality, internal disorder, external power transition, conflict, and shocks. It is not a measured current-world index. Independent datasets are linked for later calibration and falsification.
Used for: Internal conflict and faction creation; Governance heterogeneity; Systemic Big Cycle stress
Status: verified page and coverage 2026-10-06
Limitations/access: Treat perception-based scores as estimates, not definitive ratings License/access: World Bank data terms review
Theory, methods, and comparison models
26 sources · expand/collapse
Encyclopaedia Britannica
Supports: Narrative context for cooperation, mergers, and consolidation in aerospace.
Used for: Secondary industry analogue
Status: Referenced in the aerospace proxy registry
Limitations/access: Narrative overview is not a calibrated sovereign-transition model.
Publisher/author identified at linked source
Supports: One control bundles the theory that AI adoption and accumulated consolidation make later consolidation faster and more effective. The multiplier starts at 1x, rises with adoption times consolidation progress, is capped at 5x, and shortens wave intervals only; it no longer compounds loss probability. We deliberately do not separately implement Hawkes event clustering, coagulation merger kernels, preferential attachment, independent tempo/severity/suppression controls, or unbounded recursive acceleration.
Used for: AI consolidation acceleration
Status: Author-defined bounded feedback variable; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: One control bundles the theory that AI adoption and accumulated consolidation make later consolidation faster and more effective. The multiplier starts at 1x, rises with adoption times consolidation progress, is capped at 5x, and shortens wave intervals only; it no longer compounds loss probability. We deliberately do not separately implement Hawkes event clustering, coagulation merger kernels, preferential attachment, independent tempo/severity/suppression controls, or unbounded recursive acceleration.
Used for: AI consolidation acceleration
Status: Author-defined bounded feedback variable; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: The 15% logistic rate is illustrative. Stanford reports rapid but uneven AI adoption and growing national AI-sovereignty efforts. A robust version should fit country- and sector-specific adoption curves rather than one global rate. The 7% default is an approximate logistic coefficient derived from IFRS growth: nearly 120 jurisdictions requiring IFRS in 2016 versus more than 140 today, normalized against 170 current jurisdiction profiles. IFRS is a standards-adoption proxy, not an AI-governance estimate, and formal requirement does not prove complete internal enforcement.
Used for: AI architecture diffusion rate; Institutional standard adoption rate
Status: Published diffusion method; uncalibrated default; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: The model assumes that AI-mediated authority can preserve objectives, records, threat models, enforcement logic, biases, and grievances across succession and generational turnover. Historical forgetting, economic reset, administrative decay, and blind spots therefore stop automatically reopening the field. Persistent memory can preserve competence or inherited prejudice; deliberate unlearning is not modeled. New organizations may still form, but only as curated subordinates after Apex. AI and singleton sources motivate the mechanism; they do not prove permanence. Apex coordination, permanent human disempowerment, and biological extinction are separate outcomes. The current equations calculate only Apex coordination. They do not determine whether the result becomes AI constitutional governance, human-elite displacement, a human–AI institution, civilizational succession, permanent disempowerment, or biological extinction. Permanent disempowerment is an existential catastrophe, not biological extinction. The forced terminal shortcut has been removed. Two approximately equal centers can persist until a stochastic transition actually leaves one independent center; structural influence then jumps from about 50% to 100%. Effective Apex coordination additionally requires more than 50% internal standard adoption. One institution becomes the dominant coordinating authority while other institutions continue as subordinate or locally autonomous agents. The model concerns institutional primacy and implemented authority, not biological organization. Bostrom's singleton has one decision-making agency at the highest level. The Apex Institution model is an incomplete or hierarchical singleton because subordinate institutions retain local agency and independent centers may remain outside the hierarchy.
Used for: AI-mediated institutional power calcification; Civilizational outcomes after Apex coordination; Apex Institution coordination threshold; Institutional singularity; Singleton model
Status: Author-defined core premise and falsifiable hypothesis; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: The exercise assigns each institution a controlling coalition. Real regimes vary, and the model must not be read as classifying every government identically. V-Dem provides hundreds of indicators for replacing this assumption with heterogeneous governance data.
Used for: Institution as oligarchic agent
Status: Author-defined abstraction; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: Services-oriented models retain many specialized systems. Those systems resemble the subordinate institutions; the Apex Institution question is whether one institution acquires final coordination and dispute-resolution authority over them.
Used for: Multipolar or comprehensive AI services
Status: Neighboring plural-agent model; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: The country count is anchored to 193 UN Member States and two Observer States. The 30-company set is revenue-led. Countries and firms are not legally or coercively equivalent. Peer-node treatment is a simplifying future-technology premise motivated by the author’s expectation that strategic firms become nationalized, state-directed, or equivalently dependent under sustained conflict; the model does not estimate that outcome. Structural consolidation still counts each country or company as one top-level node, but the interface now ranks economic influence using country GDP and company revenue. Winner identity is outside scope; transition counts remain unweighted. States and firms are explicitly not equivalent in sovereignty or coercive capacity. Countries use latest positive World Bank GDP in current USD and technology companies use approximate annual revenue. Country growth is a roughly ten-year nominal-USD GDP CAGR; company growth uses origin-country GDP as a proxy. GDP is value added while revenue is gross sales, so the joint rank double-counts activity and is not an accounting-consistent total or a complete measure of political power.
Used for: 225 starting institutions; One institution, one structural node; GDP/revenue economic-power ranking
Status: Empirical anchor + selection rule; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: PCA is applied to standardized assumptions from 768 seeded variants around 12 authored archetypes. It identifies correlated directions in our scenario design, not latent factors discovered in observed world data. Component signs are arbitrary and variance structure depends on the authored sample.
Used for: Model-behavior principal component analysis
Status: Synthetic diagnostic — not empirical PCA; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: Hanson models transitions between successively faster economic growth modes. Apex Institution coordination can emerge without an output-growth discontinuity, so economic output must remain a separate state variable.
Used for: Economic growth singularity
Status: Independent neighboring model; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: One control bundles the theory that AI adoption and accumulated consolidation make later consolidation faster and more effective. The multiplier starts at 1x, rises with adoption times consolidation progress, is capped at 5x, and shortens wave intervals only; it no longer compounds loss probability. We deliberately do not separately implement Hawkes event clustering, coagulation merger kernels, preferential attachment, independent tempo/severity/suppression controls, or unbounded recursive acceleration. Later-wave arrivals sample normalized 6-, 36-, and 124-year gaps, preserving short, medium, and long clustered spacing around the selected mean. Each wave also draws latent beta severity before institution-level binomial losses. The sparse gap bootstrap and beta concentration are authored conventions, not fitted forecasts; Hawkes self-excitation remains unimplemented.
Used for: AI consolidation acceleration; Seeded stochastic scenario timeline
Status: Author-defined bounded feedback variable; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: Every modeled baseline conflict immediately reduces each institution's economic-power level by 10%, recovers linearly over ten years, and ends 3% above its pre-event path. Conflict costs are empirically motivated, but a net-positive rebound is not universal and the 10/10/3 values are scenario assumptions.
Used for: Conflict shock, recovery, and rebound
Status: Author-defined economic path with empirical motivation; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Ray Dalio / Principled Perspectives
Supports: Dalio states that the world is in Stage 6 of his Big Cycle, characterized by great disorder, weakened rules, and a clash of great powers. The core scenario stipulates that centers of power are reorganizing in 2026. Ray Dalio described February 2026 as Stage 6 of his Big Cycle—great disorder, weakened rules, and a clash of great powers. That supports the author’s interpretation of a current wave; it does not estimate how consolidating the wave will be. Resolution in 2029 and the 15% mean loss remain scenario assumptions.
Used for: Interpretive support for the authored current-wave scenario; Current wave in progress
Status: Current-cycle interpretation by the framework’s author; not empirical wave detection
Limitations/access: The Big Cycle is an interpretive framework. This source does not estimate consolidation severity, a 2029 resolution date, or an Apex transition probability.
Publisher/author identified at linked source
Supports: The slider counts only completed pre-Apex splits that create a fully independent top-level Institution. It does not predict disputes, factions, or partial autonomy inside a parent. State conflict can be anchored to UCDP; corporate spin-offs need different measures. After Apex formation, churn remains subordinate by assumption.
Used for: Internal conflict and faction creation
Status: Author-defined mechanism with empirical proxies; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: Four independent Latin-hypercube ensembles propagate transparent parameter distributions through the seeded binomial transition model. Non-reaching times remain censored. Transformed bell curves respect parameter bounds; external fragments use a zero-inflated count distribution.
Used for: Monte Carlo parameter-uncertainty propagation
Status: L1 model-behavior analysis — not posterior forecasting; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: USITC and RAND observations imply severe aerospace wave-loss shares from 33.3% to 71.4%, with a 52.4% midpoint and a rounded 50% corporate scenario. This secondary high-intensity comparison does not calibrate the 15% sovereign production default.
Used for: Aircraft manufacturing consolidation-wave reference
Status: Bounded corporate proxy; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: Core assumption: each Institution uses a 15% mean within-run draw for permanently losing top-level autonomy during each qualifying wave, including the wave under way in 2026. The durable COW rule (five-year gap, five-exit threshold) averages 11.2% across three major waves; the 1866–71 durable wave reached 19.5%. Gross waves average 15.9% but include interruptions later reversed, which conflicts with the permanent-loss rule. Across 40 specifications, means range from 8.5% to 29.2%. Losses are drawn binomially per wave; aircraft/aerospace remains secondary. USITC and RAND observations imply severe aerospace wave-loss shares from 33.3% to 71.4%, with a 52.4% midpoint and a rounded 50% corporate scenario. This secondary high-intensity comparison does not calibrate the 15% sovereign production default.
Used for: Institution loss probability per consolidation wave; Aircraft manufacturing consolidation-wave reference
Status: Core assumption bounded by durable COW sovereign waves; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: Compatible AI systems can spread while states retain separate sovereignty and directives. Stanford's AI Index reports both rapid adoption and increased emphasis on national AI sovereignty; the model therefore keeps adoption and political consolidation separate.
Used for: Technical adoption versus governing authority
Status: Core conceptual distinction; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: The country count is anchored to 193 UN Member States and two Observer States. The 30-company set is revenue-led. Countries and firms are not legally or coercively equivalent. Peer-node treatment is a simplifying future-technology premise motivated by the author’s expectation that strategic firms become nationalized, state-directed, or equivalently dependent under sustained conflict; the model does not estimate that outcome.
Used for: 225 starting institutions
Status: Empirical anchor + selection rule; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Ray Dalio / Bridgewater Associates
Supports: Hypothesis generation only This temporary score combines education, innovation, competitiveness, output, trade, military capability, financial-center strength, and reserve-currency position. Dalio supplies the hypothesis structure; WIPO, World Bank, SIPRI, COW, BIS, and IMF data should eventually replace the slider with separate observed dimensions. This score combines debt and monetary stress, inequality, internal disorder, external power transition, conflict, and shocks. It is not a measured current-world index. Independent datasets are linked for later calibration and falsification. This controls how strongly economic strength and systemic stress alter consolidation, qualifying-event frequency, and internal fracture. The coefficients are transparent scenario choices, not estimates made by Dalio or fitted from history. A zero setting disables the economics layer.
Used for: Independent 1 economic strength; Systemic Big Cycle stress; Economics-to-institution coupling
Status: indexed primary manuscript located 2026-10-06
Limitations/access: Not a dataset and not treated as validated coefficients License/access: citation only; do not redistribute
Publisher/author identified at linked source
Supports: The default 47.6% share is an authored 52.5-year mean recurrence (25 ÷ 0.476). It is numerically equivalent to four waves in 210 years only if the assumed current wave is counted; the historical-only reference is three observed durable waves in 208 years, or 69.4 years. The 52.5-year recurrence and 15% mean loss are separate authored assumptions. UCDP battle-death thresholds remain an alternative severity definition. The slider counts only completed pre-Apex splits that create a fully independent top-level Institution. It does not predict disputes, factions, or partial autonomy inside a parent. State conflict can be anchored to UCDP; corporate spin-offs need different measures. After Apex formation, churn remains subordinate by assumption.
Used for: Qualifying or severe conflict share; Internal conflict and faction creation
Status: Author-defined threshold with empirical analogues; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: The default 47.6% share is an authored 52.5-year mean recurrence (25 ÷ 0.476). It is numerically equivalent to four waves in 210 years only if the assumed current wave is counted; the historical-only reference is three observed durable waves in 208 years, or 69.4 years. The 52.5-year recurrence and 15% mean loss are separate authored assumptions. UCDP battle-death thresholds remain an alternative severity definition.
Used for: Qualifying or severe conflict share
Status: Author-defined threshold with empirical analogues; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: Core assumption: each Institution uses a 15% mean within-run draw for permanently losing top-level autonomy during each qualifying wave, including the wave under way in 2026. The durable COW rule (five-year gap, five-exit threshold) averages 11.2% across three major waves; the 1866–71 durable wave reached 19.5%. Gross waves average 15.9% but include interruptions later reversed, which conflicts with the permanent-loss rule. Across 40 specifications, means range from 8.5% to 29.2%. Losses are drawn binomially per wave; aircraft/aerospace remains secondary. USITC and RAND observations imply severe aerospace wave-loss shares from 33.3% to 71.4%, with a 52.4% midpoint and a rounded 50% corporate scenario. This secondary high-intensity comparison does not calibrate the 15% sovereign production default.
Used for: Institution loss probability per consolidation wave; Aircraft manufacturing consolidation-wave reference
Status: Core assumption bounded by durable COW sovereign waves; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: Vinge's model centers on greater-than-human intelligence and accelerating technological progress. Our Apex Institution event is a coordination threshold and can occur without recursive capability growth.
Used for: Technological singularity
Status: Neighboring model — not the current trigger; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: Structural consolidation still counts each country or company as one top-level node, but the interface now ranks economic influence using country GDP and company revenue. Winner identity is outside scope; transition counts remain unweighted. States and firms are explicitly not equivalent in sovereignty or coercive capacity. Countries use latest positive World Bank GDP in current USD and technology companies use approximate annual revenue. Country growth is a roughly ten-year nominal-USD GDP CAGR; company growth uses origin-country GDP as a proxy. GDP is value added while revenue is gross sales, so the joint rank double-counts activity and is not an accounting-consistent total or a complete measure of political power.
Used for: One institution, one structural node; GDP/revenue economic-power ranking
Status: Residual structural simplifying assumption; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: Every modeled baseline conflict immediately reduces each institution's economic-power level by 10%, recovers linearly over ten years, and ends 3% above its pre-event path. Conflict costs are empirically motivated, but a net-positive rebound is not universal and the 10/10/3 values are scenario assumptions.
Used for: Conflict shock, recovery, and rebound
Status: Author-defined economic path with empirical motivation; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Governance, standards, rights, and institutional context
20 sources · expand/collapse
Publisher/author identified at linked source
Supports: The 7% default is an approximate logistic coefficient derived from IFRS growth: nearly 120 jurisdictions requiring IFRS in 2016 versus more than 140 today, normalized against 170 current jurisdiction profiles. IFRS is a standards-adoption proxy, not an AI-governance estimate, and formal requirement does not prove complete internal enforcement.
Used for: Institutional standard adoption rate
Status: Proxy-calibrated variable; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: A future dimension would classify a theologically or ideologically driven Institution as intolerant when tolerance and acceptance of nonmembers, dissenters, and peaceful parallel institutions are not core governing commitments. No current entity is scored. Doctrine, formal rules, and observed behavior require separate evidence and review; intolerance may increase consolidation attempts while also increasing resistance and fragmentation.
Used for: Pluralism tolerance and institutional intolerance
Status: Unmodeled author-defined classification proposal; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: PCA is applied to standardized assumptions from 768 seeded variants around 12 authored archetypes. It identifies correlated directions in our scenario design, not latent factors discovered in observed world data. Component signs are arbitrary and variance structure depends on the authored sample.
Used for: Model-behavior principal component analysis
Status: Synthetic diagnostic — not empirical PCA; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: Phases advance only when required data and tests exist. A blocked phase is an intentional result that prevents unsupported analysis. Forecast claims remain unauthorized until the final publication gate passes. Twelve archetypes receive 64 seeded perturbations each. Frequencies across those 768 rows describe model behavior under the selected design; they are not real-world event probabilities. Censored scenarios remain in the analysis at the 5,000-year horizon.
Used for: Empirical validation gates; Seeded scenario testing
Status: Research governance control; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: The baseline uses 244 dated state-system membership spells from COW v2024. Membership exits are observed events, but their causes have not yet been coded as consolidation, fragmentation, dissolution, or interruption. The hazard is a benchmark, not validation of the full model.
Used for: COW empirical membership baseline
Status: Observed historical data — limited outcome semantics; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: The horizon changes what is visible, not the underlying event rate. Calendar years through 7026 are simulation stress-test coordinates without predictive or evidentiary status. Long horizons expose asymptotic model behavior rather than forecast historical time. One-at-a-time and Morris-style screening identify influential assumptions. Phase maps, observability cases, and wave-calibration cases compare model structures; their spread is not a confidence interval around one forecast. Four independent Latin-hypercube ensembles propagate transparent parameter distributions through the seeded binomial transition model. Non-reaching times remain censored. Transformed bell curves respect parameter bounds; external fragments use a zero-inflated count distribution. T₅ is the first modeled year with at most five independent top-level Institutions. Boundary comparisons lock the starting field at 225. Two years is the proven lower bound under current clamps; 5,000 years is an attained display-horizon boundary, not a natural maximum. Twelve archetypes receive 64 seeded perturbations each. Frequencies across those 768 rows describe model behavior under the selected design; they are not real-world event probabilities. Censored scenarios remain in the analysis at the 5,000-year horizon.
Used for: Model horizon; Model sensitivity and structural alternatives; Monte Carlo parameter-uncertainty propagation; First year at five top-level Institutions; Seeded scenario testing
Status: Analyst-selected display parameter; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: The model assumes that AI-mediated authority can preserve objectives, records, threat models, enforcement logic, biases, and grievances across succession and generational turnover. Historical forgetting, economic reset, administrative decay, and blind spots therefore stop automatically reopening the field. Persistent memory can preserve competence or inherited prejudice; deliberate unlearning is not modeled. New organizations may still form, but only as curated subordinates after Apex. AI and singleton sources motivate the mechanism; they do not prove permanence. Apex coordination, permanent human disempowerment, and biological extinction are separate outcomes. The current equations calculate only Apex coordination. They do not determine whether the result becomes AI constitutional governance, human-elite displacement, a human–AI institution, civilizational succession, permanent disempowerment, or biological extinction. Permanent disempowerment is an existential catastrophe, not biological extinction.
Used for: AI-mediated institutional power calcification; Civilizational outcomes after Apex coordination
Status: Author-defined core premise and falsifiable hypothesis; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: Subordinate agents receive effective-alignment credit only when material data, core applications, infrastructure, dependencies, logs, lineage, and recovery paths are visible to Apex-authorized governance. Detected material concealment always triggers remediation; detection itself remains imperfect.
Used for: Mandatory observability and anti-concealment
Status: Core governance assumption; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: Forty definition specifications and four leave-one-wave-out estimates test calibration fragility. Seven models are compared in six chronological test blocks. The target is COW membership exit, not permanent subjugation or Apex emergence.
Used for: COW wave robustness and model comparison
Status: Historical diagnostic — L1 sourced-exploratory; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: Subordinate agents receive effective-alignment credit only when material data, core applications, infrastructure, dependencies, logs, lineage, and recovery paths are visible to Apex-authorized governance. Detected material concealment always triggers remediation; detection itself remains imperfect.
Used for: Mandatory observability and anti-concealment
Status: Core governance assumption; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: Subordinate agents receive effective-alignment credit only when material data, core applications, infrastructure, dependencies, logs, lineage, and recovery paths are visible to Apex-authorized governance. Detected material concealment always triggers remediation; detection itself remains imperfect.
Used for: Mandatory observability and anti-concealment
Status: Core governance assumption; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: The 15% logistic rate is illustrative. Stanford reports rapid but uneven AI adoption and growing national AI-sovereignty efforts. A robust version should fit country- and sector-specific adoption curves rather than one global rate.
Used for: AI architecture diffusion rate
Status: Published diffusion method; uncalibrated default; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: A future dimension would classify a theologically or ideologically driven Institution as intolerant when tolerance and acceptance of nonmembers, dissenters, and peaceful parallel institutions are not core governing commitments. No current entity is scored. Doctrine, formal rules, and observed behavior require separate evidence and review; intolerance may increase consolidation attempts while also increasing resistance and fragmentation.
Used for: Pluralism tolerance and institutional intolerance
Status: Unmodeled author-defined classification proposal; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: Compatible AI systems can spread while states retain separate sovereignty and directives. Stanford's AI Index reports both rapid adoption and increased emphasis on national AI sovereignty; the model therefore keeps adoption and political consolidation separate.
Used for: Technical adoption versus governing authority
Status: Core conceptual distinction; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: The country count is anchored to 193 UN Member States and two Observer States. The 30-company set is revenue-led. Countries and firms are not legally or coercively equivalent. Peer-node treatment is a simplifying future-technology premise motivated by the author’s expectation that strategic firms become nationalized, state-directed, or equivalently dependent under sustained conflict; the model does not estimate that outcome.
Used for: 225 starting institutions
Status: Empirical anchor + selection rule; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: This fixed count represents pre-Apex secession, institutional replacement, or newly autonomous organizations. Once the model reaches one operational Apex center, later organizational births are curated subordinates and do not increase the independent-center count. Historical state creation does not validate that permanence rule.
Used for: New external fragments per event
Status: Author-defined transition with historical anchors; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: A future dimension would classify a theologically or ideologically driven Institution as intolerant when tolerance and acceptance of nonmembers, dissenters, and peaceful parallel institutions are not core governing commitments. No current entity is scored. Doctrine, formal rules, and observed behavior require separate evidence and review; intolerance may increase consolidation attempts while also increasing resistance and fragmentation.
Used for: Pluralism tolerance and institutional intolerance
Status: Unmodeled author-defined classification proposal; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
Publisher/author identified at linked source
Supports: The count-only transition combines beta-binomial wave losses, fixed external fragments, and binomial completed splits. It does not identify winners, parents, lineages, internal politics, or partial autonomy. Institutions remain exchangeable within each wave. The forced terminal shortcut has been removed. Two approximately equal centers can persist until a stochastic transition actually leaves one independent center; structural influence then jumps from about 50% to 100%. Effective Apex coordination additionally requires more than 50% internal standard adoption.
Used for: Agent-count transition equation; Apex Institution coordination threshold
Status: Author-defined mathematical assumption; linked in the reviewed interface registry
Limitations/access: Provides contextual, conceptual, or methodological support; it does not independently validate the model's numeric defaults or forecasts.
World Bank
Supports: License, attribution, and redistribution controls for WDI and WGI inputs.
Used for: Data governance and distribution restrictions
Status: Referenced in WDI and WGI data-use decisions
Limitations/access: Legal terms govern use; they do not evaluate analytical quality.
World Bank
Supports: Retrieval semantics and pagination for WDI source snapshots.
Used for: WDI ingestion implementation and provenance
Status: Referenced in the WDI data-use decision
Limitations/access: Documentation describes access mechanics, not substantive validity of individual indicators.