AGI Compute Diffusion & Post-AGI Governance Equilibrium Simulator

Scenario Presets

Model Parameters

Top 1% Compute Share 15%

Share of total planetary compute held by top 1% nodes.

Open Weights Accessibility 0.85

Effective algorithmic efficiency ratio available to non-monopoly nodes.

Enforcement Cost Multiplier 1.40

Cost ratio required for central power to raid or suppress sovereign compute.

Initial Sovereign Nodes 100

Number of autonomous sovereign hardware clusters modeled.

Model Dynamics

Directly drag compute sliders or click nodes in the heatmap grid to simulate hardware transfers or confiscations. Telemetry re-evaluates Gini index (G) and Centralization Index (C) instantly via linear algebra computations.

Gini Index (G)

0.12

Compute Inequality Metric

Governance Centralization (C)

0.19

Equilibrium Threat Level

Enforcement Friction

1.40

Defense Cost Ratio

Active Compute Nodes

100

Simulated Network Entities

Lorenz Compute Curve D3.js Realtime

Compute Share Density Grid Interactive Node Heatmap

Click any cell to toggle regional compute boost or containment.

Compute diffusion: normalized shares and calibrated indices

Your browser does not support video.

Read the explanation

The saved model creates one hundred nodes, with the top one percent rounded to at least one node. At a fifteen percent entered top-group share, that one node initially receives point one five. The remaining point eight five is divided across ninety-nine nodes using rank to the negative one-quarter power, then normalized. An accessibility value point eight five adds an equal floor to every non-top node. The total added mass is point eight five times point eight five times one half, or point three six one two five. Normalizing the new total reduces the original top node share to point one five divided by one point three six one two five: about eleven point zero one nine three percent. Both bars share thirty pixels per percentage point. The slider is an input share before this transformation, not necessarily the final displayed concentration. The underlying Gini formula sums every pairwise absolute share difference and divides by twice node count squared times mean share. Two equal shares yield zero; point two five and point seven five yield point two five. These illustrative bars share one thousand pixels per raw Gini unit. The simulator sorts effective shares for its Lorenz curve, but the reported Gini is not always that raw inequality. At top-share point one five and accessibility point eight five within a point zero zero one tolerance, it assigns Gini point one two directly. Elsewhere it multiplies raw Gini by one point one minus point six times accessibility and rounds to two decimals. Thus the headline index is a calibrated score and may not match the raw Lorenz distribution. It is not measured geopolitical inequality or evidence of future AGI diffusion. At the calibrated share, accessibility and enforcement multiplier one point four, centralization is assigned point one nine. Otherwise it combines the reported Gini with accessibility and inverse enforcement cost, rounds and clips between point zero one and point nine nine. The first regime boundary is point two five, shown on a shared two-thousand-pixel index scale. Values below that boundary receive a decentralized label, values below point five five a federated label, and larger values a centralized label. These labels follow the arithmetic; they do not establish causal political outcomes. Enforcement does not change the node-share distribution itself. After sorting, top flags are assigned to the largest final shares, so they need not identify the originally allocated top group under every input. Exported nodes and indices retain these assumptions, not a verified forecast. No external model or backend request is made.