Skip to content

viz

Visualization helpers. Most practitioners will understand the framework through these.

Requires: pip install minimal-oversight[viz]

plot_masking_dashboard

minimal_oversight.viz.plot_masking_dashboard(node_names, sigma_raw, sigma_corr, masking_threshold=1.3, figsize=(12, 5))

Side-by-side bar chart of σ_raw vs σ_corr with masking index.

Nodes with M* > threshold are highlighted.

Side-by-side view: \(\sigma_\text{raw}\) vs \(\sigma_\text{corr}\) (left) and \(M^*\) per node (right). Nodes above the masking threshold are highlighted in red.

from minimal_oversight.viz import plot_masking_dashboard

fig = plot_masking_dashboard(
    node_names=["gen", "rev", "test", "merge"],
    sigma_raw=[0.46, 0.38, 0.33, 0.13],
    sigma_corr=[0.69, 0.60, 0.56, 0.42],
    masking_threshold=1.3,
)

plot_autonomy_buffer

minimal_oversight.viz.plot_autonomy_buffer(c_op, p_min, governance_gap, h_w_range=None, figsize=(8, 5))

Plot autonomy buffer B_eff as a function of process entropy H(W).

Shows the capacity cliff at H_crit.

Shows \(B_\text{eff}\) as a function of process entropy \(H(W)\), with the capacity cliff at \(H_\text{crit}\) marked. Blue region = feasible; red region = infeasible.

from minimal_oversight.viz import plot_autonomy_buffer

fig = plot_autonomy_buffer(c_op=0.86, p_min=0.75, governance_gap=0.02)

plot_pipeline_risk

minimal_oversight.viz.plot_pipeline_risk(node_names, sota_scores, masking_indices, figsize=(10, 5))

Horizontal bar chart ranking nodes by governance priority.

Combines SOTA score with masking index annotation.

Horizontal bar chart ranking nodes by SOTA priority score \(S(v)\), annotated with \(M^*\). Colors: red (\(M^* > 1.5\)), orange (\(> 1.2\)), green (healthy).

from minimal_oversight.viz import plot_pipeline_risk

fig = plot_pipeline_risk(
    node_names=["rev", "gen", "test", "merge"],
    sota_scores=[1.32, 0.45, 0.62, 0.79],
    masking_indices=[1.77, 1.35, 1.49, 1.65],
)

plot_scope_frontier

minimal_oversight.viz.plot_scope_frontier(sigma_raw, p_min, coverage_range=None, figsize=(8, 5))

Coverage-cost frontier plot.

Shows how total governance cost increases with required coverage, and how average delegated competence decreases.

Coverage-cost frontier: how total governance cost increases (blue, left axis) and average delegated competence decreases (red, right axis) as coverage requirements tighten. Demonstrates why coverage constraints are not decorative — they prevent cherry-picking easy tasks.

from minimal_oversight.viz import plot_scope_frontier
import numpy as np

fig = plot_scope_frontier(
    sigma_raw=np.random.default_rng(42).beta(3, 2, size=50),
    p_min=0.30,
)