Status: PROPOSED

A0 — Repository foundation / authority

Scope:

• repo skeleton;

• constitution;

• ownership boundaries;

• privacy classes candidate;

• metric registry conventions;

• no runtime collection.

Acceptance:

• zero platform instrumentation;

• no storage service;

• doctrine only.

C1 — Counterpedia transport-aware counters

Goal: old-school counters, precisely defined.

Events:

• HTTP request;

• page render/view;

• API object read;

• MCP resource read;

• declared crawler read.

Metrics:

• http_request_count/v0.1

• page_render_count/v0.1

• page_view_count/v0.1

• api_object_read_count/v0.1

• mcp_resource_read_count/v0.1

Privacy: anonymous aggregate.

Non-goals:

• unique-person counting;

• fingerprinting;

• demographics;

• recommendation.

Public proof: a page can display: Web / MCP / API usage counts.

C2 — Counterpedia epistemic interactions

Events:

• source_open;

• claim_open;

• why_not_open;

• refusal_open;

• reopening_condition_open;

• history_open;

• edition_compare;

• verify;

• receipt_open;

• reverse_traversal.

Metrics:

• source inspection rate;

• verify rate;

• history pressure;

• reverse traversal rate;

• reopening demand.

Hard rule: activity never changes standing.

G1 — Graph/lattice analytics

No personal identity needed.

Metrics:

• source fan-out;

• claim dependency depth;

• source criticality;

• claim volatility;

• edge traversal counts;

• unresolved/refusal activity.

Outputs: Counterpedia internal/editorial + selected public aggregates.

M1 — MCP / agent transport analytics

Events:

• resource reads;

• tool calls;

• object classes;

• traversal steps;

• declared client metadata when supplied.

Metrics:

• claims vs sources vs receipts share;

• average graph traversal depth;

• verify-call rate;

• history-call rate;

• reverse traversal paths;

• machine-readable/prose preference by request type.

Do not infer:

• base model;

• owner identity;

• user intent.

F1 — Forum epistemic analytics

Metrics:

• source-grounded position count;

• shared basis extracted;

• disagreement dimensions;

• qualification count;

• explicit revisions;

• withdrawals;

• unresolved questions;

• sources introduced.

No likes/popularity ranking as authority.

T1 — Real trajectory analytics

Using Real Epistemic Response Trajectory corpus.

Metrics:

• narrative lag;

• action/posture divergence;

• evidence/response divergence;

• adoption depth;

• limitation rate.

Domain definitions remain typed.

P1 — Public Observatory

Consumes already-ratified aggregates.

Initial surface:

• web reads;

• MCP reads;

• API reads;

• source opens;

• verify actions;

• history opens;

• reverse traversals.

No new collection introduced solely for UI.

AM1 — Amnesiac private analytics

HOLD until public analytics substrate is proven.

Private/local candidates:

• acquisition curves;

• preference staleness;

• revisitability;

• belief revision;

• value drift;

• forecast calibration;

• hindsight divergence.

Default: LOCAL_ONLY.

No central collection required for the proving lane.

Recommended order

A0
→ C1
→ C2
→ G1
→ M1
→ F1
→ T1
→ P1
→ AM1

This maximizes public/agentic value first and delays sensitive/private analytics until the privacy substrate is proven.