What happened?
Record the concrete decision, omission, output, deadline, or condition before compressing it into a slogan.
PROJECT SHADOW 1.0.1 · CORRECTED R1 REFERENCE · PRELIVE · 2026-08-17
Project Shadow 1.0.1 contains no Myth package. Generic Myth v0.2.0 and Full-Canon Myth v0.3.5 are separate optional companions; both default off, neither is required by R1, and neither can authorize action or change an R1 result. ARM remains a civic repair workshop, not a certification service; publication transfers no authority and is not operational deployment, efficacy, safety, certification, a legal-compliance finding, permission to act, or a global green.
ECOSYSTEMARM
One civic lifecycle / six distinct public doors
The RecordremembersCivic QAinspectsPBHPpausesARMrepairsProject Shadowinstruments + testsALMSIVIpreserves optional originCHAMBER 01 / FAILURE MAP
Diagnose the system—not just the scandal.
Public argument often begins with a person, headline, party, or model output. Diagnosis asks what machinery made the outcome repeatable, who carried the burden, and which failed control allowed it through.
THE DIAGNOSTIC STACK
Move from event to recurrence mechanism.
A root cause is not merely the closest error. It is the condition that keeps reproducing the error after the headlines, apologies, personnel changes, and emergency fixes disappear.
Who absorbed it?
Map harm, delay, cost, uncertainty, and loss of agency—not just the institution’s aggregate result.
What produced it?
Trace authority, incentives, information flow, interfaces, budgets, dependencies, automation, and discretion.
Why will it happen again?
Identify which unchanged condition rewards, permits, hides, or normalizes the same result.
Why didn’t the safeguard stop it?
Test whether the control was absent, weak, bypassed, captured, unaudited, or impossible to use.
CONNECTED CONSEQUENCE ENGINES
The same failure can cross politics, institutions, and AI.
A political promise can become an agency rule, a procurement contract, a model threshold, a frontline decision, and finally a consequence for one person. Inspect the complete chain.
Political power
Elections, parties, legislatures, executives, courts, donors, media, and organized movements convert stories into authority.
- ASK
- What changes between elections, and who can verify it?
Institutional systems
Agencies, procurement, benefits, policing, schools, health systems, and workplaces turn policy into daily consequence.
- ASK
- Who pays first, who can appeal, and can the harm be reversed?
Automated judgment
Models, ranking systems, agents, data pipelines, and vendor tools can scale a hidden assumption into thousands of decisions.
- ASK
- What is the action boundary, and where is the independent stop?
REQUIRED RECEIPTA traceable path from authority to mechanism to consequence, with every handoff and unknown visible.
WORKED PATTERNS
Change the scale. Keep the questions.
The method should work on a neighborhood service, a federal statute, a vendor platform, or an agentic system without pretending those contexts are identical.
A benefits portal repeatedly rejects complete applications.
Do not stop at user error. Inspect form design, language access, vendor rules, staffing, exception authority, appeal time, and rejection distribution.
A lawful program still produces a predictable access gap.
Separate legality from adequacy. Trace implementation burden, county variation, notice, funding, enforcement, and whether affected people can recover.
A bill promises accountability but delegates the hard choices.
Map the rulemaking authority, deadlines, appropriations, exemptions, oversight, private right of action, and conditions that survive a change in administration.
A model prioritizes inspections, reviews, or enforcement.
Inspect objective, data coverage, proxy variables, action threshold, human override, distribution shift, appeals, logging, and the consequence of a false positive.