Define the public purpose
Convene service operators and appropriate oversight functions. Select one civilian use case, map the existing baseline, identify affected communities, and publish the questions the work is intended to answer.
Start with a bounded question.
Advance through evidence.
A proposal for moving from a concept to a reviewable evaluation. Each stage has work to complete, evidence to produce, and a decision to make. Progress is conditional; operational deployment is not presumed.
01Define the mandate02Design a controlled evaluation03Test interfaces and local fallback04Run a bounded civilian exercise05Publish a decision and reassess
An illustrative first 100 days
A policymaker could use an initial 100-day period to establish scope, authority, and evaluation criteria. This is a proposed planning agenda, not an approved timetable or a promise of operational capability. External agreements, independent review, and funding may take longer.
Set the boundaries
Document decision rights, information categories, provider dependencies, costs, privacy and protection concerns, and proposed stop conditions. Test whether a simpler improvement to existing channels could meet the need.
Publish a reviewable evaluation plan
Invite independent scrutiny of the comparison, measures, and safeguards. Decide whether a controlled exercise is justified. Publish the decision and unresolved questions; defer any next stage that lacks the necessary authority or protections.
01. Define the mandate
Choose one civilian problem and establish who is responsible for addressing it.
Work to complete
- Identify the service, disruption, participants, intended beneficiaries, and existing response process. Include operators and representatives of affected communities in scoping.
- Map authority, data categories, provider dependencies, and any constraints on cross-border sharing. Identify which questions require qualified legal or regulatory review.
- State the hypothesis, alternatives, exclusions, budget assumptions, and reasons a new arrangement might be unnecessary.
- Deliverable
- A public concept note, a responsibility map, a preliminary impact assessment, and an initial cost model.
- Decision gate
- Proceed only when the purpose is narrow enough to test and each material responsibility has an identified owner.
- Pause or stop when
- Pause if the proposed exchange lacks a defensible authority, a clear civilian purpose, or a practical path for independent review.
02. Design a controlled evaluation
Specify what would count as evidence before constructing a demonstration.
Work to complete
- Define an existing-practice baseline and a simpler alternative. Choose outcomes and comparison conditions suitable for the use case.
- Document assumptions, uncertainty, failure cases, and data limits. Set reviewable criteria for interpreting operational and rights impacts.
- Agree on independent evaluation, protected access, disclosure, incident handling, and who may suspend the exercise.
- Deliverable
- A versioned evaluation plan, measurement definitions, approved data plan, and written stop conditions.
- Decision gate
- Proceed only after reviewers can explain how the evaluation could disprove or substantially weaken the hypothesis.
- Pause or stop when
- Pause if the proposed measures are unverifiable, if harms are excluded, or if the comparison is designed to favor the demonstration.
03. Test interfaces and local fallback
Check the proposed exchange in a controlled environment before any live operational reliance.
Work to complete
- Use synthetic, public, or otherwise appropriately authorized information. Verify that formats, provenance records, and uncertainty are understandable to the intended recipients.
- Test disconnection, revocation, provider outage, delayed messages, and disputed warnings without putting essential services at risk.
- Record integration work, staff burden, verification delays, and the cost of maintaining a usable alternative.
- Deliverable
- An interoperability report, a dependency register, a fallback demonstration, and revised cost assumptions.
- Decision gate
- Proceed only when participants can verify messages, revoke sharing, and continue their essential functions independently.
- Pause or stop when
- Pause if fallback fails, access cannot be revoked, or the exchange exposes information beyond its approved purpose.
04. Run a bounded civilian exercise
Compare the proposed process with current arrangements under documented conditions.
Work to complete
- Use an agreed scenario and authorized participants. Separate the observation of a disruption from claims about cause or intent.
- Measure the agreed service and process outcomes. Record false alarms, missed warnings, delays, unequal impacts, and privacy or governance failures.
- Keep reviewers independent from delivery targets. Preserve evidence, deviations from the plan, and reasons for emergency exceptions.
- Deliverable
- An exercise record, comparison results, an incident account, and an independent assessment of limitations.
- Decision gate
- Consider further work only when the evidence can be independently reviewed and the results meet the pre-agreed criteria.
- Pause or stop when
- Suspend immediately for unauthorized sharing, unacceptable civilian harm, failed operational control, or an inability to audit material decisions.
05. Publish a decision and reassess
Decide whether to stop, revise, repeat, or propose a separately authorized next stage.
Work to complete
- Publish results with uncertainty and the full scope of measured costs and harms. Explain which claims the evidence can and cannot support.
- Require a response to independent findings and a documented renewal decision. A demonstration does not automatically authorize an ongoing program.
- Set an expiry or review date for any continuation. Repeat the assessment when participants, vendors, purposes, data, or operating conditions change.
- Deliverable
- A public decision record, corrective-action commitments where appropriate, and a bounded proposal for any next stage.
- Decision gate
- Expand only with a defensible evidence base, updated authority and safeguards, and an explicit funding and oversight decision.
- Pause or stop when
- End or redesign the activity when benefits are not established, harms exceed the agreed limits, or essential safeguards cannot be sustained.
Resources and funding decisions
No budget, savings estimate, or staffing commitment has been established for MAR. A credible request would need to price the actual use case, not a generic promise of AI-enabled resilience.
A scoped cost model should include service-operator time, integration, secure communication, independent verification, privacy review, accessible warnings, training, evaluation, maintenance, and withdrawal. Recurring obligations should be shown separately from demonstration costs.
Funding could be divided into limited stages, each with a stated deliverable and an explicit continuation decision. An inconclusive result should lead to a reasoned decision to revise, repeat, or stop, rather than an assumption that a larger deployment will resolve uncertainty.
Who would need to participate?
- The institution responsible for the civilian service and the operators who understand its practical constraints.
- Appropriate legal, privacy, civil-liberties, procurement, and security review functions.
- An evaluator with the independence and access needed to assess results and harms.
- Representatives of affected communities, including people whose access to warnings or essential services may be uneven.
- Authorized counterparts for any proposed cross-border exchange, with clearly defined responsibilities.
Evidence status
All 20 source connections remain untested hypotheses. The platform and implementation plans are independent proposals. The available material establishes no validated MAR effect, combined outcome, or probability of success. Review the evidence standard.