A decision begins with an investigation.
Pick a scenario, make your own next-step call, then reveal the frozen candidate's reference route and reasoning. The reference results below were computed from authored synthetic fixtures; the page is a teaching walkthrough, not a live execution of the Python engine or a validated answer key.
What should happen next?
What this does and does not establish
These are synthetic cases with outcomes written in their author's expectations and calculated by the frozen Python kernel during development. An agreement does not establish a correct real-world decision. Economic assumptions, feasible choices and approval rights still require independent evidence. All results are provisional and subject to review.
Five checks. One proposed next action.
A navigator earns its place by explaining why to investigate further—and when to stop. These checks are based on established decision-analysis and governance principles; their integrated sequence is being tested, not claimed as an original theorem.
Can we act lawfully?
Identify binding requirements, the authority holder and whether an authorised option can meet the deadline.
What expired?
Record overdue evidence and approved fallback. Missing approval triggers escalation—not implied consent.
What are we optimising?
When competing objectives imply different choices, return the decision to the accountable owner.
Could evidence matter?
Screen all plausible uncertainties. A possible upside is an upper bound, never proof an investigation pays.
Will a real test pay?
Compare specific test accuracy, cost, delay and feasibility. Stop when no justified timely study has net value.
Record the reason.
State the next action, evidence requested, owner, cost and deadline. Humans retain approval and legal responsibility.
Be precise about what we know.
This release demonstrates the navigation experience with frozen example outcomes. It does not claim that the method outperforms established Decision Quality, that scenario answers are ground truth, or that the browser executes the research-grade Python kernel.
Until an independent comparison with conventional problem structuring and a minimal Decision Contract, this remains research software and a teaching aid.
- Validate every payoff, uncertainty range and source independently.
- Verify statutory and institutional authority from governing documents.
- Never upload confidential customer, bank, medical or public-sector records.
- Test for missing alternatives, scenario interactions and evidence acquired too late.
- Do not interpret a proposed action as organisational permission.