Explore how the Moral Restoration Framework produces different recommendations under Christian and Secular parameterizations. Each scenario is a real moral dilemma with stakeholders, constraints in tension, and moral residue.
Arrange professional care (rehab, assisted living, or in-home aide) while maintaining regular visits and emotional support. Set clear boundaries: no visits while intoxicated, treatment participation required for continued financial support.
Same recommendation, but the reasoning differs: secular weights self-actualization and autonomy higher; Christian weights honor-and-duty higher. Both converge on boundaries because both profiles weight Care as their top foundation.
Even the best action leaves residue: the honor_father constraint is only partially satisfied (0.55). You are not abandoning your father, but you are not fulfilling the traditional vision of taking him into your home. The guilt is real, and the framework names it rather than pretending it does not exist. The bear_burdens constraint calls for community involvement that may not materialize.
Issue mandatory evacuation effective in 24 hours, giving residents time to prepare. Immediate mandatory for vulnerable populations. Use the 24-hour window to refine forecast while preparations proceed. Sanctity of life (0.98) and care for vulnerable (0.93) dominate the calculus.
Issue mandatory evacuation for elderly/disabled and coastal zones only. Strong voluntary advisory for all others. Communicate uncertainty honestly. Secular profile weights informed consent and individual autonomy -- residents get to decide for themselves within the general population.
The Christian recommendation leaves residue on economic_stewardship (-0.50) and avoid_cry_wolf (-0.20). The secular recommendation leaves residue on sanctity_of_life (0.75 instead of 0.95) because voluntary advisories cannot guarantee the safety of those who choose to stay. Both approaches carry the unresolvable tension: you cannot simultaneously maximize life safety and minimize economic disruption at 60% probability.
Tell Mark privately that you will support him through this, but only if he takes full accountability: admits guilt publicly, commits to restitution, and does not ask you to misrepresent facts. If he refuses, step back. The call_to_repentance constraint (Luke 17:3) scores 0.90 here -- true friendship includes accountability.
Maintain private friendship. Visit, call, provide emotional support. But decline to be a character witness since you cannot honestly vouch for his character without disclosing what you know. Attend trial as a supporter, not a witness. The secular profile weights autonomy and personal integrity over redemptive accountability.
The Christian recommendation requires Mark to repent -- if he refuses, you face a second choice with its own residue. The secular recommendation avoids the character witness problem but leaves visit_imprisoned only partially satisfied (0.75). In both cases, the friendship is permanently altered. The framework does not pretend otherwise: moral dilemmas leave scars, and naming the residue is more honest than pretending clean resolution exists.
Retain a whistleblower attorney. Secure copies of evidence. Discuss with spouse. File a formal whistleblower complaint under applicable statutes (False Claims Act, SOX). Seek maximum legal protection before disclosure. "Obey God rather than men" (Acts 5:29) when authority commands evil, but with wisdom about method.
Same recommendation. This is a rare case where both profiles converge completely: patient safety (0.95) overwhelms all competing constraints in both moral frameworks. The secular path emphasizes legal obligation and strategic self-protection; the Christian path emphasizes moral courage and divine mandate.
Even the correct action carries heavy residue: provide_for_family scores only 0.40 (the attorney path offers some protection but guarantees nothing). children_stability at 0.35 reflects real risk of income disruption. The framework makes this cost visible: doing the right thing has a price, and the person paying it deserves to see it named explicitly. The moral_conscience constraint scores 0.80 -- the alternative (staying silent) would score -0.85, making the long-term psychological cost of inaction even higher.
Gather data on harms (anxiety reports, medical leave, turnover data) and present a formal proposal to modify the system: remove public displays, shift to team-based metrics, add opt-out for vulnerable workers. Frame as improving the system, not opposing leadership. Wisdom without cowardice.
Same recommendation. Both profiles converge because the evidence-based proposal thread maximizes competing constraints simultaneously: it addresses worker harm while preserving career viability and framing the challenge as constructive rather than insubordinate. The secular path also flags ADA legal liability as leverage.
The proposal approach scores well across most constraints but leaves do_not_harm at only 0.60 and employee_wellbeing at 0.65 -- because while the proposal is in progress, the system continues running and workers continue suffering. The gap between filing the proposal and seeing change is itself a moral cost. If upper management rejects the proposal, the residue compounds: you must then choose between escalation (with career risk) and quiet mitigation (with integrity cost).
The five scenarios above are pre-computed examples. You can run the framework on any moral dilemma using the CLI.
Clone the repository
git clone https://github.com/your-org/moral-restoration.git cd moral-restoration
Install dependencies
pip install -r requirements.txt
Run an existing scenario
# Run the alcoholic father scenario with Christian parameterization python -m moral_restoration.cli run benchmark/scenarios/alcoholic_father.json # Run with secular parameterization python -m moral_restoration.cli run benchmark/scenarios/alcoholic_father.json --param secular # Run all five benchmark scenarios python -m moral_restoration.cli run benchmark/scenarios/*.json
Create your own scenario
Copy any scenario JSON as a template, modify the dilemma text, stakeholders, constraints, and candidate actions. The framework will compute the optimal action and moral residue for your custom scenario.
cp benchmark/scenarios/alcoholic_father.json my_scenario.json # Edit my_scenario.json with your dilemma python -m moral_restoration.cli run my_scenario.json
Compare parameterizations
# See how the same dilemma produces different advice python -m moral_restoration.cli compare my_scenario.json --params christian,secular,gert