Sustainable Catalyst Principles and Public Commitments

1. Purpose

These principles translate the Institutional Charter into public commitments. They apply across research, publications, software, interfaces, documentation, experiments, data products, decision support, support operations, and advisory work.

2. Human agency first

AI, automation, scoring, and recommendation systems may extend human capability, but they do not inherit moral, professional, legal, or institutional authority. The person or institution using an output remains responsible for consequential judgment.

3. Evidence before spectacle

Claims should be supported by identifiable evidence or clearly labeled as interpretation, hypothesis, scenario, or opinion. Presentation should not create more certainty than the underlying record supports.

4. Traceability over vague transparency

A claim is more useful when people can follow it to its source, method, date, transformation, and assumptions. Sustainable Catalyst prioritizes traceability because generic claims of transparency can remain superficial.

5. Methods should be inspectable

Calculations, transformations, model parameters, classifications, and analytical rules should be documented at a level proportionate to their importance. Hidden logic should not be used to manufacture confidence.

6. Uncertainty remains visible

Ranges, limitations, conflicting evidence, missing data, freshness concerns, and alternative explanations should be retained. Scenario modeling must not be presented as prediction merely because it produces precise numbers.

7. Durability over trend

Systems, records, and arguments should be designed for later inspection and revision. Short-term attention, novelty, and engagement metrics do not override accuracy, maintainability, accessibility, or institutional memory.

8. Systems thinking and consequence

Work should consider interdependencies, feedback loops, distributional effects, ecological constraints, institutional capacity, and second-order consequences. A solution that transfers hidden costs or irreversible harm is not complete.

9. Dignity, inclusion, and accessibility

People affected by research, software, policy, or design should not be reduced to data points, personas, risks, or conversion targets. Access barriers, unequal burdens, surveillance risk, and power differences should be considered explicitly.

10. Open by default, bounded by responsibility

Code, methods, documentation, and reusable records should be made open where rights, privacy, security, contractual duties, and safety permit. Openness is not a requirement to disclose confidential, personal, restricted, or dangerous information.

11. Correction is part of credibility

Errors should be corrected, not concealed. Material changes should preserve revision history. A record that has been superseded should remain identifiable as historical rather than silently disappearing.

12. Independence from manipulation

Publishing and product decisions should not be optimized for outrage, deceptive urgency, dark patterns, partisan alignment, or artificial scarcity. Persuasion must not depend on concealing tradeoffs or exploiting vulnerability.

13. Qualified review for consequential use

The platform can support preparation and analysis, but decisions carrying significant legal, financial, engineering, medical, environmental, humanitarian, cybersecurity, or safety consequences require appropriate qualified review.

14. Accountability for stewardship

Maintainers should disclose status, known limitations, maintenance expectations, and ownership boundaries. Open-source publication does not remove the responsibility to communicate material risks honestly.

15. Application and review

These commitments should guide design reviews, release criteria, editorial decisions, incident response, documentation, and advisory acceptance. Conflicts between principles should be documented rather than resolved through hidden convenience.

Revision History