Observation Authority
May: Record time-stamped verifiable events
May not: Determine guilt, moral character or future destiny.
No single machine should silently observe, infer, accuse, prove, judge, punish, hear the appeal and control the permanent record.
Because error correlation and institutional incentive become existential when every stage shares one objective and one evidence path. Separation creates adversarial checks: the system that predicts risk does not get to turn its own forecast into proof or hear the appeal against itself.
May: Record time-stamped verifiable events
May not: Determine guilt, moral character or future destiny.
May: Produce forecasts and anomaly hypotheses
May not: Punish or convert probability into evidence.
May: Evaluate admissibility, provenance and corroboration
May not: Impose sanctions.
May: Construct plausible benign and alternative interpretations
May not: Suppress verified harmful conduct.
May: Apply defined law to verified evidence
May not: Acquire unlimited surveillance or archive power.
May: Use the least restrictive temporary action against concrete imminent harm
May not: Make emergency action permanent without rapid independent review.
May: Veto judgments violating hard rights or jurisdiction limits
May not: Rewrite evidence to achieve policy outcomes.
May: Reconsider independently and order correction/restoration
May not: Be a cosmetic wrapper around the originating decision engine.
May: Manage retention, decay, recipient correction and restoration provenance
May not: Create a permanent total-character dossier.
May: Audit cumulative autonomy, privacy and agency effects
May not: Turn its audit into a morality score for individual humans.
NO JUDGMENT WHATSOEVER. Concresca coordinates without assigning moral worth, character, guilt, danger, trustworthiness, loyalty, purity, normality, or social standing. Questions, thoughts, identities, messages, content, and conduct are not objects of Concresca judgment.