Correction Protocol

Correction begins not with a desire to change something but with a bounded finding. The action must answer to what has been documented and to the effect that must be corrected or safeguarded.

Correction must not become a new frame exempt from testing. It must itself remain answerable to effect, side effects, and new material.

The levels express action type, not moral severity. Several levels may be combined, but each must be justified separately.

Finding category provides direction but does not automatically determine the action. Consequence, uncertainty, reversibility, urgency, and binding requirements must be assessed separately.

The action must be recorded as a plan that can be implemented and reviewed. Vague formulations such as “should be strengthened” or “must improve” are insufficient.

• Finding ID and the part of the finding addressed.
• Concrete change, scope, and limitation.
• Responsible role and decision authority.
• Start, deadline, and intermediate points.
• Resources, dependencies, and binding requirements.
• Expected effect and possible side effects.
• What will be measured, observed, or documented.
• Criteria for completion, continuation, alteration, or rollback.
• Who performs independent review.

Auditor, decision-maker, and implementer may be different roles. It must be visible who recommends, who decides, who implements, and who tests the effect.

• Clarify impartiality and conflicts of interest.
• Distinguish technical assessment from formal decision.
• Make responsibility for non-implementation visible.
• Ensure that effect is not reviewed only by the person reviewing their own work.

Before substantial correction becomes final, affected parties should be able to comment on facts, feasibility, side effects, and alternative actions.

Comments must be assessed against the finding. They must not merely be archived as an appendix without a visible response.

The fact that an action was performed does not show that the condition was corrected. Review must examine actual effect and whether new errors or capture forms emerged.

When an action does not work, this must be recorded as a new methodological condition, not concealed by extending the original explanation. The action may have been wrongly selected, poorly implemented, or directed at the wrong element.

• Suspend or roll back actions that create greater error or risk than the condition they were meant to correct.
• Reopen the finding where new material alters the grounds.
• Distinguish implementation failure from effect failure.
• Close the matter only when finding, action, effect, residual uncertainty, and continuing responsibility are documented.
• Preserve revision history so former grounds and changes remain traceable.