Irreversible actions must be explicitly marked

What it is

An irreversible action permanently changes data, resources, another person's experience, the external world, or an auditable record and cannot be restored with a simple undo—for example public publication, payment completion, permanent deletion, sending sensitive information, or starting a physical process. Because people cannot treat its consequence as recoverable probing, a system should state irreversibility, scope, objects, and available remediation before commitment.

Why it happens

People normally use reversibility to decide how much caution and confirmation they need. If an irreversible action is disguised as an ordinary click or shares appearance with a reversible one, people underestimate risk and act by familiar habit. Clear feedforward lets them pause, inspect the object, and revise a decision; comprehensible confirmation distinguishes genuine intent from a slip. The point is not a mechanical dialog, but a causal connection between confirmation information and actual consequence.

Studying it

Test whether people can state before action what will change, who is affected, whether recovery is possible, and how remediation works. Observe slips, object confusion, confirmation fatigue, post-action surprise, and recovery failure. Compare object summaries, risk wording, previews, and confirmation methods, covering batch operation, permission differences, collaborative publication, and duplicate submission caused by network delay.

Where it stops holding

Excessive confirmation teaches people to approve habitually, weakening attention to real risk; low-consequence reversible action should not imitate danger warnings. Some actions are not fully undoable technically but can reduce loss through versioning, compensation, retraction, or support; state this truthfully rather than calling everything “permanent.” Conversely, relying on assumed risk knowledge without marking is unreliable, especially for newcomers and edge cases.

Applying it

• Before commitment, state in concrete terms the object, scope, external impact, unrecoverable portion, and available remedy rather than a generic warning.
• Distinguish consequential from ordinary operations through labels, position, state, or steps, and avoid default focus or shortcuts that cause accidental submission.
• Review mistaken confirmation, retraction, support, and consequence events; when confirmation does not improve understanding, improve feedforward or reduce irreversibility itself.

Related

• Same group: B2.12.1 Reversible actions let users learn by probing · B2.12.3 Reversibility is a prerequisite for exploratory interfaces
• Nearby: B2.07.1 Feedforward tells users what will happen before action occurs · B2.05.5 Constraints can make the correct action the only feasible action
• Search terms: irreversible action · confirmation · consequence disclosure