intentional-uxlisted

# Intentional UX Analyze an experience as a path from a person's intention to evidence that the named outcome occurred. Prefer evidence over impressions, and preserve uncertainty rather than compressing it into a single UX score. ## Composition contract Intentional UX owns the task tuple, state graph, evidence levels, interaction costs, and experience acceptance checks. The most specific domain skill owns implementation. When Exemplar is active, Exemplar sets the quality target and anti-performance filter; it does not replace path evidence or relabel an Assumed finding as Measured. Chef's Choice may select useful capabilities but does not change these ownership boundaries. Accessibility remains authoritative for semantic and assistive-technology compliance. For an exceptional experience pass, use the task graph and evidence to choose the highest-value improvement, then require both observable success evidence and Exemplar's completion gate. ## Start with a task State one tuple for each materially different person or input mode: ``` person/input mode -> start -> outcome -> success evidence ``` Examples of input mode include touch, pointer, keyboard, voice, and switch access. “Success evidence” is the observable confirmation the person needs, not just a system state change. Then model the path as a state graph. Include the ordinary path plus any confirmation, wait, correction, recovery, abandonment, and context-switch states that can change completion or effort. See [t