Feedback and Feedforward
Feedback = communicating the results of an action and the current system state (from control/information theory). It answers "What happened?" — bridging the Gulf of Evaluation. See The Gulfs of Execution and Evaluation.
Feedforward = information that tells you what you can do and how, before acting. It answers "What can I do? How?" — bridging the Gulf of Execution. Delivered via Signifiers, constraints, mappings, and the conceptual model.
Requirements for good feedback
- Immediate — even a 0.1s delay is disconcerting; long delays make people give up.
- Informative — cheap beeps/blinks tell you something happened but not what or what to do; poor feedback can be worse than none (distracting, anxiety-provoking).
- Prioritized — unimportant info unobtrusive; important signals capture attention. In emergencies, uncoordinated alarms create dangerous cacophony (operators waste time silencing them).
- Not excessive — too much feedback is worse than too little ("backseat driver"); people disable over-signaling devices and then miss critical alerts.
Relationship to models
Both feedback and feedforward must be presented in a form matching the user's goals and expectations, and both rely on a good conceptual model. At the behavioral level, feedback resolves expectations → satisfaction or frustration (see Three Levels of Processing — Visceral, Behavioral, Reflective).
Confidence
1.0 — explicit; feedforward named explicitly in ch.2.
Sources: ["DOET ch.1 'Feedback' p.23-25; ch.2 p.71-72"]