Skip to content

Milliseconds matter — design the routines people can adopt

Use for repeated interactions, information access, direct manipulation, or a novel control whose value depends on how perception, thought, and action fit together. Carry the relevant insight into the existing synthesis and working artifact. Load this reference selectively alongside the primary channel.

What the research contributes

Gray and Boehm-Davis (2000) analyze microstrategies: small arrangements of cognitive, perceptual, and motor operations. Their CPM-GOMS models represent dependencies and overlap; the critical path determines predicted duration. In their button experiment, a predictable return location supported earlier movement than an unknown next target. Their separate two-person prototype study used model discrepancies to diagnose possible cue and verification problems. Those diagnoses were hypotheses, rather than experimentally confirmed causes. Their framing joins embodied cognition, task, and artifact.

Gray and Fu (2004) distinguish what an interface permits from the costs that favor one routine over another. In two experiments, access costs encouraged use of imperfect memory despite accurate external information, with worse performance. External availability alone did not determine information use.

Gray, Sims, Fu, and Schoelles (2006) tested temporal resource tradeoffs through three Blocks World experiments and an Ideal Performer Model. Their account opposes a universal preference to conserve memory. It concerns local interactive routines; locally economical behavior need not produce the best overall result.

Put the insight into the interaction

These are design applications to test in the host product:

  • Map a consequential routine: finding information, acquiring a target, recalling context, acting, waiting, checking, and continuing. Ask which steps depend on others and which could overlap. A click count misses these choices.
  • Compare the cost of consulting external information with remembering or guessing. Consider reveal gestures, gaze travel, window changes, waiting, and returning to the task. Make information practical to use where it matters.
  • Inspect predictability and state cues. Can people prepare the next action? Is a pause useful verification, uncertainty, or an interface-imposed wait?
  • Explore inventions that change these relationships: an adjacent comparison, persistent reference, stable target, or preview during direct manipulation. Name the routine the candidate enables and the tradeoff it introduces.
  • Observe the strategy as well as elapsed time: consulting, recalling, rechecking, errors, and recovery. Compare early use with practiced use when skill acquisition matters. Treat inferred internal operations as hypotheses.

For example, compare a delayed hover reveal with an adjacent persistent value in a tuning tool. Predict whether people will consult the value or estimate from memory; observe consultation, accuracy, and interference with the task. This proposed comparison is an extrapolation, not a result of Gray's studies.

Interpretation limits

These studies inform routine interaction, not a universal objective to minimize time. Preserve useful verification and meaningful task goals. Do not copy their operator durations into modern performance budgets or invent a savings estimate. Model estimates need stated assumptions; measured benefits need task evidence. The interaction-contract response budgets answer a different question: how the product acknowledges and reports work.

Sources and intake basis