Density control aims at task completion, not at showing as much available information as possible

What it is

Sensors can emit forty labels; this moment of the task needs “turn this one.” Drawing all forty is showing inventory, not helping anyone finish. Task-completion density scores overlays by whether the task completes and whether errors fall, not by whether every available fact was lit.

Show/hide, on-object binding, reduction, attention competition are means. This is what those means push toward: finishing, not filling.

Why it happens

Information has a search and interpretation cost. One more label is one more possible fixation and one more possible mis-bind. The product default is “show if we can” — missing one looks unfinished, one more looks like value. For the wearer, value turns into noise fast: the one needed to finish is buried in thirty-nine, time spent excluding. If the objective is written as “maximize coverage,” the optimizer wins coverage and loses completion.

Completion-orientation forces a ruthless cut: a layer with no causal link to the current success criterion, even if it “looks professional,” even if sales wants it, should not sit on the task path. It can live in a second level (look again, ask again). The default layer keeps only what can change the next action. The measure of the default is not item count. It is “does dropping this make completion fail or slow.”

Studying it

An information-dose experiment: the same task, overlays from 0, 1, the necessary set, necessary-plus-related, up to the full inventory. Plot completion time and errors against item count, looking for a U: too few starve for cues, too many stall in noise.

Dose: item count, whether non-causal items are included, whether second-level information is callable. Endpoints: completion rate, time, errors, which items people later recall actually using.

Subjective “richness” often correlates negatively with completion. Do not take it as the success metric.

Where it stops holding

Exploration, inspection, teaching want to see more; completion-orientation will starve them. Their success criterion is coverage, and should be written as a different objective, not the assembly cut. Safety monitoring’s “completion” includes “do not miss”; one more real alarm is preferable to an assembly-style necessary set. Experts and novices have different necessary sets: experts are hurt more by full-inventory noise, novices stop when one causal cue is missing. Shipping the sales-demo full kit as the default is the most common stolen objective.

Applying it

• Write the task’s success criterion first (turned correctly, arrived, seated). Then list the default layer: each item must name the action it changes. If you cannot write that sentence, demote to second level or delete.
• Pick the default from the dose curve, not from “we have this data.” Stop at the bottom of the U; full inventory belongs to debug or an expert switch.
• In review ask “if we drop this, does completion fail?” “No, but it looks more complete” is the delete signal.
• How to check: the same assembly on the necessary set and on the full inventory. If the full set is slower or more error-prone, density is already serving display. If the necessary set stalls on a missing item, promote that one item — do not add the inventory back as a block.

Related

• Same group：N5.12.1 Overlays should show and hide with task relevance, not stay resident · N5.12.2 Information bound to a specific object is easier to associate correctly than information floating in view · N5.12.3 Handling the real environment and overlays at once costs extra attention; they compete for bandwidth · N5.12.4 When the environment is already information-rich, overlays must reduce, not add
• Nearby：N5.07 Overlay Density · N5.13 Safety Risks from Occluding Reality
• Search terms：task-completion density · just enough information · information foraging