Distance selects which input is available

What it is

Ten metres, two metres, or half a metre from a public screen, the channels a body can offer the system are not the same. Distance selects input. Out of reach of the glass, what remains is presence, heading, coarse whole-body motion. Pointing and gesture become plausible at mid range. Reliable touch waits for arm’s length. Vogel and Balakrishnan wrote public ambient displays as a continuum from implicit to explicit, public to personal, with distance as the main axis. This entry is only which input is physically and socially live at a given distance. It is not how to stop the switch from chattering, and not how dense the type should be.

Why it happens

Touch has a hard geometric ceiling: a finger must enter contact range of the surface. Mid-air gesture and skeleton tracking have another: too far and the person is a few noisy pixels; too near and the camera loses the body and pointing angles go extreme. Far-field speech is pinched by room noise and by speaking in public at once. Each channel is therefore machine-readable and socially doable only inside a band.

Hall’s proxemic distances add the social layer. A fine gesture at public distance is theatre at empty air; a full-body flourish at personal distance reads as intrusion. An input scheme that demands touch from ten metres cannot be reached; one that still demands an over-shoulder gesture at face-on range feels redundant and in the way. Distance is not a visual “make the letters bigger” problem. It is the existence condition of a channel. People are also moving, so available channels are a menu that opens and closes with position, not a mode chosen once at boot.

Studying it

Vogel and Balakrishnan’s zoned prototypes, and Greenberg and colleagues’ proxemic interaction, take position, orientation, and motion as the factors that select input. In the field, overlay a histogram of passer-by distance on the channel actually used: how many people enter touch range, how many only dwell at mid range.

Independent variables: distance and heading to the screen, the set of open channels (touch / gesture / presence-as-input / phone). Dependent variables: channel actually used in each distance band, wrong-channel attempts (poking air from far away, still waving large at close range), abandonment.

A lab prompt to “walk to this tape and then gesture” turns distance into the task and measures compliance, not natural choice. Sensor range belongs in the method: if a depth camera drops skeletons beyond six metres, “far gesture failed” is equipment range, not preference.

Where it stops holding

Seated far-field (a living-room television, a classroom projector) barely changes distance, so a channel can be chosen once and need not follow walking. A timetable that declares itself non-interactive lets distance affect reading only; it does not open an input menu. When a phone mediates, someone can drive the wall from far away and the distance to the glass no longer gates the channel—pairing does. When a hand is occupied or an accessible alternative is required, voice or a switch may still be needed inside touch range; the default distance-to-channel map must be breakable.

Applying it

• Draw a channel table on the standing bands of the real site: passing-by opens only implicit input (being seen, being approached); mid range opens pointing; arm’s length opens touch. Do not make a band depend solely on a channel that band cannot perform.
• In each band, let the cheapest channel of that band be the first acknowledged input; open deeper channels only after the person has entered the matching distance.
• Put sensor range on the same table. Beyond the distance where the camera drops a skeleton, do not pretend gesture is still live.
• Verify by marking distance lines on site and watching what natural passers-by do in each band, and what the system accepted. A band full of “done but no channel” or “system waiting for an input nobody performed” means distance and channel are misaligned.

Related

• Within the group: K7.05.2 Switching thresholds need hysteresis · K7.05.3 Information density must follow viewing distance
• Adjacent: K7.02 Discoverability of Interactivity · K5.01 Focus-based navigation · X1.04 Human–robot proxemics
• Search terms: proxemic interaction · implicit interaction · distance-dependent input · Vogel Balakrishnan