Input needs a visible owner

What it is

When two hands land on a public screen and the system files them as one pointer’s stream of events, nobody present can answer “whose was that.” Input attribution tags each contact, gesture, or utterance to a recognizable actor, and makes that ownership visible to the actor and to everyone else. It comes before how conflicts are judged: until inputs have owners, there is no fact of two people grabbing the same thing. It is also not about whether names and faces should appear on the glass—that is a different question.

Why it happens

Single-user UI treats “the current pointer is me” as default. A public large screen deletes that default: several hands, several bodies, sometimes a phone casting from a distance. If the event stream has no identity, feedback can only be drawn “somewhere on the screen,” never “on your hit.” People guess from time and space proximity. Wrong guesses become wrong learning: they think they dragged what the neighbouring hand dragged, change force next time, and the system looks more insane.

Attribution is also social information for bystanders. Tabletop territoriality work shows that people use bodies and objects to mark “this is my zone.” If every cursor is one colour and every selection one highlight, territory cannot be read, and a newcomer will put a hand into someone else’s ongoing act. Failed attribution is then both a perceptual miss and a social collision, even when the two people wanted compatible things.

Studying it

CSCW observation of tabletops and public walls (CityWall, DiamondTouch and kin) codes “whose hand,” not only contact count. Experiments can compare unidentified multitouch, per-contact coloured cursors, and input zoned by standing position. Peltonen and colleagues’ CityWall analysis is especially about how people coordinate “this is mine” under parallel use.

Independent variables: whether contacts or cursors carry a stable identity, whether the cue is colour, position, or a mark near the body, whether input arrives from a personal device. Dependent variables: times another person’s input is taken as one’s own, interruptions during parallel use, spoken “yours or mine.”

In the lab the pair has been introduced and shares a task, so politeness damps the need for attribution. Strangers in the field are the condition where attribution is actually required. Logs that store only an anonymous contact stream cannot reconstruct attribution failure later; actor identifiers (even “left/right stance”) have to be kept from capture time.

Where it stops holding

A strictly queued, one-at-a-time kiosk does not need parallel attribution; it needs a session identity for “whose turn,” which is exclusive in time. A screen driven by one remote control collapses attribution to focus ownership. When two people truly share one object (pushing a map together), an aggressive “this is your cursor” splits the common object. Children often ignore colour codes and grab by feel; colour attribution is weak for them.

Applying it

• Give every concurrent input a stable, distinguishable feedback: differently coloured contact glows, a trail stuck to that hand, or a highlight zoned by where the person stands—and do not swap identity mid-gesture.
• Make ownership readable to others present, not only to the operator. A newcomer should be able to see which patch is already taken.
• Input arriving from a personal phone still needs a mark on the public screen for “this stream is from that side / that person,” so the wall is not a sourceless animation.
• Verify by asking two people who did not rehearse together to operate at once, then to point at which changes they caused. Every mismatch with the log is ownership that was not visible.

Related

• Within the group: K7.04.2 Concurrent use needs an arbitration rule · K7.04.3 Personal information does not belong on a shared screen
• Adjacent: C1.24 Multiple cursors and multi-user pointing · V2.05 Shared cursors and selections · V3.05 Transfer and arbitration of control
• Search terms: input attribution · multi-user display · territoriality · CityWall