Arbitration rules must be predictable to users, such as first-come or nearest-person

What it is

Even after the system knows “this hand is A's, that hand is B's,” two people who both intend to operate still require a decision about who has control. That decision is arbitration, not identification. The rule has to be guessable in advance: first-come, nearest-person, whoever faces the screen, or an explicit hand-raise claim. If people cannot predict who will be heard next, they either race the gesture or stop using the system.

Why it happens

Arbitration sits downstream of attribution: two already-legal input streams contend for one output channel. First-come depends on the timestamp of entering engagement, so clock jitter and delayed recognition of the enter motion matter—someone whose enter pose is recognized late feels they raised first and still lost. Nearest-person depends on distance from the sensor to torso or hand; anyone furniture pushes toward the screen will win steadily, and anyone at a comfortable viewing distance will lose steadily. Facing direction, gaze estimates, and spoken claims each buy a different kind of predictability. A simple internal rule is not enough; the loser still needs to understand “why not me.” Unpredictable arbitration is experienced as capriciousness even when every technical decision matches some heuristic.

Studying it

Have two people reach for the same control at once. Compare success rate, wait time, spoken contention, and “I thought it was my turn” reports under different published rules. Contrast a one-sentence on-screen rule with no explanation. After the rule is learned, measure whether losers step back instead of repeating the gesture harder. Recognition delay on the enter motion will distort the felt fairness of first-come; log both sensor timestamps and participants' subjective order.

Where it stops holding

True co-editing (two people dragging different corners of one map) needs composition, not mutex arbitration; first-come turns collaboration into turn-taking. When one person is clearly watching and the other operating, nearest-person can steal control for whoever leans in to see. Children and adults have different reach, so a nearest rule systematically favors the body that can stand closer. Rules are not culturally neutral either: in some settings “raise a hand” is more acceptable than “stand nearer.”

Applying it

• Pick a rule that fits in one sentence, and while several people are visible keep showing who currently has control (name, color, or stance mark) rather than flashing only at the moment of conflict.
• Give the loser a predictable next step: wait, claim explicitly, or move to a marked stance. Do not leave “do the gesture harder” as the only strategy.
• Accept the design with a two-person simultaneous-reach task in the real venue. Ask the loser to restate the rule; a failed restatement means the rule is still unpredictable, independent of recognition rate.

Related

• Same group: C4.17.1 When several people share the sensing volume, the system must decide who owns the current action · C4.17.2 Attribution depends on identity features, not on the motion itself · C4.17.3 Misattribution treats one person's motion as another person's command
• Adjacent: C4.15 Conditions for entering engagement · C1.24 Multi-cursor and multi-user pointing
• Search: arbitration · first-come first-served · nearest-person