Whether re-entry requires repeating the enter action depends on whether the grace has expired

What it is

When the hand reappears, the system has a binary choice: treat the appearance as continuation of the same engagement, or as a fresh entry. The dividing line is whether grace has expired. Re-enter inside the window and the enter action is waived; the user keeps working. Re-enter outside the window and the enter action must be performed again before engagement resumes. The claim is how expiry decides “must they enter again,” not what the enter action should look like.

Why it happens

Engagement is a bistable with hysteresis. The grace timer is the width of that hysteresis: before it expires, state stays at “engaged” and a returning hand only reattaches tracking; after it expires, state has already fallen to “not engaged,” and a returning hand is protocol-equivalent to a passer-by’s hand—it must pass the enter condition. Users rarely perceive the timer. They only notice that sometimes dropping and raising the hand still continues an adjustment, and sometimes they must first do the “start” pose. If expiry still waives the enter action, grace is fiction and any late hand can resume the last session. If the enter action is still required before expiry, grace is also fiction and a brief occlusion is treated like a true leave.

Studying it

Build two occlusion durations that straddle the grace boundary: shorter than the window and longer. Dependent measures: whether an enter prompt appears after return, whether an enter pose is required, and how accurately users predict “do I have to activate this time.” Vary visualization of the window (none / countdown / fading engagement halo) and see whether predictability rises. Do not mix discoverability of the enter action into the dependent measures; that is a different topic. Ask only: for users who already know the enter action, does the system waive when it should and require when it should.

Where it stops holding

If the enter action is almost free (a short dwell already in view), asking for it again after expiry barely interrupts. If the enter action is heavy (a compound pose, a spoken passphrase plus a gesture), a forced redo after expiry is felt as punishment, and people learn not to drop their hands. If an idle timeout runs in parallel with grace, the relationship between the two timers must be specified: whichever arrives first counts as leave. A system with no enter action, listening in a Midas style throughout, has no “redo enter” to talk about; expiry then only means the command channel closes.

Applying it

• Write a state table: re-entry inside the window waives the enter action; re-entry outside requires it. Treat the table as an implementation constraint, not folklore.
• Near expiry, give a perceptible fade (halo dimming, a short tone) so the user can decide to keep the hand in view and continue, or accept the coming exit.
• Test “occlude 200 ms then return” against “drop for three seconds then return”: the first must not show an enter prompt, the second must. Getting both wrong means the state machine is not branching on expiry.

Related

• Same group: C4.19.1 A re-entry grace period can keep engagement after brief tracking loss · C4.19.2 An overly long grace keeps the system live after the user has abandoned the action · C4.19.3 An overly short grace treats brief occlusion as deliberate disengagement
• Adjacent: C4.15 Conditions for entering engagement · C4.16 Disengagement and timeout
• Search: grace expiry · re-engagement · enter action