Motion can occupy a loading gap

What it is

Network or decode often leaves a gap after the tap and before the next interface can paint. If that gap is occupied by a transition that already had a reason to exist, people do not also sit through a blank stall. The motion is not there to make two seconds feel like one; it hides a round-trip inside a path. Duration perception may shorten as a side effect; this claim holds when the gap is real and the transition is real.

Why it happens

Latency from tap to first interactive frame, left naked, is labelled sluggishness. A transition rewrites that interval as “going to the next place”: a path is in progress, and the destination can decode and lay out offstage. When the transition ending and content readiness fall near the same moment, the causal story is complete and there is no ownerless blank frame in the middle.

The transition needs its own spatial or identity reason. A looping spinner with no such reason is not a cover; it only admits the gap is still there. A skeleton is a different filler: it admits “the page arrived, the content has not.” Transition-cover is “the page is still on the way.” Do not use one motion for both kinds of gap.

Studying it

Hold real latency fixed and compare blank-then-cut against transition-in-parallel-with-load on sluggishness ratings, abandonment, and whether people say “I was waiting on the network” or “I was entering the next page.” Independent variables: latency relative to transition duration, whether the transition has spatial meaning. Dependent variables: sluggishness, attribution (network / navigation / freeze), abandonment.

Report physical latency separately from transition duration, or you cannot tell a successful cover from a wait that merely changed names.

Where it stops holding

• If the gap ends before the transition can do any work, there is nothing to cover and the transition only holds the user.
• If the gap is far longer than any trackable path, the transition will finish and still expose loading; pick up progress at the end state rather than stretching the transition to force a cover.
• Vestibular-sensitive users would rather see a short blank than a large displacement covering a few hundred milliseconds.

Applying it

• Overlap predictable short round-trips (cache miss, image decode) with a page transition that was going to run anyway.
• Pre-render the destination during the transition so it is interactive at the end, rather than starting the request after the clip.
• Check: on a slow network, inspect the timeline between tap and interactive. The cover worked if blank frames disappear without the interval growing.

Related

• Same group: F7.06.2 The cover must not outlast the real delay · F7.06.3 Padding delay with motion is a net loss
• Nearby: F7.01.3 Motion fills a wait and changes how long it feels · I2.07 Perceived performance · I1.01 Instant-causality threshold
• Search terms: latency masking · loading gap · perceived performance