Why the opposite direction
The explanation given is one of relative motion. The edge of the bar is what actually reports progress, and it moves to the right. When the pattern inside is travelling left, the difference between the two is larger than the edge's own movement, so each increment of real progress reads as a bigger step than it is. Run the pattern the same way as the bar and that exaggeration disappears.
What the thread offered as evidence
- A cited study reporting that animation can increase the perceived speed of a download by up to about eleven per cent over a bar with no animation.
- The same source is credited with a second technique: making the interior pattern pulse faster as the bar approaches completion.
- One contributor made a side-by-side animation of the two directions and invited readers to cover one and watch the other, which is a fair way to check a perceptual claim.
- The figure belongs to the study the thread cites rather than to the thread itself, and nobody in it reproduced the result independently.
The two jobs a loading bar is doing
One answer separated them usefully. The widening filled area reports how much of the work is done. The moving pattern inside reports that work is happening at all — which is the more urgent question when a bar has not visibly advanced for several seconds. They are different signals, which is why an indeterminate spinner still helps even though it conveys no progress whatsoever.
Where the technique stops helping
- It changes how long a wait feels, not how long it is. Applied to a genuinely slow operation, it buys seconds rather than minutes.
- A bar that animates while nothing is happening is worse than an honest one, because the motion is a claim that work is ongoing.
- Motion that continues indefinitely stops reassuring and starts reading as a stall, which is the point at which an estimate or a description of the current step does more than any animation.