Final landing accuracy is set by the closed-loop phase

What it is

How close a pointing movement's final landing spot ends up to the target is mainly determined not by the quick initial approach at the start of the movement, but by how many rounds of closed-loop correction actually get completed afterward: the initial approach is only responsible for shrinking the gap to a rough range, while the repeated compare-correct cycles that follow are what actually converge the landing point onto the target. This means the answer to "how accurate was this movement" is mostly hidden in that small final stretch, not in how fast or how straight the movement looked in flight.

Why it happens

Each completed round of closed-loop correction, in principle, reduces the current error by some proportion; more rounds compress the remaining error further. How many rounds a given movement can fit in total depends on total execution duration divided by the minimum time one round of correction requires — cut the time short and few rounds fit, so the error may still be fairly coarse when the movement stops; allow more time and several extra rounds can run, pushing the error lower. In other words, final accuracy is essentially the product of "how many rounds of correction the time budget can buy," not simply a function of how skilled the movement is.

Studying it

A common way to verify this controls the total time participants are allowed for the same pointing task and compares landing error across duration conditions: if longer allowed duration systematically produces smaller landing error, and the rate of that decline roughly matches what a single closed-loop correction round should achieve, this supports the explanation that accuracy is set by the later closed-loop phase. Another approach segments the trajectory — splitting the movement into an early fast-approach portion and a later fine-adjustment portion — and measures each segment's separate contribution to the variance of the final error; the later segment, despite covering only a small share of total distance, is typically found to contribute the largest share of that variance.

Where it stops holding

This holds on the premise that the allowed time is enough to fit at least one round of correction; if the task duration is so short that not even one round fits, the movement falls back entirely to open-loop, and accuracy is then set by the variability of the parameters fixed before launch, not by the closed-loop phase. Also, "more rounds means higher accuracy" doesn't hold without limit — once error has already dropped below some threshold, the marginal improvement from additional rounds falls off quickly, and what limits accuracy from that point on may shift to the device's own positioning resolution or the smallest controllable increment of body output, rather than an insufficient number of correction rounds.

Applying it

• For tasks requiring high landing precision (fine alignment, small-target selection), if the time budget allows, don't rush to judge a hit while the movement is still in its fast-approach phase — leave enough time for the tail-end correction phase so users can complete a few more small adjustments, rather than applying one uniform short timeout across every phase.
• If a fine-precision operation's completion time genuinely must be kept short, rather than squeezing what little correction-time budget remains even further, enlarge the target or provide snapping assistance so the gap that needs to be closed by correction is smaller to begin with, reducing dependence on the number of correction rounds.
• How to check: test the same users' pointing task across several tiers of allowed duration and plot landing error against allowed duration. If error keeps dropping with more time and then levels off, that leveling-off point is roughly the duration at which correction rounds stop being the bottleneck for the current device and target size, and it can be used as a reference for setting timeouts or feedback cutoffs.

Related

• Same group: A8.07.1 Closed loop relies on continuous feedback to close in on the target step by step · A8.07.2 Each correction needs at least one full feedback-loop cycle · A8.07.3 Feedback delay beyond the loop cycle causes overshoot and oscillation · A8.07.5 Closed loop degrades to open loop when the feedback channel is lost
• Nearby: A8.08 Two-Phase Structure of Target Acquisition · A8.09 Speed-Accuracy Tradeoff
• Search terms: endpoint accuracy · closed-loop precision · movement time budget