This population has a higher mis-tap rate, and undo beats confirmation

What it is

Even with filtering and target enlargement layered on top, people with pathological tremor still mis-tap more often than the general population — compensation reduces the rate, it doesn't eliminate it. Given this residual elevated mis-tap rate, relying on an easily reachable undo to reverse an action after the fact works better for this population than inserting a "confirm" step before the action that decides whether to cross that line.

Why it happens

The core problem is that a confirmation dialog is itself another precisely-targeted pointing action — it places a second target, demanding the same precise landing, on the same noisy motor channel that just produced the original mis-tap. The mechanism behind the error hasn't been bypassed at all; it has simply been given one more chance to fail in the same way. Undo runs through a different channel entirely: judging "was this outcome correct" relies on visual recognition and cognitive judgment, capacities that are usually unaffected by damage to the motor system, and once that judgment is made, all that's needed is to trigger an undo control that can be made large and highly forgiving — no repeat of a high-precision action is required at the critical moment. In other words, confirmation puts the gate exactly where errors are most likely to occur; undo moves the gate to a place where they aren't.

Where it stops holding

The undo path only works when the action is actually reversible and the undo window covers the time it realistically takes the user to notice the error. For a message that can't be recalled once sent, a payment already executed, or other irreversible or high-cost actions, undo provides no protection — these cases still need a gate before the action, but that gate itself should also be implemented at low precision (a voice confirmation, a near-full-screen confirm button, a dwell confirmation with generous tolerance) rather than an ordinary-sized confirm button, or the same problem simply resurfaces at the confirmation step.

Applying it

• For reversible actions, prioritize an always-visible, large-tolerance undo affordance over inserting an ordinary-sized confirmation button before the action.
• For irreversible or high-cost actions where a pre-action gate must remain, resize and redesign that gate's trigger specifically around this population's mis-tap rate — don't simply reuse a confirmation button sized for the general user.
• To verify: have target users complete the same set of tasks under real or high-fidelity simulated tremor conditions using both a "confirm then execute" flow and an "execute then undo" flow, and compare the number of resulting irreversible errors and subjective frustration across the two flows — not just the success rate of a single action.

Related

• Same group: A8.16.1 Intention tremor grows stronger as the hand nears the target · A8.16.2 Pathological tremor amplitude far exceeds ordinary tolerance design · A8.16.3 Compensation requires both input filtering and target enlargement
• Nearby: A10.08 Error tolerance and graceful degradation · J3.09 Mis-tap prevention and undoability
• Search terms: undo versus confirmation · error recovery · motor-impaired input · destructive action