Faster is not necessarily better without correctness
Aliases: haste versus accuracy · speed accuracy tradeoff · error-aware efficiency
What it is
Shorter completion time is efficiency only when the goal was reached correctly. Skipping a check, submitting a wrong value that the system silently repairs, or stopping in an erroneous state all shorten the clock. The speed–accuracy tradeoff is the option to buy time by sacrificing verification. Chasing “faster” on its own treats that strategy as a design win. Time has to be read with correctness, omissions, and whether damage can still be undone.
Why it happens
Operators set an internal stopping rule: submit when sure enough. If the interface lowers friction, weakens confirmation, or postpones the cost of error, the rule fires earlier—time falls, and errors or invisible mis-postings rise. Laboratory time pressure, incentives, and “please finish quickly” push the rule even earlier. A time gap between designs may then be a difference in caution, not a clearer path. Time optimization without a correctness constraint rewards the people willing to skip.
Studying it
Define correct completion, completion-with-error, and non-completion in advance. Compute primary time only on correct completions, and report error rate and severity alongside. If the tradeoff itself is the question, manipulate speed pressure in instructions or deadlines and plot time jointly with errors, rather than crowning two separate metrics. Check whether fast successes cluster among people who skipped verification; code “confirmation page never opened” as a process measure when needed.
Where it stops holding
For skilled users on low-risk, reversible actions, speed is the goal and extra confirmation is a tax. Irreversible transfers, deletions, and dosages are the opposite: slightly slower and correct beats faster-and-wrong. Gamified or contest-like tests inflate the tradeoff and do not transfer to unpressured voluntary use. Autocomplete and defaults shorten time while possibly bypassing inspection; “correct” then depends on whether responsibility still sits with the user.
Applying it
- Never write a time reduction as an improvement when correctness fell; show the two together.
- Put a guardrail on irreversible actions: the rate of skipped critical steps must not rise as time is optimized.
- Instruct participants to work at their usual care unless the study is about time pressure; avoid “as fast as you can.”
- Review the fastest quartile and confirm the clock advantage did not come from mis-clicks, skipped steps, or tacit moderator permission.