Power Law of Practice
Performance commonly improves quickly during early practice and more slowly later, but individual learning is not universally described by one power function.
Direct explanation
What it means
Repeated stable tasks become faster through learning, strategy change, and skill. Averaging people can make a power curve appear even when individual curves are better fit differently.
Interfaces should preserve stable expert paths while avoiding the claim that repetition alone fixes poor design.
Research anchor: Nathan J. Evans, Scott D. Brown, Douglas J. K. Mewhort, Andrew Heathcote, 2018
Research and interpretation
What the research found
Early practice data often showed large initial improvement followed by diminishing gains.
Modern work finds systematic practice speedup but warns that averaging and model choice can make power functions appear more universal than individual data justify.
Popular advice versus careful use
Common shorthand: Every repeated task improves according to one predictable power curve.
Expect diminishing gains as a descriptive pattern, measure individuals over time, and do not use practice to excuse poor design.
Tₙ ≈ a + b·n⁻ᶜ (descriptive candidate, not universal)How designers apply it
- 01
Keep command positions stable for repeated operations.
- 02
Introduce shortcuts after users understand the underlying action.
- 03
Measure novice and experienced performance separately.
Original UI examples
desktop developer tool
Code review
Stable shortcuts and row actions support repeated triage.
Experienced reviewers gain speed without hiding discoverable labels from newcomers.
mobile operations
Warehouse scan
The same confirm sequence remains stable across items and gives immediate errors.
Practice builds fluency while exceptions still receive explicit guidance.
Difficult admin flow
Less effective
Keep an obscure interaction because users will eventually learn it.
Better fit
Fix unnecessary friction, then provide stable accelerators for repeated legitimate complexity.
Learnability is not permission for avoidable confusion.
Limits and failure modes
Curves vary by task, strategy, interruption, transfer, skill, and performance measure.
Common mistakes
- Assuming all users converge on one curve.
- Comparing experts and novices without accounting for strategy.
When it should not dominate
- To defend inaccessible or rare confusing tasks.
- When the workflow or mapping changes continuously.
Responsible use
Accessibility
Do not make repeated practice the accommodation; retain labeled, keyboard, assistive, and error-tolerant paths at every expertise level.
On mobile: Keep gestures and control placement stable across releases.
Ethics and context
Do not externalize product complexity as unpaid user training or use practice metrics for unfair worker surveillance.
On desktop: Support shortcuts, batch actions, and customization without removing discoverability.
Designer checklist
- Separate novice and longitudinal measures.
- Keep mappings stable and feedback clear.
- Test alternative curve fits and individual variation.
Check your understanding
Which statement applies Power Law of Practice most carefully?
Practice and continue learning
See ideas in real interfaces
These external interfaces are editorial references, not examples created to demonstrate this principle.
Sources and further reading
- 01Open source
Refining the Law of Practice
Nathan J. Evans, Scott D. Brown, Douglas J. K. Mewhort, Andrew Heathcote · 2018 · Psychological Review
Peer reviewed review
Continue the graph
Related principles
theory
Flow
Flow is a deeply absorbed experience associated with clear goals, immediate feedback, concentrated attention, and a workable balance of challenge and skill.
phenomenon
Paradox of the Active User
People often begin productive action immediately instead of investing first in broad instruction that might improve later efficiency.
principle
Stimulus-Response Compatibility
Responses are generally faster and less error-prone when the relationship between a signal and its control matches learned or spatial expectations.