Textbook of Usability

The Power Law of Practice describes how performance on a task improves with repetition. The time to complete the nth trial is given by:

$$T_n = T_1 \cdot n^{-\alpha}$$

where $T_1$ is the time on the first trial, $n$ is the trial number, and $\alpha$ is a learning rate parameter (typically 0.2–0.6). The law predicts the characteristic shape of skill acquisition curves: steep initial improvement that gradually plateaus, continuing to improve indefinitely but with diminishing returns.

The power law has been demonstrated across an extraordinary range of skills: typing, cigar rolling, reading inverted text, menu navigation, text editing, and computer game performance. It is one of the most universal findings in experimental psychology.

Implications for design and evaluation:

• Novice performance is not a good predictor of expert performance; the two differ by orders of magnitude
• Usability tests with first-time users reveal different problems than tests with practised users
• Learning curves can be projected from early trials to estimate expert performance
• Designs that favour novices (e.g., menus over shortcuts) may limit long-term expert productivity
• Keyboard shortcuts become valuable only after practice has made them automatic

The law implies that no amount of practice produces infinite speed — there is always a floor set by the underlying cognitive and motor architecture — but practice continues to improve performance long after gains seem to have stopped.