The Psychology of Human-Computer Interaction
Source: https://archive.org/details/psychologyofhuma0000card ↗
This book founded human-computer interaction as a quantitative science.
Card, Moran, and Newell — working at Xerox PARC and Carnegie Mellon — introduced the GOMS model and applied Fitts's law to predict how long real users take to accomplish real tasks with real interfaces.
Before this work, interface design was opinion; after it, there were testable models.
The Keystroke-Level Model alone gave designers a way to compare alternatives without building prototypes.
Nearly every serious HCI curriculum still starts here, and the engineering psychology tradition it established runs through everything from touch-target sizing on mobile to modern A/B testing methodology.
If you design or evaluate interfaces and have not read this, you are working without knowing the foundations of your own discipline.
Central argument
Card, Moran, and Newell argue that human interaction with computers can be modeled with the same rigor applied to physical and cognitive engineering systems. Their central contribution is the GOMS model — Goals, Operators, Methods, and Selection rules — which decomposes user tasks into measurable cognitive and motor units, and the derived Keystroke-Level Model, which predicts task completion times from first principles without requiring user testing. The core thesis is that interface design is not a matter of taste or intuition but of quantifiable human performance, and that designers can therefore compare interface alternatives analytically before a single prototype is built.
Critique
The framework's predictive power depends on expert, error-free users performing discrete, well-defined tasks — a population and a context that describes a shrinking share of actual product usage. Modern digital products are defined less by procedural efficiency than by exploratory, emotionally driven, and socially embedded behaviors that GOMS has no vocabulary for. A CPO relying on this tradition alone risks optimizing the measurable at the expense of the meaningful, confusing task speed with product value.
Why it matters for product
The Keystroke-Level Model offers product leaders a concrete counter to the instinct to prototype everything — when the task is sufficiently defined, analytical modeling can resolve design debates in hours rather than sprint cycles, which directly affects discovery velocity and where teams spend their build capacity. More broadly, the engineering psychology tradition this book established is the intellectual ancestor of the quantitative UX metrics — time-on-task, error rates, touch-target sizing — that modern product teams use to write acceptance criteria and interpret A/B test results; understanding the assumptions baked into those methods tells you exactly when they can be trusted and when they mislead.