March 17, 2022
Jef Raskin did not want interface design to be a craft. He wanted it to be a science. The word he used for that science was cognetics, and he set it out near the beginning of The Humane Interface: New Directions for Designing Interactive Systems (Addison-Wesley, 2000). Twenty-two years after the book appeared, the term has not entered general use, and the discipline Raskin described still has not fully formed. That gap is worth examining, because the reasons cognetics did not take hold say something about how the software industry actually treats its users.
What Raskin Meant by Cognetics
Raskin defined cognetics by analogy to ergonomics. Ergonomics is the applied study of how physical tools and environments fit the human body — the height of a chair, the reach to a control, the force needed to press a pedal. Cognetics, in Raskin’s framing, was the applied study of how interfaces fit the human mind: attention, memory, habit formation, the limits of conscious processing.
The analogy carried a deliberate implication. Ergonomics is not a matter of opinion. You do not debate whether a control should require forty pounds of force to operate; you measure the population, you find the distribution of strength, and you design within it. Raskin argued that interface decisions could be approached the same way. The properties of human cognition that matter for interface design — how long it takes to make a decision, how many things a person can attend to at once, how readily an action becomes automatic — are measurable, stable, and largely outside the designer’s control. Design should be fitted to them, not the other way around.
This was a pointed claim in 2000, and it remains one. It says that a great deal of what passes for interface design is unscientific in the strict sense: it is guided by taste, by precedent, by what looks impressive in a demonstration, rather than by measured facts about the people who will use it.
The Distinction From “Cognitive Psychology”
Raskin was careful to distinguish cognetics from cognitive psychology itself. Cognitive psychology is the basic science — it studies the mind to understand it. Cognetics, like ergonomics, is the applied discipline that takes the findings of the basic science and uses them to design things. A cognitive psychologist asks how short-term memory works. A practitioner of cognetics asks how many items an interface may reasonably ask a user to hold in mind at once, and designs accordingly.
The distinction mattered to Raskin because he had watched the basic science accumulate for decades without reliably reaching the people building software. The research existed. George Miller’s 1956 work on the limits of immediate memory was nearly half a century old when The Humane Interface was published. The findings were not secret. They simply were not being applied with any discipline by the people designing the interfaces that millions of people used every day.
Why the Term Did Not Catch On
A reasonable question, two decades on, is why “cognetics” did not become a working part of the field’s vocabulary the way “usability” or “user experience” did. Several reasons suggest themselves.
The first is that the industry preferred a softer framing. “User experience” — the term that won — is capacious and forgiving. It accommodates measurement, but it does not demand it. A team can claim to be doing UX work while relying entirely on intuition and opinion. Cognetics, as Raskin defined it, did not allow that. It insisted that interface claims be grounded in measurable properties of human cognition. That is a higher bar, and higher bars are adopted slowly.
The second is that Raskin’s own discipline was demanding in a way that ran against commercial incentives. If you take cognetics seriously, you cannot ship a feature because it tested well in a focus group’s expressed preferences; you have to ask whether it reduces the user’s actual error rate and time-to-completion, measured. That is expensive, and the results sometimes contradict what the people paying for the software want to hear.
The third reason is simply that Raskin died in 2005, five years after the book, and the term had no equally forceful champion afterward. Vocabulary in a field tends to spread through the persistence of its advocates. Cognetics lost its advocate early.
What the Discipline Would Have Required
It is worth being concrete about what a mature cognetics would have demanded, because the demands are where the resistance lay.
It would have required interface decisions to be stated as testable predictions: this layout will produce fewer selection errors than that one; this command structure will become automatic after fewer repetitions. It would have required those predictions to be checked against measurement rather than against the preferences of stakeholders. And it would have required the willingness to abandon a cherished design when the measurement contradicted it.
Raskin’s own work modeled this. The Humane Interface does not merely assert that modes cause errors; it explains the cognitive mechanism by which they do, and ties the claim to observable, repeatable error patterns. It treats the quantification of interface efficiency — using models drawn from the human-factors literature — as a normal part of the design process rather than an academic afterthought. The book practices the science it names.
The Partial Inheritance
Cognetics as a named discipline did not arrive. But pieces of Raskin’s program did survive, scattered across the field under other names.
The human-factors tradition that Raskin drew on — the work of Card, Moran, and Newell on the modeling of human performance, and the older results from experimental psychology — continued in academic human-computer interaction and in specialized practice. Quantitative usability testing, when it is done rigorously, is cognetics in spirit if not in name. The accessibility field, which by necessity must reason about measurable human capacities, carries some of the same discipline.
What did not arrive was the general expectation. Raskin wanted cognetics to be the default frame for interface work — the assumption that, of course, you ground design in measured properties of the mind, the way an engineer of course grounds a bridge in measured properties of steel. That expectation never became general. Interface design in 2022 remains, for the most part, a craft informed by science rather than a science applied through craft. Raskin wanted the order reversed.
Why It Still Matters
The argument for cognetics is not a historical curiosity. The cognitive facts Raskin built on have not changed — human short-term memory has the same limits it had in 2000, attention is no less finite, and habituation works the same way. Interfaces, meanwhile, have grown more numerous and more demanding of attention. If anything, the case for a disciplined, measurement-grounded approach to interface design is stronger now than when Raskin made it.
The term may never spread. The discipline it named is still worth building. Raskin’s contribution was to insist that the science was available, that the application was possible, and that the failure to apply it was a choice — one that the field has continued to make, and could choose differently.
Related: