Highly Human Software conducts research in human performance modeling, tools to support computational cognitive modeling, and applications of modeling to error-tolerant system design.

Human Performance Modeling

Developing effective and usable software is expensive and time-consuming. Usable software is efficient to use, easy to learn, easy to remember, error-tolerant, and subjectively pleasing. This Usability is often the critical factor that determines the success or failure of a product.

Half of all software projects fail or are canceled, often because of poor usability. Human performance modeling is a proven technique that, when inserted into the build and test cycle of software development, can find many usability problems faster, cheaper, than user testing and more reliably than general design guidelines. The modeling process often indicates not only where users might have problems, but also why and what interface elements to redesign.

A good software engineering process is clearly defined and includes the use of both design guidelines and usability testing. Adding human performance modeling to this process can make the application of guidelines and the design of testing instruments more effective and efficient, resulting in a better end product. Highly Human Software develops engineering models of human performance in a way that fits well within a dynamic software engineering environment.

GOMS

The best-developed engineering technique for gaining deep insights into user needs is GOMS analysis. GOMS is an acronym that stands for Goals, Operators, Methods, and Selection Rules, the components of which are used as the building blocks for a GOMS model.

Highly Human Software utilizes a form of GOMS based on a simple cognitive architecture known as the Model Human Processor (MHP). This representation of human performance consists of separate components for cognitive, motor, and perceptual processors (and associated buffers), as well as for long and short-term memory. We use a formal GOMS language to create human models that are both computational and human-readable. This allows for models that are both efficient to use and easy to understand.

GOMS Analysis

GOMS models are used to represent the procedural knowledge required to perform tasks from the user’s perspective. That is, they represent the “how to do it” information about a task, rather than “how it works. This makes GOMS useful for answering questions, such as “How long does this task take?”, or “Which of two procedures is easier for the user?”

GOMS models can also form the basis for both quantitative and qualitative predictions of how well people can perform tasks using a particular interface, but GOMS analysis does not replace empirical user testing entirely. It can, however, catch many usability flaws and can replace expensive user testing, especially during initial design iterations. This is especially important when the intended user audience consists of highly skilled people whose time is expensive and difficult to obtain. GOMS can also be used before an interface is even built, allowing rapid exploration of the design space.

In effect, the GOMS model acts as a simulated user reacting to events and performing the task just as real users. If the simulated user has trouble or makes mistakes, it is likely that a real user would as well. This allows the analyst to predict various usability properties of an interface design such as consistency and clarity of procedures, cognitive complexity, and potential for human error.