At the Interface

Illustration by Heather Miller

In the late 1980s, computer engineer Cherri Pancake made a discovery that startled her: Despite the millions of dollars invested in computer hardware and the explosive growth in software, no published research focused on how people actually use these devices.

The issue came up when Sue Utter, Pancake’s master’s student at Auburn University, wanted to develop software for the supercomputing industry. “My first question was: Who uses these computers?” says Pancake, now a professor in the School of Electrical Engineering and Computer Science at Oregon State University. “What problems do they work on and how do they think about these problems?” When a search of the published literature found no studies on those issues, she saw an opportunity to use skills she had learned in her past as a social scientist. “I couldn’t help it. The anthropologist came out,” she adds.

In fact, as a graduate engineering student, Pancake had avoided her more than six years of experience as an ethnographer in the highlands of Guatemala. The engineering culture, she found, was not terribly open to someone in the social sciences.

Cherri Pancake
In OSU’s Kelley Engineering Building, Cherri Pancake leads efforts to adapt software to the preferences of users. (Photo: Hannah O’Leary)

Pancake got her bachelor’s in environmental design at Cornell and took graduate courses in anthropology at Louisiana State. A stint in the Peace Corps took her to Peru. Later, as curator of the Ixchel Ethnological Museum in Guatemala City, she collected stories in isolated Mayan communities that had developed their own customs and even distinct languages. Human behavior intrigued her.

In communicating with her Indian field crew, Pancake observed a skill that would prove to be useful in her engineering career: When conversing in a second language, people tend to take pains to be clear. They choose words carefully and provide context. The same deliberate approach to communication, Pancake has learned, can benefit scientists and engineers.

Cross Cultural

Since then, engineering has changed dramatically. As director of the Northwest Alliance for Computational Science and Engineering (NACSE, pronounced “nacks”) at Oregon State University, Pancake combines knowledge of human behavior with expertise in computational systems and software. She and her colleagues work with universities, industry, government agencies and nonprofits to take knowledge from the confines of technical specialties — climatology, hydrogeology, seismic motion in the Earth’s crust, wildfire science — and adapt it for use in other fields. Most of their clients are natural resources managers who use models to make decisions based on scientific input. Their responsibilities range widely from managing wildfire to preparing for tsunamis and reducing fraud in crop insurance programs.

The term for what Pancake and her colleagues do is “usability engineering.”

“Anyone who uses a computer or another device these days knows when things are not usable,” Pancake says. “The idea of usability engineering is: What can we do to large complex software and data systems to adapt them to how users think rather than forcing users to think like the software?”

When Pancake began studying how programmers and software users approach their work, she faced skepticism. She was among the first to ask them questions about why they did things a certain way and whether they had ever tried other methods and failed. “I got them to start recording for me when they reached a dead end,” she says, “when they thought something was going to work but proved not to. And I would have them save scrap paper that they used to jot down notes while they were working. I gave them a box and asked them just to throw it there instead of the trashcan. They thought I was nuts.”

At the interface between computers and people, Pancake also applied lessons from her physical anthropology past. “We are humans. We are in these bodies, and they are a blessing and a curse,” she says. “If you get down to it, how we think and even the errors we make have to do with our physical limitations, with how we learn things, how much attention span we have, how many things we can remember at a given time, what goes wrong when we’re trying to form logical models of things.”

Uniformly, the users of these machines hated the tools and the languages that were being provided to them. And uniformly, the industry people thought they were doing a great job of providing tools.

The supercomputing industry (including Intel, HP and IBM) funded her research. “They were trying to understand how to create better tools by understanding why people found their machines so difficult to use. Uniformly, the users of these machines hated the tools and the languages that were being provided to them. And uniformly, the industry people thought they were doing a great job of providing tools.”

With a staff of 15 full- and part-time researchers, NACSE operates out of the Kelley Engineering Building on the OSU campus. Pancake and her colleagues, including Chris Daily, chief scientist, and Dylan Keon, GIS coordinator, have brought together an interdisciplinary team. All have expertise in two or more specialties, from computational geography to crop and soil science.

“I’ve found that if you’ve trained in a science, it’s relatively easy to learn the techniques of a new discipline,” says Pancake, “but what’s very hard is to communicate with people in a different discipline. To me the great pleasure is in being able to combine all of those threads to solve problems in a different way.”