Newsroom

CORVALLIS, Ore. - Almost 30 years ago, before the dawn of the personal computer revolution, Cherri Pancake was working in the mountains of Peru on a Peace Corps assignment. She was struggling to communicate agricultural concepts and build working relationships with area farmers who spoke only Quechua, the ancient language of the Incas.

Pancake didn't speak Quechua very well. Most of the area residents were illiterate. For a culture that lacked modern farm technology, agricultural innovations had to be created on the fly. And to get her messages across, Pancake eventually ended up drawing comic books with her limited artistic skills - it wasn't a textbook style of communication, but it worked and the locals loved it.

Pancake had no idea at the time that some of the "ethnographic" communication skills she had studied and was now perfecting with indigent farmers in Peru might later pave the way for a new generation of user-friendly computer software programs - programs that don't leave people around the world muttering in frustration and disgust.

"When I first got into computer science, I almost used to hide my background in ethnographic research," said Pancake, who is now a professor of computer science at Oregon State University and a leader in ways to make computer hardware and software easier to use and more understandable.

"At the time, I was afraid if people knew I used to work in physical and cultural anthropology, they wouldn't take me seriously as a computer scientist," she said. "It was only later that I found out how useful and relevant those skills were to some of the problems we have in using modern computers."

The core of the problem, Pancake said, is that the people who build computers and the people who use them speak different languages and have different perspectives. This makes some software about as easy to use as an English textbook would be for a Peruvian farmer who only spoke Quechua.

"Take the familiar example of a personal computer that has frozen up and simply won't respond," Pancake said. "To restart the computer, the user is supposed to simultaneously press the buttons 'Ctrl,' 'Alt,' and 'Delete.' To the user, this is ridiculous and is another example of how hard computers are to learn.

"But to the software designer who created that mechanism, it was a logical evolution from earlier mechanisms and even has the usability 'advantage' that it is impossible to press those keys accidentally," Pancake said.

Bridging that gap is not a trivial problem, Pancake said. Computer hardware and software design is now a multi-billion dollar industry, but millions of casual computer users and even many professionals repeatedly voice frustration and irritation at the way programs work and their difficulty of use.

It's not that computer and software designers want to create clumsy programs that alienate customers or defy understanding, Pancake said. They are just unaware of how much can be done to improve program usability, and have never been professionally trained in those skills.

"Our computer science programs at OSU are among only a few in the nation that offer courses in usability engineering," Pancake said. "These approaches are fairly rare, partly because so few usability experts end up at universities. Most of them go into industry."

In addition to her work at OSU, Pancake also directs the Northwest Alliance for Computational Science and Engineering, which is a multidisciplinary research institute housed at OSU that investigates how to make software and information systems more useful to scientists and engineers. Increasingly, Pancake said, she is turning to "ethnographic methods" of learning to surmount some of these obstacles, instead of the scientific or experimental methods that are second nature to most university professionals. The same approaches that worked to help train, communicate with and learn from Peruvian farmers are now being used to train computer scientists.

"The ethnographic approach is almost the opposite of the traditional scientific method, where scientists form a hypothesis first, then create experiments to test it," Pancake said. "Ethnographers specifically avoid forming any hypothesis. Instead, they function as impartial observers, eliciting information from human subjects through questions that are designed to be open-ended and unbiased."

Pancake said her first in-depth experiences using ethnographic methods occurred when she was in Peru, studying how hand-woven textiles had been produced early in the century, with skills most of the local people had long since forgotten.

Eliciting what the local people knew about the textiles, what styles were made in what families, how colors were used, and other details was a complex investigation, Pancake said, in which people were encouraged to speak, interact freely and reflect on their past while she tried to identify patterns, recurring themes and tidbits of information people were barely aware they knew.

The successful exploration of historic textile production in Peru actually has a lot in common with understanding how people use computer software and why errors occur, Pancake said.

"With computers, I'm trying to learn why a person doesn't like a particular program, what mistakes are made over and over, what are people's experiences and frustrations, what steps they are following that lead to errors," Pancake said. "I'll sit with people and watch what they do every step of the way, ask them what they are thinking when they try certain things, and what they think will happen next."

The people who build these systems, she said, often have very understanding of the experiences and frustrations of the people who use them. And the situation is becoming worse as computer technology becomes more powerful and allows ever more complex programs to perform a wider range of tasks.

The ultimate challenge is to identify what people use most often and how they structure their tasks, so that the complex capabilities of computers can be exploited while minimizing their difficulty of use.

Success stories with these approaches are already available. Pancake's group worked with the U.S.D.A. Forest Service to improve the "interfaces" that people use to retrieve certain types of data. They developed a much simpler system that's specifically aimed at helping first-time users. And the improvements even helped the original group of scientists make some key discoveries that they hadn't been able to see using the old, clumsier interface.

"In reality, we could probably improve every computer program or software that's ever been created," Pancake said. "That's the new challenge in building these systems and in educating our students. We haven't looked closely enough at how people use computers, and it's time we did."