The State of Computer-Human Interaction

by Quinn Norton

You might say it's the toughest problem to solve in the modern world of computing; it's certainly the hardest to define. This month more than 1,800 designers, programmers, academics, professional researchers, industrial engineers, artists, and musicians gathered in Portland, Oregon, for another bash at the question, How do you make these monstrous electronics we've created easier and more pleasant to use?

Welcome to CHI 2005, the annual meeting of the Association for Computer Machinery's special interest group on Computer-Human Interaction (ACM SIGCHI).

Interfaces on Parade

The rooms are full of experiments and reports on the cutting edge of user interface design: exotic laser 3-D scanners that can track and record the path of a finger through the air; experiments in better ways to scroll documents and select objects with a ballooning version of the standard cursor; fixes to the more mundane, everyday troubles of modern technology, such as tiny tweaks to dozens of attempts to stuff a Qwerty keyboard into the tiny confines of the modern mobile device.

And then there are the crazier ideas. One short paper featured an edible positive-feedback user interface for programmers. Dan Maynes-Aminzade of the MIT Media Lab constructed a debugging interface that monitored malloc, realloc, and free calls in running threads. When memory was malloced, a jelly bean was dispensed into a tall, sealed transparent tube. When freed, the jelly beans fell into bowls, to be consumed by the hungry debuggers.

Maynes-Aminzade's equipment provided a tangible depiction of memory allocation and potential leaks in a program, with effects beyond a simple visual display. The temptation of those jellybeans also encouraged programmers to release all their memory as soon as possible. On the other hand, it might be that greedy programmers ended up writing equally memory-greedy programs just to get hold of more beans. And what of the impact of gourmet jelly beans? Or the possibilities of using cake in a similar construction? Clearly, more research is needed.

"What we're trying to do at CHI is have it all," says Gerrit van der Veer, a professor of computer sciences at Vrije University in Amsterdam, and this year's CHI chairman. Following the Dutch tradition of mixing art and science in academic disciplines, he's encouraged more playful experimentation and explicitly invited artists to participate in the conference: the end keynote was by Michel Waisvisz, the experimental electronic music performer.

Van der Veer also introduced an interactivity area featuring displays that verge more on art or fun as much as the hard computer science and psychology that the American CHI community is used to, like this year's "Virtual Raft" from the University of California at Irvine, a physical-virtual space mash-up in which participants carried digital stick figures bearing torches from one computer to another on a Tablet PC raft.

But is including all these constituencies enough? SIGSOC, the predecessor of SIGCHI, began in 1969. The CHI conference itself dates back to 1983, when tests of user interfaces involved examinations of how easy Unix was to learn, and the Mac was a glimmer in Apple's eye. Has all this consideration shown any concrete improvement in knowing how to build a good interface?

The Usability Hard-Liner

Related Reading

Information Architecture for the World Wide Web
Designing Large-Scale Web Sites
By Louis Rosenfeld, Peter Morville

Jared Spool has long represented the harder, data-intensive end of the CHI--he views himself carefully as a usability professional and nothing else. While his company, User Interface Engineering, is mainly a research firm, he will consult with companies about their usability issues; prices start at $100,000.

For that kind of price, you get a money-back guarantee that the changes he recommends will affect your bottom line to the exact degree he predicts. It's a uniquely concrete offer in a field still struggling to persuade the world that it can deliver real results.

Has it worked? Spool says no one has asked for their money back yet.

He describes the method: UIE examines the financial fundamentals of a company and tries to ask the right questions. If a customer wants their registration numbers up, UIE asks why, and what they hope to get from it. If they want higher conversions to sale, that's what UIE focuses its research on. "We don't put anything in our reports that doesn't affect the bottom line," says Spool. "Sometimes we'll see something, and we'll think that it would be an obvious improvement. But if we can't see a way that it will make the client money, we leave it out."

Given all of these concrete promises of improvement, Spool seems an odd person to be delivering his particular bit of controversy to CHI--that usability, as a field, isn't making much progress.

"I'm in a field that's 20 years old, and we barely know what we're doing," he says. "How do you say that, and then 'Now pay me a lot of money to do it'?"

"It's the big white elephant in the room," he says. Even when companies find something that works, usability experts can't say why it works--the field is simply too young and understudied to have a fundamental understanding of what makes a good or bad interface.

Usability professionals are loath to explain this problem, he says, "because they are a persecuted minority." As important as this work is, few companies ever concern themselves with whether their interfaces are any good. This contributes to usability professionals being the black sheep of the development family. They often feel like their jobs are perpetually on the chopping block, and thus they are often resistant to self-examination. Admitting their methods are far from perfect puts them into jeopardy.

For obvious reasons, not everyone in Spool's field agrees with him. His debate partner at CHI, Eric M. Schaffer, sees usability as a growing field with offshore potential, a service industry with consistent results, similar in many ways to bespoke software.

It was this kind of clash of ideas, expressed in a debate on the scalability of usability, that van der Veer wanted to see more of in CHI. Many of the panels used to be people agreeing with each other, says van der Veer. "The community needs controversy to be healthy," he says.

CHI in Context

For all the strange and interesting ideas that crop up in CHI--whether it's Scott Carter's examination of how topical bloggers imagine themselves (PDF) or how people in offices share iTunes music--it seems that many of the ideas never make it out into the mainstream where they can get played with.

Some of them, like Kick Ass Kung Fu's total immersive gaming experience, are fairly difficult to actually do. Others are Wizard of Oz interfaces with no code behind them. A few are pure thought experiments, and "some of them may be hoaxes," posits van der Veer.

Still, if the large turnout for events like the open source birds of a feather session is anything to judge by, code is getting more interested in CHI. With going concerns such as The GNOME Usability Project and long standing interest and investment from companies like Sun, Microsoft, and Apple, it's arguable that the field isn't such a persecuted minority anymore.

Progress over the first 20 or so years has been slow; that's a point everyone seems to agree on. But computer-human interaction and the principles that have arisen from CHI's past are beginning to show up in more products and fields than ever. Hopefully, in the long run that will mean our technology will be more pleasant to use, and our expectations about what it can do for us will be set higher--and met.

Quinn Norton is a writer and photographer whose work has appeared in Wired News, The UK Guardian, Make Magazine, Seed, and more.

Return to the O'Reilly Network.