Pleasing Users Isnt Easy

Opinion: Developers need to build performance and more into their systems if the goal is to make users take notice.

In a column last October, I warned builders of systems that users literally dont see things unrelated to what the users are trying to do. This makes users almost useless in diagnosing system problems.

Ive now learned that users cant even be trusted to tell you if something makes them happy—assuming they notice it at all. This suggests that anyone charged with making systems work better is going to be badly misled, especially over the long term, by user satisfaction studies.

Im talking about a phenomenon that psychologists call "hedonic adaptation." I first ran across it in an article by James Surowiecki in this months issue of Technology Review. Surowiecki wondered why surveys of American happiness show little improvement since the 1940s—and an actual decline since the 1970s—despite huge gains in every measurable aspect of life that ought to be a happiness driver.

Whether people are well off, in any absolute sense, seems to have little impact on how they feel about it. One researcher, Paul Pearsall, established a base line by asking 1,000 people to score their own happiness on a scale from 1 to 10. Pearsall arrived at an average of 6.8 on that scale. He then asked 150 lottery winners to do the same thing, both immediately after winning and again two weeks later. Their initial scores in the range of 8.6 to 9 declined during that time to an average of 6.7. Pearsall asked 200 paraplegics to give him the same two reports, one just after their injury and another two weeks later. Their initial scores of 1 to 2 on a 10-point scale rose over that time to an average of 6.9.

If its your job to figure out what people really like, this is a forbidding set of data points. It seems as if any given persons happiness is more a personal characteristic than a response to anything that we can change in that persons environment. It hardly seems fair; even if you try to be sensitive to users, you cant win. But research like this cant be ignored, no matter how much we dont like the results. We either find better measures to guide our designs or imitate the man who looked for his lost keys under the street lamp because the light was better there, even though hed actually dropped them half a block away.

If you want to build systems that people genuinely like, it looks as if you need to make people continually aware of how well the system is working for them so that they dont adjust their expectations and start to lose their sense of wonder. Think of the kind of little feedback line you get from a Google search: "Results 1-10 of about 459 for hedonic adaptation. (0.21 seconds)." Its subtle but never lets you forget what youre getting.

I wonder if making users happy with what you give them needs to be a design goal in itself if you have to compete against other offerings in an environment such as online retailing. Merely making your own system better, by any number of objective criteria, may not yield the expected results. It may take an independent engineering effort to make a system pleasing or delightful or whatever adjective you like to describe a system that makes people happy to have the privilege of using it.

I find this especially ironic in the week that follows the Consumer Electronics Show in Las Vegas, the annual mass pilgrimage to the shrine of technology delight. Apple and Sony seem to make this journey with ease, in terms of the physical design of their products and the experience of using them. Other pilgrims, such as Hewlett-Packard and Canon, despite genuinely excellent core technologies, seem less adept at creating that sense of satisfaction.

/zimages/3/28571.gifClick here to read more about the Consumer Electronics Show.

Enterprise success is increasingly driven by peoples response to customer-facing technologies. The difference between a Web site that attracts repeat visitors and one that does not, or a standards-based media player that becomes a popular icon and another that winds up in a clearance bin, is largely determined by reactions that cant be predicted by objective measures of performance. Dont let engineers tell you "But they have to like it." Its better to study this than to hope you get lucky.

The understanding of happiness is going to become an important tool for successful system builders—like the skills of user-interface design in the 1990s. A developers learning, it seems, is never done.

Technology Editor Peter Coffee can be reached at

/zimages/3/28571.gifCheck out eWEEK.coms for the latest news, reviews and analysis in programming environments and developer tools.