The Design of Future Things

In Don Norman's latest book, whose title echoes that of his 1988 bestseller The Design of Everyday Things, he turns his attention to the increasing number of "smart" products that don't just react to us but predict, pre-empt, correct and even supersede our activities. As one might expect from somebody who, in his earlier book, found water taps and light switches fraught with difficulties, Norman considers with ambivalence - of the sort that characterises the best accounts of new technologies - the idea that automobiles may soon improve sloppy driving, refrigerators monitor what we eat and electronic assistants comment on our spending.

According to Norman, we will soon live in partnership with our products, negotiating everyday life with a variety of artificial intelligences. The problem is not with autonomous systems, which when successful keep us safely out of the loop, nor with augmentative ones, which leave us firmly in control. Rather, it is with semi-autonomous systems, which by design or through inadequacy must coordinate and trade control with us. This spells trouble because people and machines cannot share "common ground", the set of experiences, knowledge, motivations and goals that make communication possible. Without common ground, machines and people struggle to understand one another, sometimes with disastrous consequences.

The solution, Norman suggests, is to use "natural interaction" to control and monitor systems, allowing awareness and operation at the periphery of attention to prevent unexpected and disastrous breakdowns.

Ironically, given Norman's concluding call for a "science of design", he elaborates this largely through the sorts of case studies (including, here, teakettles, horses and the Segway Transporter) that artist-designers use to convey complex ideas.

His examples don't always hang together, and analytically oriented readers may complain that "natural interaction" amounts to little more than an exemplary brand name and a set of largely anodyne design principles (for example, "be predictable"). Nevertheless, the notion will remind designers to eschew arbitrary and impoverished forms of interaction.

The basic argument of the book emerges among numerous examples of trends in "smart" products. The vast majority concern automobiles - presumably reflecting Norman's consulting contracts, though his ardent descriptions of driving suggest a personal passion - but smart homes, robotics and education also enter the discussion.

Though something of a magpie collection, this eclectic range of topics is one of the book's strong points: an eye-opening break from the PCs, websites and mobile phones dominating most research on interaction.

Norman puts together a resolutely psychological account. Unlike commentators such as Adam Greenfield who highlight the potential for context-aware networked technologies to disrupt privacy and lead to coercion by convenience, Norman reacts to such possibilities with a rueful shrug (those pesky machines!) or sees them as desirable (systems should steer people to act correctly). The concern here is with the individual's immediate experience: the satisfaction of using machines that work, the frustration of those that do not.

Norman's chatty and occasionally self-congratulatory tone is sometimes disturbing, at least to this immigrant admirer of British reserve. The way he loops and echoes his ideas throughout the book, while allowing detail and substance to accrue, tends to obscure its structure.

Still, one doesn't read Norman's books to agree with them, or even their overall perspective. Instead, his strength is in identifying genuinely important areas and issues, and allowing us to look over his shoulder as he uses his personal experiences and reflections to tease out some sort of big picture.

From this perspective, this book succeeds in gathering trends on "smart" products and considering what makes them satisfying or maddening to individual users. It's not the last word on the subject, but it's a good place to start.