[Reposted from my Red Herring blog, 2005.]

Whatever happened to the paperless office? A decade ago, futurists and pundits were confident that personal computers, CD-ROMs, and the Internet would render books and magazines obsolete, turn paper money and checks into curiosities, and bring about the paperless office. Of course, none of these predictions has come true. Books and magazines are still around, and while total paper use has declined in the last couple years, offices use more paper now than before personal computers became commonplace.

What this suggests is that the relationship between the paper and electronic worlds is more complicated than we first thought. On its own, this is hardly surprising: it's a truism that we overestimate the magnitude of technological changes in the near term, and underestimate them in the long run. But understanding why early predictions about the death of paper haven't come true will help us map out some of the possible futures of paper in the coming age of pervasive computing. This is a world in which computers are small and cheap enough to be embedded in virtually any built object; information can be associated with everyday objects and places; and networking technology allows devices to communicate and cooperate on behalf of their owners.

If paper was supposed to be made obsolete by the personal computer, what future could paper have in a world in which computers fly off the desktop and are everywhere, in everything?

First we need to understand why we still read, write, communicate, and collaborate with paper, even as electronic substitutes have become cheaper and easier to use. After a decade of the paperless office not coming to pass, we have a good sense of what's going on.

For one thing, new media wasn't the dawn of a revolution in the way content was created, distributed, and read, but the culmination of a decades-long evolutionary process bringing computers into publishing. In the 1960s and 1970s, publishers adopted mainframe computers to handle in-house editing and typesetting—work that ultimately yielded a printed book, newspaper, or magazine. In the 1980s, with the emergence of word processors, authors were able to produce their own electronic documents. Finally, in the 1990s, the falling costs of multimedia personal computers and appearance of Web browsers made it possible for readers to interact with digital texts. (The fact that had publishers had built up vast databases of machine-readable data that could go online also gave the Web an extra kick.)

For another, paper turned out to have certain qualities—or affordances, as Abigail Sellen and Richard Harper put it in their excellent The myth of the paperless office—that computers couldn't match. Few of us, it turns out, read in a straightforward way: we graze across several manuals and papers simultaneously, underline and annotate documents, or take notes as we write—all functions that can be only incompletely duplicated with computers. In many workplaces, paper documents are tools for organizing workflows, or serve as indicators of where work is building up: what is easily visible in the form of slips of paper moving about an office becomes obscured when that data moves onscreen. Spaces like whiteboards are better-suited to collaborative work than computer screens, as they allow multiple users to work simultaneously, to easily build on each other's ideas, and to see how a group's thinking is unfolding. Documents aren't just a recording or storage medium; they're a medium through which coworkers work, and through which collaboration happens.

Finally, a lot of skilled work is more than rule-following, and computerized workflows have a hard time accommodating to the realities of such jobs. In a police department that equipped officers with laptops and software for writing and transmitting electronic crime reports, for example, officers found that the sensitive work of dealing with victims and talking to witnesses didn't mix with typing on keyboards, or the rigid formulas that the software wanted to impose on the interview process. Indeed, Stanford sociologist Aneesh Aneesh has argued that this creativity-killing aspect of computerized work is exactly what many companies want when they install workflow systems.

The fact that paperless office hasn't come to pass, though, doesn't mean that it never will. After all, while desktop computers and clunky monitors still bear a family resemblance to the terminals and CRTs of a bygone era, technologies like smart dust, RFID, wireless, and flexible displays are completely different. Likewise, the vision of computing that drives pervasive computing research is nothing like today's chunky interactions with keyboards and monitors. Should it come to pass, the model will deeply affect the ways people will interact with information, and the ways information will interact with the world.

Most important, computers will be everywhere, and we'll interact with them constantly, but we'll hardly be aware of them. Think of what driving a car is like today. A luxury car can have a hundred microprocessors or more in it, monitoring and controlling everything from the ignition to the exhaust to the temperature of the seats. Yet the experience driving a Mercedes C-Class or a BMW 700 series car (so I'm told—I've never been behind the wheel of either one) is nothing like checking your e-mail or fiddling with a spreadsheet. An even better, or at least more financially accessible, model is the iPod. It's a big hard drive, some software, and thousands of files; but from an interaction standpoint, it's nothing like a computer.

A second change is that new mobile devices will allow us to access information in "real space" as well as "real time."

Finally, in the workplace, one will no longer have to choose between dealing colleagues and computers: digital devices will support face-to-face interaction, sociability, and creativity, rather than competing with them, forcing them into narrow channels, or redesigning the workplace.

So what does this mean for the future of paper?

The bad news is that while the "electronic piñata" will continue to get fatter and fatter, a smaller and smaller proportion of digital information will ever get onto the page. A lot that information is going to support machine-to-machine communication, and except in cases where there are liability concerns, won't get printed—and probably stored only for limited times. Likewise, "printing" an MP3 or video is nonsensical. Even instant message chats and camera phone pictures, despite their family resemblance to older media, are printed far less frequently than e-mails or digital pictures.

But the good news is that we'll keep sending around e-mails, PDFs, and other documents that need to be printed for one reason or another. This leads to one scenario, in which paper continues to be the celluose tail wagging at the end of the silicon dog. Paper remains a part of knowledge work, packaging, signage, and other familiar industries and products, but is slowly eclipsed by new display technologies, and it plays a vanishing role in entirely new kinds of work and social activity.

This is a simple extrapolation of current trends, and as such is a future that has a fair degree of likelihood. But there's a second, which you can see in weak signals coming from academic laboratories and a few startups: one in which paper becomes a digital medium.

Could paper become the hot new medium of the next digital age? The idea isn't quite as farfetched as you might think.

A number of scientists are experimenting with printing electronics on coated paper. Regular paper, with its network of wood fibers, is far too coarse to print circuits; but coated papers and paperboard composites (which mix paper and other materials) show promise. This spring, in fact, Swedish company Cypak AB announced that it had printed a computer on a paper-plastic surface; the device's capability places it somewhere between a Sinclair Z80 and a Radio Shack TRS-80. (Interestingly, the company isn't selling the device as a low-end computer, but as a smart, high-end RFID tag.) This may sound pretty underwhelming, but anyone who bought a Z80 in, say, 1979 can remember how fast the personal computer evolved in the next ten years.

But why even try to make a paper computer? Why not stick with silicon or plastic? Paper is cheap, its properties are well-known, and yet it's versatile. With the right processing or coating, it can be fashioned into everything from packaging for appliances, to a medium for ultra-sharp photographs, to tissues. And you can make it by the mile. Paper, in other words, is a malleable, mature, slightly boring technology: perfect for cheap, flexible devices.

And of course, paper is appealing because of its affordances: its flexibility, its brightness, and its familiarity. For companies trying to make paper-like electronics, it makes sense to try to actually, well, just use paper.

The Cypak case points to a scenario in which paper, far from being an afterthought in the digital world, becomes an integral part of it. Paper becomes a substrate for a new generation of flexible electronics, highly portable displays, active and reactive packaging, and dynamic signage. At the same time, printing becomes a key manufacturing technology for flexible electronics, as desktop printers acquire the capability to turn out highly customized electronics (as discussed in an earlier column), while industrial printing becomes a tool for manufacturing trillions of data tags, sensors, and processors.

In other words, it's likely that in the future computers won't make paper obsolete. It's even possible that computers will become paper.