Today's operating systems are conceptually upside-down. They developed the hard way, gradually struggling upwards from the machinery (processors, memory, disks and displays) toward the user. In the future, operating systems and information management tools will grow top-down.
Computing power should make life simpler, not weigh you down with fancy features. Computing power should unify your life online, help you pull threads together -- not add more virtual shoe boxes for information to get lost in. I have time for one screen in my life. I need to be able to tune in one single information structure and know that my whole digital life -- every document, every file type -- is in there. And I need to be able to tune in this structure from any Net-connected device anywhere.
But operating systems have been traveling in the exact opposite direction, away from unity and simplicity. Today, most users' documents are distributed over many computers (often three "main" ones: at home, at work and a laptop). Inside each computer, documents are scattered as if someone had dumped them out of a low-flying airplane: some in the file hierarchy or on the desktop; mail in the mailer; bookmarks in the browser; images, other multimedia types, calendar and address information in other boxes. If you own a PDA, Internet-enabled cell phone or other digital gadgets, you have even more boxes to lose things in.
This is not merely unacceptable, it's crazy. No one can work effectively in such an environment. No wonder "can't find my g*ddamned data!" keeps showing up in surveys asking, "What bothers you most at work?" No wonder Bill Gates said, in the summer of 2002, "Right now, file space in any PC is a cesspool." No wonder I said a year earlier, in a PC Expo speech, that "the file system is dead -- that permanent bureaucracy that grows inside all our computers like crab grass." (Gates and I, just two peas in a regular old pod.)
Today's information environment is, in this sense, a huge step backward from the world of, say, 1946. In 1946, you could say "Pull the Schwartz file," and the whole Schwartz dossier would be there -- letters, memos, reports, photos, jottings, resumes, publications, bills, contracts and receipts -- the whole story.
Current operating systems have traditionally been built bottom-up: Start with the machine, then connect it somehow to the user. Their goal is to package the processor, memory, disk and other peripherals (which are unmitigated nuisances to manipulate directly), so that you can manage them by remote control. Instead of moving bits around the disk, you drag file icons around the desktop.
The next-generation operating system starts with the user. It ignores the underlying hardware -- and as a result, such systems are inherently less efficient than today's primitive, machine-centered ones. Instead, it reflects the shape of your life. Its role is to track your life event by event, moment by moment, thought by thought.
A life is a sequence of events in time. The future of information management is narrative information management, in which all of your stored documents are arranged as a "documentary history" of your life.
All the digital documents you create or receive, all the "community" documents you want or need to see, are laid out in one narrative stream with a past, present and future. The stream flows because time flows; the future (where your calendar notes, meeting reminders and plans are stored) flows into the present, then into the past. E-mail shows up at your "now line," and flows into the past. Everything you've got is on your stream.
Every word in every document is indexed automatically. All metadata, such as a document's origin and type, is indexed automatically. Image-indexing software is still primitive, but it won't always be. As soon as it works well (chances are within five years), it will be folded in. To find the information you need in this huge all-inclusive collection, you "focus" your stream as if it were a light-beam. Focus on "Schwartz," and the Schwartz dossier (lost since 1946) re-emerges -- not just one document, not just a list ranked by computed "relevance" and therefore in basically random order. What you get is the Schwartz story, from earliest contacts to appointments still to come. All kinds of documents are part of this focused stream, because it takes all kinds to tell a story.
The user interface will keep going in the direction it's been heading. It was basically one-dimensional in the DOS age of the 1980s -- you typed a command line, and the operating system typed lines back. With the Mac of 1984 and Windows 3.0 of 1990, the user interface (UI) became fully two-dimensional. The next-generation UI will be three-dimensional -- not literally (for now) but pictorially, in the sense that a printed tablecloth is a 2-D image but "Washington Crossing the Delaware" recedes into imaginary space behind the painted surface. The computer screen will no longer be a glass bulletin board with windows and icons stuck to it; it will be a viewport with an "information landscape" in simulated 3-D on the other side.
Why? Because your hardest task in today's information gale is to master the big picture. The mind's picture-processing capacities are amazing, and you need to use them to see as many electronic documents at a glance as you can.
That's why people cram their desktops with icons. Suppose you were looking at a thousand people. If you lined them up side-by-side, you'd have to walk the length of the row or stand way back to see them all. But if you put them in a column and stood right in front, slightly to one side or above, then you'd see the whole parade in one glance. Foreshortening does that for you. Next-generation UIs will use foreshortening to show you a deep parade of digital documents instead of today's flat chaos. Simulated 3-D interfaces are a perfect match to the needs of a narrative stream -- which just happens to be a parade of documents.
Software revolutions will keep happening in the way they've been happening, without requiring that you throw out your old software and data and start over. To install Unix in, say, 1976, you had to rip out your old operating system first. By 1990, you laid down Windows 3.0 like carpeting: DOS stayed put underneath. The coming info-management revolution will be the same sort of low-trauma affair. Your new software will sit on top of Windows and your Windows applications.
The approaching operating system and info-management revolution is no mere speculation. Microsoft's "Longhorn" is supposed to be ready in two years, and its goal is to solve the information management problem. (Bill Gates, summer '02: "The one question we're trying to solve with Longhorn is, 'Where's my stuff?'") Other companies have already taken steps in that direction. What system will win, I can't say. But the future of narrative information systems is a marketing and user-adoption question, not a technology challenge. The software exists today and some narrative stream or other is the future of operating systems.
David Gelernter is professor of computer science at Yale University and chief scientist at Mirror Worlds Technologies Inc., the New York-based maker of the Scopeware information management software. A beta version of Scopeware Vision is currently available for free download from the company's Web site.