Mirror Worlds: or The Day Software Puts the Universe in a Shoebox . . . How It Will Happen and What It Will Mean

By David Gelernter Oxford University Press, New York, 1991, 237 pages, $24.95.

Reviewed by Ernest Davis

A mirror world is, or rather will be, a computer system that allows its users to see, study, explore, and understand some substantial setting in the real world--a school, a hospital, a city, the whole world. Users can move about in mirror worlds: geographically, from room to room or street to street; or topically, from viewing academic structure to viewing financial structure; or through levels of abstractions, from a high-level overview to low-level detail; or through time, to the state of the setting a week or a year ago. At each point, the mirror world provides its information in a picture, detailed but easily grasped.

The information in a mirror world is always up-to-the-minute, continually updated by a battery of sensors and input devices on the site. The user can even communicate with other users who are "visiting" the site through their own mirror worlds.

David Gelernter believes that mirror worlds will soon be with us and that they will be overwhelmingly important--far more important than such glitz as robots, computers in education, or "virtual reality" systems. Mirror worlds will be the great public works of the next century--magnificent accomplishments of engineering and art, comparable to the Piazza San Marco in Venice. He is not entirely sure whether mirror worlds will be a Good Thing or a Bad Thing, an issue he discusses in dialogue form at the end of the book, but he is quite positive that they are inevitable and will have revolutionary impact.

Gelernter has written Mirror Worlds to prepare us for their coming. His book explains for a popular audience what mirror worlds are, how current research (primarily Gelernter's own research) is laying the groundwork for them, and how they will change the face of society. Gelernter writes entertainingly and energetically, in an engaging, chatty style. The book is enriched by his eclectic interests and knowledge and by his provocative opinions about all manner of things. If nothing else, his book will provoke any reader to think, which is no small virtue.

The better part of the book (chapters 3 through 6) is an explanation of Gelernter's research in programming languages, distributed systems, software engineering, and artificial intelligence, written for a popular audience. (Gelernter is a leading researcher in the first two areas and has done substantial work in the last two.) It is extremely difficult to make such material understandable to a nontechnical audience, and, on the whole, Gelernter succeeds remarkably well.

Some parts are extraordinarily good. His explanation of what he calls the principle of "espalier"--the imposition of a uniform conceptual framework on the components of a system--is the best I have seen anywhere. Equally good is his discussion of the elusive concept of software. A piece of software, he explains, is not the physical computer that implements it, as a Rembrandt is not a collection of pigments; nor is it the text of a program, as a symphony is not a series of notes written out on paper. What software is, he says, is an abstract structure of data and control.

Elsewhere, Gelernter is less successful. He tries valiantly to convey the idea of object-oriented programming through an elaborate metaphor of an actor working off a script inside a command post overseeing a landscape, but I doubt that it will mean much to either the average or the technical reader. Personally, I found myself mentally translating it into the standard jargon as I read along. His discussion of artificial intelligence, both in content and in style, is very much along the lines of Roger Schank's work, both in content and in style. I strongly disagree with it, but this is not the place to discuss it.

But the software is only infrastructure: What really matters here are the mirror worlds. Gelernter describes at great length his vision of what mirror worlds would do for us. Regrettably, I find his account rather flat. For instance, he devotes an extended passage (pages 200-201) to an overview of a city as presented by a mirror world. The mirror world displays a map of the city in which red/blue color-coding is used to indicate the "level of performance" of a whole host of things: the level of traffic in the streets, prices in the stock market, delays at the airport, crime in the subways, test scores in the schools, and cleanliness in the parks. The map also shows little bulletin boards that announce public meetings, museum exhibits, and so on. Reducing the quality of a school to a color between blue and red is admittedly a bit simplistic, Gelernter says, but it gives the user an idea, and the mirror world will happily provide more information to the interested user. But he entirely overlooks the more serious problem: the sheer pointlessness of displaying this incredible hodgepodge of unrelated parameters on a single map. This map would be an overwhelming, confusing, useless mess.

Part of the problem, I suspect, is that Gelernter himself is verbally rather than visually oriented. (He mentions that he has written some articles on Biblical literary criticism.) The diagrams in this book are remarkably uninteresting and unhelpful, given its subject and the energy with which he advocates pictorial presentation of information.

Another problem is that Gelernter doesn't seem to have given much thought to the question of what it would actually be like to use the mirror world. Consider, for example, his suggestion that users should be able to meet other users who are exploring the same place. How would it work? Would the users appear in the place in pictorial form, as if they were actually there? Suppose there are more people visiting a place than would physically fit there; what would the display show then?

Suppose you go back in time to visit a place; is your ghost forever after visible at that time and place? What, actually, is a place in a mirror world? If Joe is visiting today's council meeting and Sue is visiting the meeting of a month ago, are they in the same place? How about if they are both viewing the high-level chart of administrative responsibility in the educational system? I don't say that these are serious problems; I'm sure that there are perfectly adequate solutions to all of them. But I find it curious that there is no sign that Gelernter has thought about them at all.

When I put the book aside and ask myself how I would use a mirror world, if I had one in my office, the answer is plain: It would be great to shop with. I would go, in spirit, from store to store with the flick of a mouse, comparing prices, viewing the merchandise, and placing my order via e-mail. Heaven. But such uses are too frivolous and too bourgeois for Gelernter; he is not devoting years of brilliant research to improving mail-order catalogues. His uses for the mirror world are all civic-minded. (The city map described here is a typical example; Gelernter specifically excludes commercial advertising from his mirror worlds.)

His purpose is the radical improvement of society, which mirror worlds will accomplish in two ways. First, they will simply make it easier for concerned citizens to explore the ins and outs of the world, to find out what is happening in their communities, and to make their voices heard in a participatory democracy. Second, and more importantly, the mirror world will give its users what Gelernter calls "topsight": the ability to take an Olympian view of the world; to see the whole picture and the underlying connection between things; to abandon the narrow concerns of interest groups and seek the broader good of the community as a whole.

Would the ability to scan the world necessarily give us a broader view? Jorge Luis Borges, in his story "The Aleph," tells of a poet who has found in his basement an Aleph, a point from which all points in the world can be seen. But the Aleph brings him no insight whatever. The poetry it inspires is doggerel, and the man spends his life in absurd vanity and petty literary jealousy. The Aleph in the basement is "a mere optical device," says Borges, "a false Aleph."

Ernest Davis is a professor of computer science at the Courant Institute of Mathematical Sciences, New York University.