WHEN THINGS START TO THINK | Chapter 2: Bits and Books

May 15, 2003
Neil Gershenfeld
Henry Holt & Company (1999)


. . . are things that think?

books that can change into other books

musical instruments that help beginners

engage and virtuosi do more

shoes that communicate through

body networks

printers that output working things

instead of static objects

money that contains behavior

as well as value

The demise of the book has been planned for centuries. This came first by fiat, with bannings and burnings, and more recently by design, with new media promising to make old books obsolete. Today’s book-of-the-future is the CD-ROM, offering video and sounds and cross-references to enhance ordinary text. Who would ever want to go back to reading a book that just has words?

Just about everyone. The state of the book business can be seen at the Frankfurt book fair, the annual event where the publishing world gathers to buy and sell books. It dates all the way back to the fifteenth century, shortly after the development of movable metal type created the need for a new kind of information marketplace. Approaching the end of the millennium the fair is flourishing, with hundreds of thousands of books and people on display. The only exception is the short lines and long faces in the hall displaying books on disk. People are voting with their fingers.

You don’t need to look beyond the book you’re holding to understand why. Think about its specifications. A book:

  • boots instantly
  • has a high-contrast, high-resolution display
  • is viewable from any angle, in bright or dim light
  • permits fast random access to any page
  • provides instant visual and tactile feedback on the location
  • can be easily annotated
  • requires no batteries or maintenance
  • is robustly packaged

A laptop meets exactly none of those specifications. If the book had been invented after the laptop it would be hailed as a great breakthrough. It’s not technophobic to prefer to read a book; it’s entirely sensible. The future of computing lies back in a book.

Isn’t it curious that a book doesn’t need a backlight? A laptop screen requires a power-hungry lamp, and even so isn’t legible in bright light or from an angle. Under those same circumstances a book continues to look great, and the reason why is surprisingly interesting.

The light in a laptop starts its journey in a fluorescent tube behind the display. It’s guided by a panel that spreads the light across the screen, losing some away from the screen. The light then passes through a polarizing filter that transmits the part of the light wave that is oriented in one direction, while absorbing the rest of it. Next come the control electrodes and then the liquid crystal. This is a fluid that can rotate the orientation of the light wave based on a voltage applied by a transistor, which absorbs still more light. Finally the light passes through color filters and a final polarizer, each taking its cut out. The result is that most of the light is wasted inside the display. The fraction that trickles out must then compete with the ambient light around the display to be seen. And since the light that does emerge must make its way through this obstacle course, it continues on in the same direction straight out, leaving very little to be seen at an angle to the display.

A piece of paper takes a much more sensible approach. The fibers that make up a sheet are actually translucent. Light striking the paper gets bent as it passes through a fiber, and since there are so many fibers this happens many times in many directions. The result of all this scattering is that the light spreads through the paper much like the spread of a drop of ink, eventually leaking back out. And just as the shape of a blob of spreading ink doesn’t depend much on the angle at which a pen is held, the light comes out of the paper in all directions, independent of how it arrived at the paper. This phenomenon, called optical weak localization, is what makes paper (or a glass of milk) appear to be white. It’s a very efficient system for converting light of most any color and orientation into uniform background illumination for text on the page. Crucially, like an aikido master redirecting an incoming attacker, this mechanism takes advantage of the light already in a room rather than trying to overpower it the way a backlight does.

In the Media Lab, as I started spending time with publishing companies as well as computer companies, I was struck by how strange it is to replace paper with displays that are guaranteed to be bulkier, take more power, and look worse. If paper is such a good system, why not continue to use it? The one advantage that a liquid crystal panel has over a sheet of paper is that it can change. Joe Jacobson came to the Media Lab to fix that.

He found that the key to making smart paper is a process called microencapsulation. This process grows tiny shells of one material around a core of another one. A familiar success of microencapsulation is carbonless copy paper. It used to be that to make a copy of a receipt when it was written, one had to slide in a sheet of an invariably messy paper that had an ink on it that could be transferred by writing on it, or touching it, or holding it, or rubbing it on your clothes. Now it’s possible to get a copy simply by writing on what looks like an ordinary piece of paper. The carbonless copy paper has been coated with small particles containing ink. The act of writing on it applies enough force to break open the shells, releasing the ink onto the sheet below. Simply touching the paper doesn’t have enough force to break the shells, which leaves the ink encapsulated where it belongs.

The beauty of microencapsulation is that it’s also cheap. There’s no need for a tiny assembly line to fill the particles; this is done by a straightforward chemical process. A solution is formed of drops of ink of the appropriate size in another liquid. Then a material is introduced that grows only at the interface between the two liquids. Once it has had time to form a thick enough shell around the droplets, the newly microencapsulated particles can be separated out from the liquid.

The toner used in a printer or copier consists of small particles that get fused to a piece of paper and absorb light. Joe’s group has developed a way to make microencapsulated particles about the size of toner (smaller than the thickness of a hair), which contain still smaller particles. The inner particles come in two types, one white and the other black. They also have a different electric charge stored on them. This means that in an electric field all of the white particles will go to one side of the outer shell and the black particles to the other. If the field is reversed, they will change sides. The result looks just like toner, because it essentially is, but it’s a kind of toner that can be switched on and off. They call this electronic ink.

The first thing that can be done with e-ink is to cover a sheet with it to make reusable paper. Instead of ending up in a recycling bin, the paper can go back into the printer after it’s used. A conventional printer has a complex mechanism to spread toner or ink over a page. Reusable paper starts with the smart toner already on it; the printer just needs a row of electrodes to switch the particles on and off. And since the process is reversible, the paper can go back through the printer to be reprinted over and over again. Reusable paper is needed because tomorrow’s paperless office has turned into today’s insatiable consumer of paper, carting in reams of paper to fly through ever-faster printers and just as quickly be thrown out. Reusable paper ends this high-tech deforestation by recycling the paper right in the printer. Instead of filling your trash with newsprint, a newspaper printed on reusable paper can go back into the printer at the end of the day to reemerge printed with the next day’s news.

Even better than reusing a sheet of paper is changing it while you watch. There’s enough room to drive a microelectronic truck through the thickness of a sheet of paper. This means that the electrodes needed to switch the particles can be moved from the printer to the paper itself. Forming a sandwich of paper, particles, and electrodes creates a new kind of display. But unlike any other display this one works just like printing and therefore can be viewed from most any angle in most any light, and it retains its image when the power is switched off. And even though the original goal of the project was to make displays that would look beautiful but be changed infrequently, the particles can switch quickly enough to approach the rate needed to show the moving images of a video.

Joe’s group began developing a printer to put the electrodes onto a sheet of paper to make electronic ink displays and soon realized that they could bring still more capabilities to a piece of paper. Integrated circuits are made by depositing conducting and insulating materials onto a wafer held inside a chamber that has had all the air evacuated so that it doesn’t react. The head of his group’s remarkable prototype printer accomplishes the same thing under air on a desktop by creating a tiny plasma column, like a miniature lightning bolt. This delivers to the surface the materials needed to print circuit elements such as wires and transistors. They developed the printer to put down display circuits to address individual pixels on the paper, but there’s plenty of space compared to a computer chip to add other functions.

The most exciting prospect of all is called radio paper. Since it takes very little energy to switch the electronic ink particles they can be driven by a solar cell, which is essentially a big transistor that can be deposited by the circuit printer. By printing still more transistors, a radio receiver could be integrated with the paper. Now your newspaper doesn’t even have to go into the printer at the end of the day. If you leave it out on your coffee table, the light in the room will power its circuits, which will receive a radio signal with the news to update the page. The circuitry can then flip the particles to reprint the page to have the day’s news waiting for you whenever you pick up the paper. Bad news for birdcages and yesterday’s fish: this would be a newspaper that is never out-of-date. While there’s still a lot of work to do to make radio paper a reality, the essential elements have already been shown in the laboratory.

Actively updating a newspaper points to the ultimate application of electronic ink: the universal book. Sheets of paper covered with the microencapsulated particles and electrodes can be assembled into a book, with control electronics embedded in the spine. This would look and feel like any other book, since it’s made out of the same ingredients of paper and toner. The only difference is that this book can change. After reading it, you can download new contents through the binding and onto the pages. This can even be done while you read. Pages other than the one you’re looking at can be changing, so that all of War and Peace could be read in a pamphlet of just a few pages.

The great innovation of Gutenberg and his peers was not the printing press, which was just a converted wine press; it was movable metal type. Before then printing was done from wooden blocks that had to be laboriously engraved. Movable type made it possible for a printer to produce a set of type that could be used to print all books, arranging the letters as needed for each page. Electronic ink takes this idea to its logical conclusion, moving the toner instead of the type. Now a reader can own just one book that can be any book, which can arrange the ink as needed for each page. Instead of going to a library to check out a book, the bits of the book can be downloaded onto the pages of an electronic book.

The electronic book ends the argument over old-fashioned books versus new-fashioned bits by recognizing that both sides have strong technical cases that can be combined. There are deep reasons why the old technologies in a book work so well, and there are new ways to emulate and adapt them. What jumps out the first time that you see a Gutenberg Bible is the glossiness of the ink. It turns out that Gutenberg made his inks by cooking a stew of oil and copper and lead that precipitated out little platelets that act like tiny mirrors, paradoxically reflecting light from a black background. The printing industry is still catching up to the formulation of inks of that sophistication. The same applies to the relatively recent discovery of how optical weak localization helps illuminate a printed page. It’s fair to ask that any successor to the book be able to do these things as well.

Drawing on all the resources now at our disposal to catch up to what Gutenberg was doing in the fifteenth century is a worthy challenge. But it’s not the real goal; it’s an essential warm-up along the path to asking what we can now achieve that Gutenberg could not. Gutenberg did exactly the same thing. He started by printing replicas of illuminated manuscripts; after all, that was where the market was. Then, as printing with movable metal type made it possible to assemble more pages than before, it became necessary to invent page numbering and tables of contents to keep the information accessible. These innovations departed from the past practice of manuscripts that were copied by hand, which were designed to minimize length at all costs. There was an interesting transitional period during which hand-illuminated manuscripts added page numbers to keep up with the fashion, before the bother of copying by hand became too hard to justify and quietly disappeared. Freed from the constraint of duplicating manuscripts, Aldus Manutius in Venice around 1500 then settled on the dimensions of the modern book. It was designed to fit in the saddlebags of traveling scholars, and his Press developed the italics font, to fill pages more efficiently than did fonts that imitated handwriting.

The arrival of electronic books now presents an opportunity to rethink how collections of books are organized into libraries. A library is, of course, much more than a book database. It’s a reading temple, a place for serendipitous browsing, a space to be with books, by yourself and with others. My only literary connection with John Updike is that we have the same favorite part of Harvard’s great Widener library—a small depression in the stone floor on the way into the stacks. To gain admittance to the stacks you pass through a gauntlet of dour gatekeepers, climb a set of stairs, turn immediately right, and then left. The generations of readers and writers pivoting at this last corner over the years have worn it down, hollowing out a bowl that provides a connection to everyone who has come before. I can’t write like Updike, but when it comes to floor abrasion I’m on an equal footing with him.

Once you’re past that corner, the bookshelves appear. I have no idea how far they extend, because it’s impossible to pass more than a few aisles without stopping to pick up an interesting title, then wondering about a book next to it, then realizing that you simply must spend more time in that section. The same thing happens with the old card catalog. Drawers holding worn cards that appear to date back to the opening of the building, if not the Civil War, speak volumes about what books are popular and what books are not. Flipping through the cards is even more likely than walking down the aisle to turn a simple query into a branching ramble through accumulated wisdom, which lasts until dinner or darkness forces an end.

In the noisy battle between the wired digerati and their analog enemies, there is a common misconception that the pleasure and utility of this kind of browsing must be sacrificed if a computer is used to access information. Too much is made of the inadequacies of transitional technology, while simultaneously too little is expected of it. An on-line card catalog that reduces each entry to a single screenful of flickering green text is incapable of communicating the popularity or context of a book. That’s not a failing of electrons or phosphors; it’s a poorly designed interface. The wear and tear of a card catalog, or of a library floor, are consequences of engineering decisions about the construction of cards, or floors, or shoes. That those technologies succeed through a combination of intended and unplanned interactions provides valuable guidance for future development, but it doesn’t mean that they are the only possible solutions.

There are many things that a card catalog, even Widener’s, can’t do. Of all major U.S. libraries, Widener is the only one that doesn’t shelve its books according to the Library of Congress indexing system. This is because Harvard’s system predates that of the Library of Congress; relabeling and reshelving the millions of volumes would be prohibitive. It’s much easier to move the bits associated with the books. Once a card catalog becomes electronic, the presentation can be freed from a single ordering defined with great effort by generations of librarians. Books can be shown by date, or location, or author, or discipline. Browsing need not be confined to books that are nearby in any one index; a smarter card catalog can show books that are related in multiple ways. And the card catalog can help provide context for the book by showing other books that have been checked out with it, or perhaps a map of where in the world the book is being read then.

Doing these things requires a visual sophistication that was beyond the means of not only early computers but also their programmers. Letting the programmers alone design the card catalog interface results in a faithful presentation of how the computer represents information internally, rather than how people use it externally. Creating books and libraries requires typographers and architects as well as bookbinders and librarians; the same thing is even more true electronically.

My colleague David Small belongs to a new generation of graphic designers who happen to be fluent programmers. Early fonts were designed to hold ink on the letter forms as well as to help guide the eye; David and his peers are using the interactivity of computers to let letters act and react in ways that are appropriate for the text and the reader. One of my favorite of his design studies presents Shakespeare’s works through the medium of a graphical supercomputer connected to a Lego set.

Initially the screen shows a hazy blob that in fact contains all of the words of all of Shakespeare’s plays. It’s a 3D space that you can fly through, so that you can grab the computer’s control knobs and zoom in to find a play, and approach still closer to find the text of a scene beautifully laid out. Rotating the space to peek behind the words reveals supporting annotations. Moving within a play and among plays is done by traveling in this space rather than by pulling books off of a shelf.

Since most of us don’t have much experience with flying through words, David adds some familiar guides. In front of the computer is a Lego set, with toy characters drawn from Shakespeare’s characters and themes. Selecting one of these physical icons highlights the corresponding parts of Shakespeare’s corpus on the screen in the background, showing first the large-scale patterns and then the individual words if you move in for a closer look. Spending time in this environment invites browsing in a way that can’t be done if the texts are frozen on pages. A good book uses a linear exposition to convey a nonlinear web of connections between the author and the reader; David’s environment comes much closer than a flat book to representing how I think about a Shakespeare play as I read it.

There’s still one more benefit of electronic books: access. There’s an implicitly elitist undercurrent in defending printed books over electronic books. The collection of the Widener library is a treasure that by design very few people will ever get to see, much less use routinely. Before printing, when books were copied by hand, they were so precious that owning one was a sign of great wealth, usually limited to religious orders rather than individual people. A single book is now so inexpensive that anyone can have one, but private libraries are still owned only by institutions and the wealthy. The budgets of most public libraries restrict their collections to a small subset of the wonders of Widener. Creating an electronic Widener is a heroic task that, even if done imperfectly, will make much more information available to many more people. Not everyone can own David Small’s supercomputer, but if the job of a local library becomes one of providing the tools to access information rather than holding the information itself, then it can do more with less.

Electronic books will be able to do everything that printed books can do, but one. They can’t replace the primacy of a historical artifact. This is a point that technologists like me frequently miss. Part of the pleasure in seeing a Gutenberg Bible is knowing that Gutenberg held the same object that you are looking at, drawing a connection across the centuries between you and him and everyone else who has come in between. A completely faithful replica that matches the specifications can never convey that.

Some information simply can’t be copied digitally. The rarest of the rare books in Harvard’s collection is bound in human skin. It is a modern witness to an old practice of people leaving a legacy of a bit of themselves for their successors in the binding of a posthumous book. The unsettling attraction of such a book is that it can never be duplicated.

For people who come to a library to experience the texture and smell and even the pedigree of an old book, there’s no point in arguing the merits of electronic ink. On the other hand, there’s no point in limiting everyone else by preserving old packages for timeless information.

In the end, the debate about the future of books is really about the relative performance of competing technologies. Books are designed by people, as are computers. There are plenty of examples of apparently irreconcilable disagreements over a new technology disappearing once the technology catches up to the specifications of what it’s replacing. Many people used to religiously proclaim that they couldn’t edit text on a computer screen; once the resolution of a screen began to approach that of the eye at a comfortable reading distance, the discussion began to go away. Now writing a text by hand is the exception rather than the rule. Such passionate debates get settled not by persuasion, but by technical progress making them increasingly irrelevant.

Along the way, the presumptions of a new technology must usually be tempered by the wisdom embodied in an old one. In the early days of the internal combustion engine it was interesting to race horses and cars. Now we have supersonic cars, but no one is arguing for the abolition of horses. Although horses are no longer the fastest means of transportation, no current car can recognize its owner with a glance, or choose a path through a narrow mountain pass, or be left in a meadow to refuel itself, or make a copy of itself when it begins to wear out. Cars still have a long way to go to catch up to horses.

The computers that are meant to replace books are destined to be transformed even more by the books. Reading is too varied and personal to be bounded by the distinction between digital and analog. I have some beloved books that I’ll never want to browse through in any other form, and piles of books threatening to topple over onto my desk that I’d love to access electronically, and some inscrutable texts that I’ll never get anywhere with until they’re available with active annotations and on-line connections.

Choosing between books and computers makes as much sense as choosing between breathing and eating. Books do a magnificent job of conveying static information; computers let information change. We’re just now learning how to use a lot of new technology to match the performance of the mature technology in books, transcending its inherent limits without sacrificing its best features. The bits and the atoms belong together. The story of the book is not coming to an end; it’s really just beginning.

WHEN THINGS START TO THINK by Neil Gershenfeld. ©1998 by Neil A. Gershenfeld. Reprinted by arrangement with Henry Holt and Company, LLC.