Power and Pedagogy: Transforming Education through Information Technology

Chapter Two - The Computer as a System

Computers are like wheeled vehicles: they come in many shapes and sizes, each serving a different purpose. Moreover, the computer has yet to mature. It is an emerging technology. Hence, to determine the potential of computers in education, we need to understand what the computer is. To start, consider two distinctions, one between transitional and mature technology and the other between artifacts and systems.

Complicated technologies take a long time to develop their potentialities. They also take capital. Developers cannot perfect their technology in endless years of laboratory work and then deliver it, refined and complete, to a grateful public. To underwrite the costs of perfecting a technology, developers must bring it to market long before it is mature. Profits from transitional implementations sustain the development work, providing resources and disclosing unexpected opportunities for use. Computers have exemplified this drawn-out development: computers have evolved through several distinct, quite profitable incarnations, yet neither the time-sharing mainframe nor the stand- alone micro indicate fully what the computer will be when the technology matures.

In common speech, we generally do not distinguish between typical technological products and the technical systems that make them usable. For instance, "television" can refer to the TV set, that ubiquitous appliance, or to the whole industry -- the networks, their broadcasting installations, the news teams and production studios, advertisers, and all. Likewise, "automobile" can refer to the car in my driveway or to the vast infrastructure -- the manufacturers here and abroad, with their suppliers, advertisers, and dealers; all the roads and bridges and the builders constructing and maintaining them; the service stations and oil producers, refiners, and marketers; and the myriad of designers, workers, police, and service people who make the system go. The car is both a separate artifact and a complex system.

Currently, "computer" usually calls to mind the artifact, the stand-alone personal computer, like the one on which I am now writing. Most of us do not think much about the complex system of which my PC is a transitory part. Computers as a system are important, however. The significance of computers for education will not be well understood by thinking simply of a lot of separate machines sprinkled through existing schools and colleges. Computers are an emergent infrastructure, a system, fully as complicated as that of the car. We need to think about what that system is and how that infrastructure will work. Computers as a system can be a powerful agent of change in education.

To grasp the computer as a system, particularly as it matures, let us concentrate, on neither hardware nor software, but on an underlying process, the digitization of information. The computer, as a system, introduces a new way of representing information in our culture, a new way of encoding ideas. When complete, it will constitute a deep transition in our history, one equal in importance to the introduction of printing, quite possibly to the development of writing itself. Essentially, the computer as a system will envelop all previous modes of representing information, preserving and empowering them by integrating once separate domains of communication into a unified, "multimedia" system.

Information in Matter and Energy
Think of the ways we commonly represent information -- a scribbled note, a neatly printed page, a reflective sign, a painted picture, a ruler uniformly marked, a measuring cup, or the symbolic forms of church or court. With these, people have encoded ideas and information in material objects, in the ink upon the page or the shape of the sculpted stone. Put most generally, through traditional ways of encoding ideas, people expend energy to transform matter in ways that they will find meaningful, making enduring marks and forms in which ideas inhere. People locate the information in the material object. When they do this according to a defined convention and art, the tangible, palpable results are our major forms of traditional communication -- documents, sculptures, pictures, monuments.

Starting with the telegraph and developing through the telephone, radio, television, and computer, people have begun to put their information into controlled pulses of energy itself. The material object, say the telephone, becomes a kind of transparent medium for an infinity of possible conversations encoded in different electrical waves that the phone will generate, transmit, and receive. Increasingly people are representing information in controlled states of energy, not in matter, as they did traditionally. The new practice requires various material tools, with which people apprehend on their human scale the information located in energy, but the information is not in the material, but in the energy. Thus the TV translates the information bearing energy into a material form that I can watch. The picture hanging on my wall is what it is because the information that it contains is in the material that makes it up. My TV, in contrast, can receive an infinity of images because the information it displays is not in the material of the set, but in the electromagnetic waves that it picks up and decodes for me.

This practice of locating information in energy states is not entirely new in our culture. One can take sound to be a form of energy, not a state of matter, and hold that through speech and song people have long encoded information in energy, using the ear as the naturally developed, material receiving apparatus. Other senses, too, especially sight, kinesthesia, and the ability to feel hot and cold, derive much information from energy states and forms of force. Some traditional tools of communication and control also provided readings of the information in energy states. The clock measures time by controlling the release of energy in uniform units. The compass provides a most informative reading of the orientation at any location of the earth's magnetic field. The governor on a steam engine directly translates a change in its energy state into a controlling action. Like the TV -- but unlike the painting on the wall -- clocks, compasses, and governors all inform their users through their changing readings, not through their static states. More strictly speaking, these instruments display information that is fortuitously located in states of energy, rather than encoding it in those states. Traditionally, only the voice and musical instruments went beyond display to encode.

Up until very recently, information encoded in energy has been, however useful and dynamic, troublesomely transient. Speech is the paradigmatic instance. It is powerful and nuanced, yet fleeting and unstable. For a time memory preserves its residue, and writing fixes a stiff representation of it in stable matter. But much is lost. This transience also characterizes many modern media that encode information in wave forms, substituting electricity for sound as the energy medium. Thus telephone, radio, and television have enabled people to encode sound and gesture in electromagnetic waves, amplifying these vastly, without making them much more enduring. Recording signals on tape and other media makes such material reproducible, and thus enduring. Yet this has been a recent, ancillary development. So far, the power of electromagnetic media has resulted from the breadth of their transient reach, not from the ease with which productions can be reproduced.

This transience of electromagnetically encoded information fundamentally affected the usefulness of broadcast media for education. Entertainment results from encountering cultural experiences for their immediate, present value -- they amuse, inspire, absorb, purge, distract, or release us now. Education involves us with cultural works of enduring importance -- we acquire skills, ideas, beliefs, knowledge, information that will empower us over time in the conduct of life. The things at stake in education are the elements of the culture that are on-going, lasting resources. Consequently, the educationally important media are the ones that represent and make such enduring ideas and skills available to people. For the most part, these have been the media that locate information in material objects, particularly in printed texts and pictures.

Commentators complain that educators have done little with the major communications developments of the twentieth century. Despite high hopes, radio and television have not become important educational resources and some infer therefore that education is resistant to technological change. This inference is wrong. The photograph, which extends the pictorial capacity to locate information on film and paper, has been seamlessly incorporated into education. It improves the capacity to work with lasting ideas and information, and educators have quickly adopted photographs in the processes of research and instruction. As conservative a field as art history took without hesitation to 35mm color slides because they served the intellectual needs of the subject. So too, recorded music has become a natural part of music education, far more so than have broadcast performances, for the recordings are stable, enduring resources that different students at different times can study, each with unique purposes in mind. Recordings suit the needs of education because they are stable, easily stored and retrieved, while broadcasts suit the needs of entertainment, absorbing us in their immediate presence.

Educators cannot resist new technologies, provided those technologies have characteristics suitable to educational purposes, foremost among those being a permanence in time. Stop for a moment to consider film, which encodes information in stable, material form yet has not come into robust use in education. Is it an exception to the rule here propounded? No. With respect to dissemination and retrieval, film is not as stable as it might seem. Film is bulky, hard to store, costly to project, and easily damaged. It can be best disseminated in a quasi-broadcast fashion with prints distributed to numerous theaters more or less at the same time, with the production playing as long as it can command a full audience and then disappearing into an archive, from which films are not easy to retrieve. These distribution constraints have made movies, until very recently, far more effective as media of entertainment than of education.

Computers as a system will change that, and much more. Broadly speaking, the communication innovations since the mid- nineteenth century have created a family of technologies for encoding diverse forms of information in energy. The computer is the most recent in this series of innovations, and it is likely, historically, to incorporate all those leading up to it into itself. What seem to us to be separate industries with separate technologies will become branches of a single comprehensive industry and technology, the computer as a system.

One can now see large corporations jockeying to capitalize on this consolidation of technologies. For instance, the major Japanese electronics firms seem to be calculating that they can best shape this process by combining business communication with the entertainment industries, buying up major entertainment conglomerates while designing ever-more computing power into home entertainment devices. The emerging system, however, may in fact be far more robust if built on a combination of telecommunications and education. Digital technologies enhance the staying power of information in time, expanding its educative power relative to its currency as entertainment. We will be developing the thesis that the computer is rapidly incorporating the modern media in one comprehensive system, a system of knowledge and education.

The Analog and the Digital
We distinguished between technologies that locate information in matter, for instance sculpting and printing, and those that locate it in states of energy, for instance radio and computers. Among the latter, we need to make important further distinctions, which have to do with the techniques people use to encode information in energy. To grasp the cultural import of the computer as a system incorporating all the media of communication, to appreciate its potential power, we need to reflect on the way that it encodes information in energy, seeing how that differs from other techniques.

Analog coding serves effectively for some specialized computational purposes, but almost all computers, from tiny palm- tops to huge supercomputers, work with information stored in digital code. Such digital code differs profoundly from the analog codes used typically in radio and television. In the paragraphs that follow, we will reflect on how digital code differs from analog and then consider five matters that determine the value of information for human activity -- production and reproduction, storage, transmission, selective retrieval, and intelligent processing. Through these considerations, we will form a sense of why the computer, as it matures, will be a very significant step in our history.

Note at the outset that we could apply this distinction between analog and digital coding to the media that use matter to carry information. For instance, painting and sculpture are highly analog media, whereas alphabetic writing is interestingly ambiguous. It is analog insofar as it is phonetic and digital insofar as it is a prescriptive set of legible conventions. But it would take us afield to pursue these distinctions with respect to media that locate information in matter, for our concerns here are primarily with the media that carry information in energy. How does the digital coding of information in energy differ from the analog?

Analog systems encode information in energy by using the properties of continuous waves so that each successive change in the amplitude of the wave will be analogous to a change in sound or appearance in the human world. Lets construct an example. Take a dishtowel. Holding each corner of one end in each hand, flap it rhythmically in front of you, making it undulate up and down. It is not hard to control the beat of the flapping, making each flap identical in duration, perhaps slow and long or quick and short. That beat is like the frequency of an analog signal. Usually it does not carry the information, but when we are surrounded by many different signals, each with a different frequency, it allows us to find the one signal we want. Observe the flapping towel, however. From beat to beat, it will have all sorts of variations, curving this way then that, depending on subtle changes in the orientation of your hands to each other and the tension they put on the cloth. If you could control the flapping skillfully enough, you could make each change in the way the towel undulated match some other, analogous change in a completely different wave, say the ever changing sounds of a symphony or rock concert. At that point, you would have encoded the concert in the flapping towel rather like the way radio encodes a concert in an electromagnetic amplitude or frequency.

Like the sound itself, the flapping is transient. Analog encoding depends on making significant changes in the energy state of the wave, a most unstable phenomena. Digital encoding is much more stable. Put down the towel and flip the light switch on the wall. The switch has gone from "on" to "off;" it was stable in its former state and is stable in its latter. The light switch is a digital device, although one that does not accomplish much in the way of communication and control. To see simple signals controlling more complicated processes occurring around you, look at another digital switch, the stoplight at the corner. It has two basic states, red or green -- amber is not really a state, but a cue that a change of state is about to happen. There are two unambiguous states, green-go, red-stop. These are easily standardized, stable, and remarkably effective in controlling complex flows of matter and energy. The stoplight is very much like the small charge in a transistor in that one state allows traffic to move and the other calls it to a halt.

Our basic red-green stoplight is a binary digital system -- binary because there are two alternatives and digital because those consist of discrete, unambiguously different states. The typical electric stove, with options on each burner running from warm to high, has a quinary digital control on its coils -- quinary because there are five alternatives and digital because each of these is distinct from the others. Thus, digital systems can in principle have different numbers of basic alternatives, but computers almost always use a binary system, building many subtle variations from a multiplicity of either-ors.

A digital state is what it is, discrete, unambiguous, disjunctive. Digital code does not capture changes similar to other changes, it presents a set of values that are what they are. Digital coding follows a principle akin to encrypting -- there is only one message, which, when encrypted, is put in a way that makes it look indecipherable. With the appropriate key, however, the cryptographer finds the message, not something like the original, but the original itself. For instance, the apparatus for recording music digitally measures sound frequencies at successive instances and records the numeric value of the frequencies. These are samples of the actual sound, not likenesses to it. Digital coding samples a phenomenon, registers the sample, and then reproduces the phenomenon from the sample. If the sampling technique and the technique of reproducing from the sample are very good, it can be extremely hard to distinguish the original from the reproduction. What is coded is an exact value, precisely what it is and nothing else.

What is encoded digitally, therefore, is actually very different from what is encoded in an analog system. The digital system encodes a sample of the thing whereas the analog system encodes an analogy to it. Again, let us construct an example. Consider a full wheel of cheddar cheese. Describing the cheese by analogy can be difficult. I might say it is about the size and shape of an old-time hatbox and that it is heavy, as if the hatbox were filled with water. Its color is like custard and it tastes -- this is the important, difficult part -- somewhat like grapefruit, although its texture in the mouth is very different, a bit like a firm fudge that crumbles and then softens into a paste as one chews it. Describing the cheddar by a sample of it is much simpler. I cut you a little piece, perhaps several from different places in the wheel. The sample is the cheese and you can sniff it or taste it directly from the sample.

When we digitally code the sample, we register what the sample is on an appropriate scale and we code that value, not some approximate likeness to it. Consider recording a singer's voice digitally. At numerous intervals the recording samples the exact sound frequency of the voice, registering in a matrix of precise values what, at each sampling instant, the frequency was. The digital recording carries no information about the voice during the intervals between the sampling instants, but it carries the exact frequency of it at those instants. If the sampling frequency is sufficiently rapid, the sound of the reproduced voice will be essentially identical to the original. Digital code allows the playback to reconstruct the voice. Thus, digital coding registers sampled values, not approximate similarities. That is its first point of difference with analog coding.

Secondly digital code differs from analog because it resists degradation far more effectively. Electrical systems, like everything else, are subject to entropy. Every circuit has in it random fluctuations. Computers are not wondrously free of such static. Minor fluxes are a big problem in analog coding because the locus of information is in tiny incremental differences in the amplitude of waves, which the random fluxes in circuits can easily affect. In the absolute, digital systems are equally subject to noise, but the locus of information is in the basic energy state, not in small changes of that state. When the significant point is simply whether a circuit is on or off, it allows for a huge threshold before an intrusive fluctuation will become significant, making a circuit that is "on" appear to be "off" or vice versa.

To construct an example, consider a binary test for whether or not it is raining: looking out my apartment window to see if the sidewalk is wet or dry. This test is subject to noise -- perhaps in this case we should call it "splash." During the summer window air-conditioners in the building adjacent condense water on hot, humid days, splotching the sidewalk. Also on the road on the other side there is a low spot where water collects from a leaky hydrant and occasionally passing cars splash it onto the sidewalk. Like the noise in the electrical system, extraneous wetness sometimes partially covers a dry pavement. This rarely confuses my binary test, however, because I establish a threshold -- it is raining if the sidewalk is fully, uniformly wet and it is not raining if the sidewalk is dry, or partially splotched from random sources of water. Given the substantial threshold possible in a binary system, very, very rarely will electrical noise cause the misreading of a bit of information.

In sum, in comparison to analog coding, digital code registers values that are attributes of the thing being coded, not likenesses to it, and those values, once coded, will be remarkably resistant to error or degradation. These characteristics make digital code immensely useful in processes of communication and control.

Digitization and Communication
Digital code records samples of phenomena, not analogies to them, and it does so by techniques that are remarkably stable and accurate. By themselves, these characteristics may not seem so extraordinary. But put in context, the context of human use, they have very significant effects on the computer as a communication system. Whatever the medium, in order to communicate people need to be able to produce and reproduce information, to store it, to transmit it, to select among it, and to process it intelligently in the course of action. These five areas determine the relative historic value of different communication techniques. Reproduction, storage, transmission, selection, intelligent action: communication techniques that perform these functions well serve human needs well. Because digital coding registers samples of things and because it resists error and degradation, it has interesting effects in each of these five areas. These effects will determine how the computer as a system can contribute to our unfolding cultural history.

We begin with the problem of producing and reproducing information. What sort of information can one produce with a typical analog medium, audio tape, for instance? The answer defines a wide range of matters -- anything that can be recorded through an electromagnetic analog to sound within certain frequency ranges -- an aria but not a painting, a speech but not a balance sheet. The analog techniques used in the audio system must be closely coupled to the phenomena they record so that the way they modulate electromagnetic waves is precisely analogous to the particular wave patterns they are recording. To use the audio system to record images or the financial transactions of a bank, complex and careful adjustments need to be made in it, radical adjustments that convert the audio system into something quite different. Here the constraints of the analog medium limit the sort of information the system can record. With the digital system, we can produce a much more flexible range of information. As a result, digital coding can absorb both the analog media for carrying information in energy and many of the more traditional media that carry information in matter. For instance, the most familiar digital application now is word processing, enabling people to manipulate electronically the material system of writing with far greater flexibility, precision, and ease that traditional means have availed. In due course, anything that we can represent with a symbolically coded sample, we can record in a digital system.

It is not a trivial task to implement this potentiality. But it is inexorably happening. The first wave of computer uses involved diverse numerical applications. The microcomputer extended these and added extensive textual applications. Recently software designers have incorporated two-dimensional graphics into many programs for general use and three-dimensional imaging for special needs. Supercomputers have begun to record vast samplings of extremely complex phenomena that were simply beyond the ken of analog media -- climate change and molecular structures, for instance. With compact discs, the audio industries have developed and marketed the digital recording of sound, which is fast being incorporated into computing systems. The television and computing industries together are rapidly generating digital systems for producing and recording moving images. Techniques for sampling nearly all the forms of information and capturing them in digital code are quickly developing. In its basic sense, the concept of "multimedia" is this practice of integrating in one system all forms of producible information. When we speak of the computer enveloping other media and incorporating them into itself, we mean the capacity, unique to digital coding, to produce and reproduce many different forms of recordable information. Multimedia implements this capacity.

The difficulties in implementing multimedia are not primarily "technical," in the layman's sense of the term. Ordinarily we think that the technical problem lies in designing an apparatus to accomplish a novel purpose. In many areas, making the apparatus is relatively simple, and it can be done in numerous different ways. What is difficult is setting a controlling standard that will establish agreement on which one of the possible ways to design the apparatus will be the one put into common use. This is in part a question of technical standards -- for instance, what sampling rates will be standard for digitally encoded sound or what screen resolution will be standard for digital high-definition television (HDTV)? But the problems of controlling standards goes far beyond the domain of technical standards -- long established branches of law and language are at stake as well.

Thus, the production and reproduction of information is not simply a technical process. It is a process controlled by law and driven by incentives. Digital coding of information will affect these domains as well. For instance, copyright makes sense in a system in which people locate information in material objects -- copying consists in expending the energy to implant the information in matter, preeminently by putting ink on a page. Copying information that is located in matter is a laborious, error-prone process, subject to legal processes. Recording and reproducing information that is located in energy has very different characteristics. It becomes extremely inexpensive, with the result that it can be done ad hoc by anyone who possesses easily available, inexpensive tools. Already, spontaneous reproduction through analog means, such as photocopying and audio and video tape, has put considerable stress on laws pertaining to the right to copy. The broadcast industries have had to develop novel ways to realize economic benefit from cultural works, ways that turn less on the right to copy and more on the right to use a work.

With digital coding the reproduction of material becomes even faster, cheaper, and vastly more accurate than it does with analog electronic media. Once something has been sampled and captured in digital code, the idea of a copy of that sample ceases to make much sense. The copy is not really a copy, but a second instance of the original. The computer radically changes the conditions bearing on the reproduction of information and ideas. Once the infrastructure is in place, the reproduction of materials has a negligible cost with respect to materials, work, or quality. In principle, in a digitally encoded culture, anyone can have instances of anything they wish without added cost to the system. It will require an elaborate process of technical, social, and legal development to achieve actualize such potentialities.

Digital coding will also transform the problem of storing information. Librarians concerned with the preservation of materials traditionally attend closely to the durability of paper and its possible substitutes. The key question they ask is: "How long will it last?" This makes a lot of sense as long as the information is located in matter. If the paper will quickly degrade, the cultural community will soon need to reprint its materials or reproduce them on some alternative material such as microfiche. The shelf-life of all this is important as each cycle of reproduction is very costly, as well as an occasion for material to be lost and errors in reproduction to creep in. With digitally coded materials, shelf-life remains limited, but the costs of reproduction and the likelihood of errors arising from reproduction drastically declines. Hence, the keepers of the heritage need to rethink the standard principles of storage and preservation. Continuous reproduction can make the quest for durability unnecessary. Since reproduction is very cheap and very accurate, the problem is not one of finding the most enduring materials and keeping them as stable as possible. Rather the problem becomes one of regularly refreshing the energy-states in which the information is located and making sure that it is scattered in enough separate instances that a catastrophic failure in one instance would not obliterate the heritage.

Other, more novel problems of storage also arise. With respect to information located in material objects, we naturally store materials in institutions adapted to the attributes of the objects. Thus we use libraries for books and museums for paintings and artifacts. Much intellectual specialization arises because people need specific skills to work effectively in these different collection of material resources. Insofar as we can record all these resources in digital code, we will store them in one, comprehensive system and we will thereby diminish in power many objective goads to intellectual specialization.

As digital coding makes information easier to store with much diminished threat of loss, so too it improves our ability to transmit information. Transportation costs and limitations have long been a significant determinant of communication capacities. Through the twentieth century, techniques of coding information in energy have greatly reduced the costs and limits on its transmission. With the substitution of digital for analog coding, these developments are extending far further as we enter the twenty-first century. Analog systems using energy as the medium have developed two major principles: point-to-point circuit switching, as through the telephone, and the use of wide information channels in broadband transmissions, as through radio and television broadcasting. Digital systems are combining and unifying these two principles, allowing the links between point-to-point switched circuits to be wide information channels, creating a single transmission net of extraordinary flexibility and power.

We are already everyday users of the basic principles essential to these changes. My mother is eighty-eight and legally blind, but she can use a push-button phone with confidence and has a good head for phone numbers and thus she keeps up familial and social connections all over, in Mexico, in Canada, and around the United States. Each time she dials someone's number, she instructs the phone system to establish connections within its circuits to link her phone with that of the person she is calling. Phones code and decode voices from a very simple electrical signal that can be easily transmitted through complex switching systems and has a narrow band for coding information, one just sufficient for the low- fidelity reproduction of ordinary speech. How much traffic the phone system can bear depends on how many separate circuits it can switch together at any time and on how many separate transmissions its trunk lines can aggregate together in simultaneous calls. You'll get a busy signal if the system runs out of switches or transmission room.

Radio and television use much wider bandwidths, and they code them more intensely, with the result that their signals can be much more complex than those of the telephone. Thus radio can reproduce sound with much greater quality that the telephone, and the amount of information transmitted via television far exceeds that used in a phone conversation. The wider bandwidth, however, makes point-to-point switching in such transmissions more complicated to do without introducing noise into the signal, and without overwhelming the capacity of connecting circuits when many parallel transmissions are traveling on them simultaneously. Various properties of digital coding facilitate the combination of circuit switching with the information intensive transmissions that characterize broadband systems.

Both analog and digital systems make use of what we will call micro-time, the actuality of incredibly brief instants. For instance, radio waves fluctuate several million times per second and each fluctuation produces some of the information we hear. The higher the frequency, the more information the signal can contain, provided we can keep the receiver tuned to the proper spot upon the spectrum and provided we can minimize interference between signals and other sources of noise. Because the information bearing medium is a continuous wave, however, we find it much easier to propagate the information onto the medium at the rate it occurs at, and at which it is to be received. In contrast, when the information has been captured in digital code, it becomes much easier to make use of micro-time in more flexible ways: capture, transmission, and delivery can be separated. The pace of capture depends on the pace of the phenomenon, what we call "real time." Transmission of the binary units, the bits encoding the phenomenon, can take place in different time -- it can squeeze into each tenth of a second, or less, the information needed for one second of conversation, giving the circuit to other conversations for the remaining nine-tenths, or more, of each second. By this technique, and others like code compression and error correction, the capacity of a circuit carrying digital data can be greatly expanded.

Further, the transmission of analog data depends very closely on the particular characteristics of the transmitting medium. With the transmission of digital data, it does not matter what the transmitting medium is, provided that medium has been adapted to transmit digital code. Thus all the different electromagnetic transmission media in common use now easily transmit digital data. More importantly, new media, useless for transmitting analog information, for instance, laser light in fiber optic cable, increasingly transmit digitized information with significant gains in speed and volume, at lowered cost, and with increased dependability. The frequencies of light waves are much higher than those of electromagnetic waves. Hence, we can pack information far more densely per unit of time into light for transmission over fiberoptic cables than we can with electricity over wires or electromagnetic signals in space. The usable bandwidth is much, much wider. The higher density allows much more intense timesharing of the circuit and the greater bandwidth means that in each instant a much larger load of information will be charging through the circuit. As a result, a system is emerging in which all forms of information -- text, numerics, graphics, audio, video -- can be transmitted, switched from point-to-point, as easily as we can with the phone.

Digital coding, thus, is making possible the use of one system to produce all forms of information, to reproduce anything in the system with low cost and little loss, to provide for its indefinite storage through this process of continuous reproduction, and to transmit any element of it to any user fast and cheaply. By themselves, these developments make oodles of good information easily accessible, threatening to overwhelm the user in a vast babel of bits. These three characteristics are of a piece with each other, setting limits on what intellectual resources a culture can provide its members. But they do not, alone, make for a well developed system of communication.

Selective retrieval, enabling people to get precisely the information they want and when they need it, has always been a key problem of culture and communication. How can you get from the culture the ideas and information that you want and need? And even more perplexing, how can the culture intimate to you and everyone else what possibilities of interest it does and does not offer in the infinity of circumstances surrounding us? Retrieval is a fundamental problem of all cultures, and it is becoming an even more pressing problem with digitally coded information. It is the fourth determinant of communication effectiveness in history and the widespread digitization of information is transforming it as well.

Throughout history, major communication advances have brought with them new ways to retrieve information. The practice of citing books and articles by title and author, edition and page, rose to full significance in the era of print. The printed book, which could be distributed in many locations in identical versions, needed some logically effective technique of reference and recall, one that would work in many different places and many different times. Prior to that people referred far more vaguely to an author and an argument or thesis, and to retrieve the actual text a scholar needed to know where a specific instance was physically located, with diverse works bound together for convenience. Today, people often handle their personal libraries in this pre-print fashion, jumbling certain books together say by size, or just shelving them as they come, able to find any particular one, not by a sense of logical order, but by having a feel for where it is by some sense of spatial juxtaposition. That works for small libraries, but it spells chaos for large collections of printed books. For those, people needed to develop far more systematic techniques of reference and recall.

With digitally coded information, the situation is much the same: people need to master new, more powerful retrieval routines to manage the cornucopia of information. These techniques relate to two different problems in the use of information -- exchanging information and applying ideas. In both exchanging ideas and applying them to problems, people need to retrieve information selectively. Exchanging materials is somewhat similar to the phenomena of point-to-point switched circuits while applying them is related to finding a station or channel in broadcast communication. Exchange requires the precise identification of start and end points and application requires the substantive sifting through extensive materials to select out the precise components pertinent to the problem at hand. Since the problems and prospects in each domain are rather different, let us consider each briefly in turn.

Our means for managing the exchange of information have already been heavily influenced by characteristics of digital coding, at least insofar as digital coding involves discrete units, as distinct from continuous waves. For instance, integer numbers are a system of digital entities: each number is discrete, autonomous, separate from any other. So too is the alphabet, which is a more restricted set of discrete elements, most simplistically twenty-six, but preferably 256, if we take extended ASCII code as the norm. Long before computers, people became adept at using numbers and letters to assign precise locators to all sorts of objects, persons, phones, buildings, accounts, parts, and so on.83 Implementation of these coding principles in digital computers enhances our capacity to manage them greatly, extending the scope, precision, and speed of the process. In substance, the problem of addressing things so that information about them can be exchanged from point-to-point is less technical than socio- political: the problem of privacy, of censorship, of deciding what limits, if any, to place on the reach of possible exchange. Whenever the power to exchange information increases significantly, it brings such problems with it. The abuse of privacy thus seems to be a structural issue, occurring at the margins where new ways to manage exchange are developing. Historically, people seem to opt for accepting the benefits of new systems of information exchange, after instituting measures to ensure that they will not be used to subvert personal security and integrity. Unfortunately, this trade-off has not always been benign as the tragic abuses of totalitarian regimes of right and left repeatedly demonstrate. As computers make it possible to exchange information that was formerly "private," easily kept to oneself, we will need to face up to difficult issues of defining limits and controlling abuses.

Retrieval that involves sifting, selecting, and applying ideas presents different problems and opportunities. Our existing techniques for doing this involve time-consuming secondary processing of materials -- indexing books, abstracting articles, cataloguing things under key words and subject headings, adding captions to pictures and tables, annotating works with cross- references and footnotes. Digital coding makes these practices more effective in three significant ways. First it facilitates the processing by creating tools to help people to index, abstract, caption, and catalogue their culture. This presents us incremental gains. Second, it makes many traditional references, which had been unidirectional from one work to another, usefully bi- directional. Only where very special indexes have been laboriously developed can I go into a library and ask for a list of works that cite a passage that specially interests me. In a digital environment, the electronic reference that implements a note will point both ways, something that will make traditional references useful in powerful new ways. Third, traditional references implemented digitally will save users much time and energy, for following out a reference will be nearly instantaneous. Currently it is often hard to maintain a train of thought in following a reference as one needs to go off to the library or bookstore, perhaps having to wait weeks for a work to arrive from a distance. Digitally coded links will be fast and transparent. Together, these three changes will significantly enhance traditional resources for the reflective retrieval of ideas and the application of them to our controlling purposes.

In addition, new retrieval resources are under development. These require no intelligent pre-processing of materials aside from the capture of them in digital code. Instead, the end-user of the material specifies criteria of interest, and the system matches materials in it against these criteria, showing the resultant possibilities and allowing the user to further winnow the results, should that be necessary. These principles have been most fully developed with respect to the retrieval of textual materials. Their novelty still engenders some confusion, and many people, among them even professional librarians, misuse the concept of "full-text retrieval." Thus some think it simply means retrieving for an inquirer the full text of a document, rather than an abstract of it. More properly it means conducting the search for matches to an inquirer's criteria of interest against the full-text of everything in a collection, rather than against a list of keywords. Techniques for such full-text retrieval are becoming both sophisticated and fast, and users can apply them to both the flow of current information generated through correspondence, calls, and news, as well as to libraries of accumulated information.

Techniques of search and retrieval have historically developed far more fully with respect to text than with other forms of information. Up to now, we use text to catalogue most other forms -- maps, pictures, numeric tables, films, recordings, and so on. Yet text processing is not the only form of intelligent recall and retrieval that we can do. We can often find our way to places with a visual- spatial memory that is much more effective that verbally forming a set of directions for ourselves. We associate both moods and ideas with various sounds and melodies and even colors and places. All this suggests that beyond full-text retrieval, there lies the domain of "non-text retrieval." In non-text retrieval we might point to a geometric relationship and request the computer to search a graphic database for other instances of the similar relation or play a chord and have the system call up musical compositions in which it occurs. Non-text retrieval should in principle be possible with digitally coded information, but for the most part it is a possibility that awaits development.

One area in which non-text retrieval has been underway for some time, however, gives an idea of its potential power -- statistical processing. Statistics can be thought of as a numeric system for selecting and retrieving information that allows for judgments of significance and relevance that are very hard by textual means alone. Also, the ability to zoom-in and zoom-out to different levels of detail on graphical materials such as maps, diagrams, and photos provides substantial non-text retrieval capacities. In general, digital code enables us to capture and link different kinds of information pertinent to complex phenomena and to represent their interactions in ways that we can see or hear, using those senses to select directly between combinations. All sorts of complex controls work this way, especially in simulation systems and innumerable computer games.

These variations on non-text retrieval really carry us into consideration of the fifth area in which digital coding is deeply influencing our culture -- the intelligent processing of information. For the most part, up to the twentieth century, communication tools used external artifacts to extend the memory, while leaving the intelligent processing of ideas to take place almost exclusively inside the human body and brain. Through cultural history, people have accumulated vast stores of memory projected outside themselves into man-made objects. Despite all that externalization of memory, the possible agents for the key verbs describing intelligent operations on information and ideas are still almost exclusively human person -- perceiving, sensing, thinking, correlating, inferring, deducing, concluding, and so on. With the computer, man-made objects are becoming useful in performing these intelligent operations.

Memory, to be meaningful, must ultimately return to a sentient human mind -- a library unread is not a culture preserved. In externalizing memory into material objects, humans have not alienated memory from ourselves, but enhanced our capacity to remember by transferring parts of the task to objects that we make and manage. So too, in externalizing intellectual activity, we do not entirely alienate it from ourselves. Instead we compensate for limitations, strengthen capacities for demanding operations, and enhance attention, precision, finesse, or speed.

To understand how the computer is accelerating the transfer of intelligence to external tools, it is important to realize that this is not a sudden novelty in our culture. We perceive the world with our senses and prepare it for thought: through most of history, people did this without the aid of instruments. That began to change some centuries ago. We can interpret the rise of modern science as the intellectual fruits of externalizing capacities for perception into instruments of observation. Clocks and chronometers permitted people to perceive time with ever greater precision. The telescope and microscope enhanced the human capacity to see distances and details. The thermometer lent accuracy to our capacity to perceive differences of hot and cold. Exact scales and rules and other measures, tuning forks, prisms, filters, balances, samples, gages, a wondrous panoply of instruments, allowed inquiring minds to develop the empirical base of observation upon which they built our stock of scientific understanding.

By working with digitally coded information, instrument designers are extending the power of perception greatly. The unmanned space-probes reporting on the solar system have perhaps been the most dramatic of these extensions, with wondrous photographs and other readings radioed back as masses of digital code. Not since the invention of the telescope has our ability to perceive the universe around us so leaped forward. But digital read-outs are all around us with the computer creeping into all sorts of mundane tools, enhancing our capacity to track and control their use. For many decades car instrumentation, for instance, was very stable, consisting of a few analog gages that indicated the car's speed and possibly the RPM's of the engine, while additionally giving key hints about the state of the car's fuel, coolant, engine oil, and electrical system. That's fast changing now with digital sensors in new and old places giving a much more exact picture of the car's condition of operation, with an onboard computer relating readings to one another -- "it's getting pretty close to empty" gives way to "range remaining fifteen miles." The computer will greatly extend the reach and accuracy of instrumentation as people apply it with increasing effects to small matters and large.

With the computer, people can externalize into their instruments more than their powers of perception. When Edison claimed that "genius is one percent inspiration and ninety-nine percent perspiration," he probably thought that the human capacity for both inspiration and perspiration were basically fixed, and by perspiration he had in mind the laborious calculations needed to test speculative insight, separating good from bad. Digital systems do not do away with the need for perspiration, but they extend what we can accomplish with a given amount of it. Most forms of calculation, correlation, combination, and connection that people can make, computers can help them make better. They can expand our abilities to sort, order, rank, and select. Even this process of externalizing powers of calculation is not entirely new historically, as one who has worked with a slide rule will realize, but it is being vastly increased. The consequences are likely to be very great.

Many people think that numeric calculation is the peculiar domain for computers, but their reach goes far beyond numbers. The computer can operate on anything that in some meaningful way can be represented in digital code through an organized data structure. And any operation that can be accurately described within the compass of binary logic -- AND, OR, NOT -- the computer can perform. Let us leave as moot whether people can, or should, or ever will, externalize into tools that one percent of their genius -- inspiration. They are externalizing in all sorts of ways that other ninety-nine percent, amplifying greatly their powers to calculate and control objects of their attention. Even if artificial intelligence, in the sense of the computer being an autonomous rational agent, is not soon coming to pass, if ever, AI, in the sense of amplified intelligence, is rapidly emerging all about us. We need to come to terms with its implications.

This, then, is the computer. It is the representation of our culture in digital code and the development of all the cultural possibilities that result. The computer makes cultural work easier to produce and reproduce, to preserve, to transmit, potentially accelerating intellectual attainment and opening cultural access in unprecedented ways. The computer greatly augments human powers of selection, memory, perception, and calculation, potentially amplifying the intelligence that each and all can bring to bear upon the panoply of questions that life puts to them. We turn to the implications of this computer for the activity of education.

Table of Contents

Chapter 3