here is little talk in America these days of bold new public initiatives; public money is scarce, and faith in public remedy even scarcer. One notable exception is new technology and education. Bill Clinton's challenge to connect all of America's schools to digital networks by the end of the 1990s is the only initiative today that echoes, if only faintly, John F. Kennedy's call to put an American on the moon by the end of the 1960s. Like the moon shot, linking America's classrooms to computer networks appeals to a technological nationalism that seems beyond partisan politics: Everyone—almost everyone—likes the idea of putting the U.S. first in the race to the future. Thus in the same legislation widely heralded as deregulating telecommunications, the Republican Congress and President Clinton were able this year to agree on regulatory requirements for universal service that for the first time include affordable connections for schools.

The New Media
and Learning

With this issue we inaugurate a series of articles on the new media and learning, drawn from a conference sponsored by The American Prospect on June 4th at the MIT Media Laboratory.

The aim of the conference and the series is to explore whether the new technologies offer genuine promise for improvements in learning or are merely a diversion from the real problems of education, and to ask what approaches to policy and the new technologies hold the most promise. In addition to the authors of articles in this issue, the conference featured:

• Congressman Edward Markey of Massachusetts on why the Federal Communications Commission should adopt an "e-rate" under the Telecommunications Act of 1996 that would make a basic level of internet access free to schools;
• Mitchell Kapor, who served on the President's National Information Infrastructure Advisory Council before resigning in protest, on what went wrong with the NII initiative;
• Seymour Papert on the use of computers for fundamental change in education;
• Sherry Turkle on how learning about computers may affect our thinking about other things; and
• Howard Gardner and Shirley Veenema on multimedia and new ideas about cognition and learning.

Audiotapes of the conference are available by calling 1-800-872-0162.

Support for both the conference and publication of these articles comes from the Spencer Foundation of Chicago.

Yet past efforts to improve education with better technology have generally not lived up to the promises made for them. In the eyes of skeptics, the current enthusiasm for computers is the triumph of hope over experience—or worse, it reflects a persistent infatuation with technological fixes for deeply rooted social problems. It would be a mistake, however, to dismiss the new initiatives on the basis of such a reading of the past. The new media are different from the earlier technologies, even from computers as they were introduced into education, and these differences improve the odds of substantial change. The computer revolution of earlier decades has now turned into a communications revolution and opened up important new possibilities for learning. The new media, moreover, are becoming essential to intellectual and artistic expression and scientific work. As the entire world of communication and knowledge is transformed, it becomes inconceivable to leave education out.

Of course, computers are now integral to work of all kinds, and public support for educational technology reflects an appreciation of that inescapable fact. Many parents want the schools to use computers for the same reason that often influences their purchases at home: They believe that computers will help prepare children for good jobs and careers. In fact, workers with computer skills do enjoy higher earnings. Instead of deploring the interest in computers—"I know a false god when I see one," the critic Neil Postman writes of computers in his recent book, The End of Education—reformers should regard the popular support for new technology as an opportunity for positive change.

The question is what form innovation may take. Some critics—such as Lewis J. Perelman, the author of School's Out, a 1992 book popular in Newt Gingrich's high-tech, free market circles—believe that the new technology demands the end of school as we know it. The new media and schooling are incompatible, they say, and schooling must go. This is a setup for failure; Americans are not ready to abandon the very idea of school, nor should they. But there are important changes in schools worth making, some of which have been on the agenda of reformers ever since progressive educators first proposed them early in the twentieth century. Ironically, the continued diffusion and evolution of the new technologies may finally help to bring those reforms about.

THE RIPENING OF TECHNOLOGY

Forecasts of technological change often fail, Anthony Oettinger observes, because an innovation is not yet "ripe." Failed predictions may convince many people it will never work, but then it ripens—its costs fall, its limitations are overcome, it suddenly matches the demands of a market or the needs of an institution—and everything changes.

The history of education in the twentieth century is littered with mistaken forecasts of technological revolutions in education. In 1913, Thomas Edison predicted that books would "soon be obsolete in the schools" because of motion pictures. Similar predictions of epochal change in education accompanied the diffusion of radio in the 1920s and '30s and television in the 1950s. In Teachers and Machines, published in 1986, the educational historian Larry Cuban argues that these expectations were repeatedly disappointed, despite effort and investment, not for the reasons that advocates usually cited—poor implementation, insufficient money, resistance by teachers—but because of a more fundamental obstacle: the logic of the classroom.

Every day teachers confront enormous problems in accomplishing their objectives, including just managing their students. "The tools that teachers have added to their repertoire over time (e.g., chalkboard and textbooks) have been simple, durable, flexible, and responsive to teacher-defined problems in meeting the demands of daily instruction," Cuban observes. In contrast, movies, radio, and television typically required a lot of setup work and advance scheduling and did not necessarily mesh with lesson plans. Administrators and reformers also initiated change from the top down without engaging teachers as active participants. As a result, except for a few enthusiasts, teachers have tended to use movies and broadcast media only to supplement regular classes and break up routines of instruction.

Computers originally seemed destined to go through the same cycle of enthusiasm and disappointment, eventually to be relegated to the margins of education. Broadly speaking, educational computing has gone through three phases. In the first, from the mid-1950s to the early 1980s, the principal interests were the development of computer-assisted instruction (CAI) and the teaching of computer programming. Though often ridiculed as mere "electronic flashcards," CAI had a more sophisticated conception as an approach that could customize instruction according to individual needs and allow students to pace themselves. After stirring initial excitement, the approach drew increasing criticism in the 1960s and '70s and had relatively little impact on the educational mainstream. As of 1980, according to a review by the Educational Testing Service, most computer education in secondary schools "consisted primarily of teaching white middle-class males to write programs in the BASIC language."

The impetus for CAI originated primarily outside of the schools. As was generally true of computers and computer science, the military was the chief sponsor of research, contributing three-fourths of all the research funds for educational technology up through the early 1980s. Perhaps the most highly publicized project that grew out of defense research was PLATO ("Programmed Logic for Automatic Teaching Operations"), based at the University of Illinois and owned by the Control Data Corporation, which hoped to build a business educating students all over the world from its central computers. By 1981, Control Data had 115 "learning centers" in the United States, making it the largest computer-based instructional system. Because of its cost, however, PLATO was rarely used by schools; the orientation was chiefly toward technical training. Control Data ultimately lost nearly a billion dollars on PLATO, a failure that became emblematic of dashed hopes in computer-based education.

The second phase of development began roughly in the early 1980s with the spread of personal computers, graphical user interfaces, and general applications software. Between 1981 and 1991, the proportion of schools with computers rose from 18 percent to 98 percent, and the number of students per computer fell from 125 to 18. Instead of just offering specialized courses in programming, schools incorporated computing into many subjects and activities. Still, computers were typically located only in special laboratories (as most still are today), and student time on computers averaged only slightly more than an hour per week or about 4 percent of instructional time. At the secondary level, most such instruction took the form of courses in "computer literacy"; at the primary level, computers were typically used for "integrated learning systems" that provided drill-and-practice in basic skills. None of this much affected the core curriculum or general educational experience. In an article called "Computers Meet Classroom; Classroom Wins," Cuban could still argue in 1993 that computers were likely to continue to have limited impact and might be expected to become more significant only in primary schools because of their greater flexibility in classroom structure.

By this second phase, however, computers were already deviating significantly from the pattern followed by earlier technologies. Much of the interest in computers was coming bottom-up from teachers and students, not merely top-down from administrators. PCs and general applications software made computing more flexible and easily adapted to different subjects and styles of teaching. Unlike motion pictures, radio, and TV, computers were far more susceptible to both student-centered and teacher-defined activities. And as computers began to be used for communication and the development of new learning communities, they took on an entirely different character from the earlier technologies.

These possibilities are all being extended in a third phase of development that has begun in the 1990s with the advent of multimedia, the explosive growth of the Internet and the World Wide Web, and the transformation of computing from a segregated activity into a ubiquitous part of the everyday work, school, and home environment. If, as Cuban argues, teachers adopt tools that are "simple, durable, flexible, and responsive to teacher-defined problems in meeting the demands of daily instruction," computers now increasingly meet those minimum requirements—but, obviously, they can also do much more.

THE COMMUNICATION REVOLUTION IN LEARNING

To some critics, the problem with computers has not been the obstacles to their adoption, but the effects on education if they are adopted. Computer-based education, critics have worried, would value "calculation" and "instrumental reason" over the emotional, aesthetic, and critical faculties. It would mechanize education, reduce its personal character, and lead students to become engrossed in relations with machines instead of developing relations with teachers and other students.

The first phase of educational computing with its emphasis on teaching machines gave some reason for this concern. But alongside the model of the computer as tutor there grew up another paradigm of educational computing that emphasized creative, student-centered learning. As the former reflected the didactic tradition of education, so the latter reflected the approach advocated by John Dewey and other exponents of progressive education, which views students as actively shaping their own understanding and teachers as facilitating that process. During the 1980s, this constructivist approach to computing, best exemplified in the work of Seymour Papert of MIT, became more prominent.

In addition, the culture that grew up around Apple's Macintosh computer offered humanist critics a more comfortable aesthetic that celebrated creativity, self-expression, and individuality—not calculation. The Mac's graphical user interface reversed the whole premise of "computer literacy"; instead of making students sophisticated enough to use computers, it made computers simple enough for students to use. The predominance of Macs in schools may have resulted from Apple's corporate strategy, but it also fit with a preference for dealing with the computer, not as an analytical engine, but as a tool useful for a variety of tasks, projects, and activities. Feared originally as an educational straightjacket, the computer turned out in many of its uses to be a new medium of expression—like writing or painting.

Of course, one virtue of the computer is that it can become, as computing pioneer Alan Kay writes, "any and all existing media, including books and musical instruments." And with the advent of multimedia, the computer has evolved into a distinctive medium that is uniquely capable of juxtaposing text, images, audio, and video. Multimedia permits an extraordinary flexibility in conveying concepts—through words, pictures, and sounds, as something that can be built or played as well as read or watched. The connections change old genres and make possible new ones. The traditional dictionary had a cumbersome and inadequate method to describe the pronunciation of words; the multimedia dictionary pronounces them. New genres, such as simulation games, are emerging that challenge the user or player to build some complex creation—a city, species, business, or world—out of some given set of resources, or that put the student into a simulated environment or through a scenario to meet a challenge or learn a skill. The computer thereby turns the passive reader into a participant; it cues the student of a need to do something, but not necessarily what to do. With multimedia the computer draws on more of the senses, and more dimensions of intelligence, enlarging the opportunity to learn for those who have been less able to learn from conventional teaching materials. And as the tools for creating multimedia become less expensive, students will make multimedia fully their own by creating work that exploits its new aesthetic and intellectual possibilities.

Multimedia has such stunning possibilities that it invites a fascination with technical virtuosity and surface effects that can become a distraction from learning. New software that combines learning and play has blurred the boundaries between them in a new hybrid variety, "edutainment." The very term expresses perfectly both the opportunity to turn education into play and the danger of learning being lost amid the games and the glitter. It is one thing to play at something; another to reflect upon it and acquire a discipline. Software that is good for play may not be good for learning in the full sense. Much educational software also just renders on a computer screen what is already available in books and merely adds gimmickry. But some uses of the new media are genuinely inspired, provocative, and engaging, and these examples suggest that that we have opened an important new chapter in the history of the imagination—and of education.

he transformation of computers into a medium of two-way communication also advances the creative and exploratory uses of the technology. Access to the Internet and the Web puts students in reach of resources and people that schools could never before provide. Even if the Internet consisted only of texts and images, it would be of immense value as it becomes the world's largest library. But it also increasingly provides access to audio and video archives, which conventional libraries generally do not offer. Hypertext links offer pathways that allow the novice to find connections among different sources, and the growing search capacities on the Web make it an increasingly powerful instrument of research.

And, of course, the Internet provides not simply published resources, but also cyberspaces—news groups and other forums for discussion; MUDs for role playing and simulation; and new learning networks that help connect students, teachers, and others for a widening variety of purposes. Electronic networks enable students and teachers to combine resources and communities and to work with one another in novel ways. Groups of students at different schools, even in different countries, work together on collaborative projects, comparing the results of environment studies or cross-cultural surveys and thereby learning not only the subject at hand but also other skills in social relationships—just the kind of learning that the early critics of teaching machines were afraid computers would stifle.

Through distance learning, both students and teachers can take courses in special subjects not locally available. In a recent article in the American Journal of Physics (December 1995), Edwin F. Taylor and Richard C. Smith—two physicists who since 1986 have been teaching online courses on relativity to a mix of students and teachers from high schools and colleges—report that they have equally good results teaching online as in person. Their first two conclusions contradict the usual expectations:

1. The computer conference setting can be personal, friendly and inclusive. The medium is largely race-neutral, location-neutral, status-neutral, age-neutral, income-neutral, disability-neutral, and would be gender-neutral except for the clue of first names. Student participation in the discussion (in part forced by our course format) is greater than in any of our face-to-face classes. Some kinds of personal warmth appear to be more freely exchanged in the absence of bodies.
2. Computer conference classes bring instruction to a range of students for whom enrollment in conventional courses is difficult or impossible. Participation can occur conveniently at any time in a busy daily schedule. Some students blossom when in front of a computer screen and accomplish tasks that they otherwise would avoid. . . .

Taylor and Smith also recognize some drawbacks in the online format—for example, students don't receive visual cues from teachers—and they are not suggesting that online instruction will replace ordinary teaching. Their online students are a special, self-selected group. But many students and teachers have special interests; schools traditionally have just had little way of meeting them.

Long before computers, progressive educators called for strengthening the contact between school and the society beyond. Computer communications make such contacts, at whatever physical distance, easier and less costly. For example, students now take electronic field trips to enter into discussions with people in specialized fields of work, to view exhibits, even to use the cameras and other physical instruments that are now being connected to the Web and that will increasingly enable students in real time to enter into events at a distance and to participate in scientific experiments.

The Web is beginning to transform the practice of scientific research as it becomes a system for publishing not only scientific literature, but also scientific data. Genetic, meteorological, geographic, and census data are readily available to be downloaded and analyzed. Some journals are linking articles, data, and bibliographies, enabling a reader to jump back from the text to the original data or sources on which they are based. With the benefit of "applets"—software programs available on the network usable for specific purposes—readers will increasingly be able to redo the analysis or to extend it. Thus, an article about supernovas can include a simulation of a supernova explosion, along with tools that allow the reader to see the simulation run under different assumptions. In short, the Web is emerging not simply as a digital library, but also as a digital laboratory—a genuinely revolutionary development in science.

n the light of these developments, the history of educational computing takes on a different significance. Evaluations of the "effects of computers," whether in education or other areas, have had a short half-life because the very nature of computing has changed so fundamentally with the development of PCs, the introduction of graphical user interfaces, the advent of multimedia, and the explosion of computer communications. Cost-benefit analyses of educational computing compared to conventional teaching have been particularly prone to obsolescence because of the declining costs of computing, especially relative to teaching. And the value of early efforts in educational computing was not measurable in the short run because the benefits did not only involve what students learned. They also involved what everyone else learned. The early efforts helped American software designers and developers as well as schools and teachers to begin climbing a learning curve far ahead of other societies. Today educational software is largely an American industry. The Internet and the Web, while global in scope, also reflect America's distinctive edge.

Some evidence suggests that current technology already offers benefits in the narrow sense of measured student learning. The "Kickstart Initiative," the final report by President Clinton's Advisory Council on the National Information Infrastructure, cites studies showing that "technology supporting instruction [has] improved student outcomes in language arts, math, social studies and science"; that "multimedia instruction—compared to more conventional approaches—[has] produced time savings of 30 percent, improved achievement and cost savings of 30 to 40 percent" and demonstrated "a direct positive link between the amount of interactivity provided and instructional effectiveness"; and that "remedial and low-achieving students" have registered "gains of 80 percent for reading and 90 percent for math when computers were used to assist in the learning process." I would not stake my life on these numbers. But even approximately equal results for computer-based education would amply justify further investment given the trajectory of costs, and in any event the more important uses of technology now extend and enliven education and discovery in ways that such studies do not capture.

The skeptics are surely right, however, that the "learning revolution," as the magazines call it, still has not had any general influence on schools. The many high-end uses of the new technology, like online courses in relativity, are appropriate for advanced secondary and college work but do not address the general needs of primary and secondary education. The market for educational software is growing rapidly, but many students, especially from middle-class families, are more likely to use it at home than at school, while students from low-income families never use it at all. New educational sites on the Internet appear daily but don't affect most classrooms. The challenge now is to go from scattered initiatives to more comprehensive changes. And that is what many reformers are trying to do by combining new technology with an educational reform agenda—one that progressives of the 1920s would have no trouble recognizing.

PROGRESSIVE EDUCATION REBORN?

Reports and commentary on education now often argue that as our current system of schooling reflects the industrial age, so we need a new approach to learning in the information age. Thus a report published in 1995 by the National Academy of Sciences, Reinventing Schools: The Technology Is Now!, says postindustrial society "calls for a new, postindustrial form of education" — one that puts students in a more central, active role in their own learning, helps them learn "to ask many questions and to devise multiple approaches to a problem" instead of forcing them to come up "with one right answer," and encourages "critical thinking, teamwork, compromise, and communication." Similarly, the Clinton administration's "Kickstart Initiative" foresees innovation that "brings the world to the classroom," "enables students to learn by doing," and "allows educators to become guides and coaches to students, rather than be 'the sage on the stage.'" On the right, Lewis J. Perelman, the author of School's Out, wants to empower students to seek out instruction individually in the electronic marketplace. While significantly different, all these proposals call for use of technology to advance student-centered, project-based approaches to learning.

To anyone familiar with the history of educational reform, such ideas will have a familiar air. For example, in the opening pages of The Child-Centered School (1928), one of the classics of progressive education, Harold Rugg and Ann Shumaker decry traditional schooling as a product of the industrial age and "mass mind." They use two photographs in the book's frontispiece to represent the contrast between "the new and the old in education." One photo shows a class of students at their desks ("Eyes front! Arms folded! Sit still!"), which Rugg and Shumaker call the old "listening" regime. A second photo shows students in small groups busily working on different projects ("Freedom! Pupil initiative! Activity! A life of happy intimacy . . ."). This was the image of the future in the 1920s, and though the tools and terminology have changed, it is still the image of the educational future that many reformers hold up today.

Perhaps the absence of acknowledgments by today's reformers is understandable. For those who claim to be anticipating a new era, old antecedents are embarrassing. Moreover, what progressive education achieved in the first half of the twentieth century—the expansion of the curriculum, addition of extracurricular activities, greater flexibility and mobility in elementary school classrooms, improved teacher training, child study teams, changes in school architecture to provide for more varied activities, and much else—is now taken for granted, and the movement is better remembered for its failings. Progressive education collapsed during the 1950s because it had lost its way and then ran into a storm of distorted charges. Taken over by professionals, it became encrusted as an ideology of the teachers colleges. From one direction, conservatives accused progressivism of subversive tendencies on the basis of its old entanglements with the left; from another direction, liberal critics accused it of promoting conformity and anti-intellectualism. Perhaps progressive education had to die to be shorn of all the extraneous baggage it had accumulated.

In their concern for active, student-centered learning and communication with the wider world, today's technological neoprogressives have revived an old and worthy tradition. And by connecting progressive ideas with computers, they may have finally found a way not only to present them in an appealing, updated form, but also to make them work. For the difficulty with the ideal of active, student-centered education was not simply the opposition it aroused, but the demands it imposed on teachers and schools. The new technology may help manage those demands.

A growing body of evidence suggests that the introduction of computers into classrooms promotes a greater emphasis on projects, with teachers acting as guides and students taking on a central role in their own learning. Alan Collins, head of educational technology at BBN Corporation, an internet services company for businesses, identifies eight major shifts that research suggests computers bring about in education—all of them moving in the direction of progressivism. Among these are a "shift from whole-class to small-group instruction" and "from lecture and recitation to coaching." When computers are introduced, Collins argues, teachers find it hard to keep students in "lockstep" and so adopt more "individualized" approaches. A study of the Apple Classrooms of Tomorrow found that teacher-led activities dropped from 70 percent in classes without computers to less than 10 percent in classes with computers, and that activities facilitated by teachers, rather than directed by them, increased from about 20 percent to 50 percent of class time. Other trends, according to Collins, include shifts "toward more engaged students," "from a competitive to a cooperative social structure," "from all students learning the same things to different students learning different things," and "from the primacy of verbal thinking to the integration of visual and verbal thinking."

The new technology alone does not determine these effects. Schools with different cultures and philosophies will make use of computers, like other tools, in different ways. A school wedded to the didactic approach can use integrated learning systems to reinforce conventional teaching methods. A constructivist approach isn't easy; it requires a great deal of institutional support. In a recent study of nine sites pursuing an educational reform agenda emphasizing "student-centered, curriculum-rich, technology-based projects," Barbara Means and her colleagues at SRI, a California research organization, found that the key factor in determining success was a coherent, schoolwide instructional vision.

One factor that in the long run may help advance this approach is cost. For the immediate future, the cost of technology is an obstacle to large-scale plans for change. The National Academy report says the technology is "now"—but, alas, the money is not. The fundamental trends, however, are implacable. The cost of labor only goes up, while the costs of computer power and telecommunications go down—steadily and sharply. Computers will become extremely cheap in the next century, and thus student-centered projects based on computers will be far less expensive than today. The obstacle to more individualized instruction and smaller classes has always been the cost of employing additional teachers. But if additional teaching comes inexpensively from computers, individualized education is more feasible. By occupying some of the students, computers can reduce the number of students teachers need to supervise at any given moment. This amounts to a reduction in effective class size. Moreover, according to Collins, unlike teachers in conventional classes, who tend to call on stronger students, teachers in classes with computers spend relatively more of their time with weaker students.

The use of computers can also help address another obstacle to change—standardized achievement tests. The new technology may encourage some change in assessment methods, but the present system will likely remain for such critical purposes as college admissions. To prepare students for those tests, schools can make use of the more didactic forms of computer-based education without organizing their whole program on that basis.

As technology may help create effectively smaller classes, so it may also strengthen the case for smaller schools. Empirical studies indicate that students in large schools take part in fewer school activities, identify less with the school, and have lower scores on achievement tests than do students in modest-sized schools. Deborah Meier, a principal in East Harlem and an advocate of smaller schools, argues that small size permits closer relations among administrators, teachers, and students and thereby fosters the kind of unified educational vision that researchers have repeatedly identified as a key to successful schools. In a small school, students are less likely to be lost amid the throng. The creation of little schools within the framework of public education makes diversity and school choice accessible on an equal and local basis.

Of course, modest-sized schools can be created from big ones without any help from technology. But the new media may help mitigate some of their shortcomings and improve the trade-offs. Many parents are concerned that smaller schools may not be able to offer as great a diversity of courses. Computer learning networks can provide them. As a small school can create a strong local learning community, so online communities can help students widen their contacts and affiliations—offering the best of both worlds. And just as computers help small businesses by enabling them to perform complex services that used to require large bureaucracies, so the new technology can help small schools manage their affairs.

Computers and computer communications may also have particular value for alleviating some sources of inequality. Computer communications enable people with disabilities to gain access to resources otherwise unavailable and to take part in groups without hindrance or stigma. Similarly, computer networks improve access to educational resources for those in small communities and rural areas. For the same reason, they may be especially valuable for those who seek to continue their education while working at a job. Members of racial and ethnic minorities may learn more through interactive software or online services because they sense no stigma or disapproval. Social psychologists Lee Sproull and Sara Kiesler have found in experimental research that lower-status participants in electronic discussions are less inhibited and more likely to speak up than when communicating face-to-face. Thus, the very groups that now lag in the use of computers and computer communications may especially benefit from access to them.

Of course, nothing guarantees that computers will be used for progressive purposes. Conservatives would like nothing better than to use the technological limitations of schools as a rationale for privatizing the schools or substituting a kind of high-tech home schooling. Inevitably, choices about technology become entangled in larger choices about politics.

THE POLITICS OF THE COMPUTER TRANSITION

The use of computers and the Internet is now expanding rapidly, but with marked disparities between rich and poor school districts. In 1995, according to a U.S. Department of Education survey, half of public schools had at least some internet access, up from 35 percent in 1994; and while only 9 percent of classrooms were connected, that was up from 3 percent a year earlier. A student in an affluent community is roughly twice as likely as one in a poor community to attend a school with internet access.

Students are now using computers differently from in the past. The computer laboratory, typically set up for computer literacy and programming courses, is evolving into general-purpose computer work area where students can do projects of all kinds, including internet work. For most schools, according to the SRI study, concentrating computers in a laboratory is still the most efficient way to provide maximum access to a limited number of machines; distributing computers through classrooms optimally requires at least 6 to 8 computers for a class of 25 to 35 students. At present rates of growth, the average school in the United States should approach that roughly one-to-four ratio around the turn of the century.

The 1996 Telecommunications Act made it a matter of national policy that schools receive "affordable" access to telecommunications. The legislation sets a new precedent by linking communications policy and education; the Federal Communications Commission (FCC) will now determine the exact obligations of the telecommunications industry in the subsidy of school connections. Cost estimates vary, depending on the assumed level of access and whether the estimates include the cost of the computers themselves. There is a wide range of possibilities between providing a school with a dial-up account for one computer and creating a high-bandwidth network linking computers on every student's and teacher's desktop. As platforms change, the standard for universal school service is likely to evolve. In the near term, if every school were to have a local area network, 60 new computers, a router, and a local server—with every district, or 4 to 6 schools, having a high-bandwidth (T-1) connection to the Internet—such a system, according to a 1994 Department of Education study, would run between $9 billion and $22 billion in onetime costs (about half of which would pay for the initial purchase of computers) and $1.75 billion to $4.61 billion annually. Of the annual costs, roughly a quarter would go for the telecommunications lines and internet service; so if only those are cross-subsidized, the schools would still be left with very large costs indeed. The FCC is exploring whether to set aside spectrum to provide schools wireless connections, which could particularly help to minimize indirect costs, such as asbestos removal, from retrofitting school buildings.

Schools in affluent districts may be able to raise these costs in local taxes. Some districts (such as my own in Princeton) have already benefited from partnerships with universities and businesses in making the transition to networked schools. But even with the most supportive telecommunications policies and voluntary support, schools in low-income communities will almost certainly need additional financing from the states or federal government to shoulder the required investments. Otherwise there seems little prospect that inequalities among schools or communities will soon diminish.

In principle, the falling cost of computers and bandwidth should increase opportunities for lower-income groups and communities. So far, however, possession of computers (and of network connections) has continued to grow more rapidly among high-income than among low-income households, thus widening the disparities, according to a recent RAND analysis of changes in computer ownership from 1989 to 1993. Eventually, if histories of the telephone, radio, and television are accurate precedents for the computer, the diffusion of computer communications will tend toward universality. But the transition could take a long time—decades. In the meantime, many groups will be disconnected from a communication network of growing value, and we will all lose the benefit of "network externalities"—the increased value of a network to each user as others are connected. For example, the value of computer communications to schools increases as teachers are able to reach more of the parents of their students and as more students can use the systems from home. Hence the rationale for using public policy to accelerate the transition to universal electronic communication for both community institutions and households in low-income areas. Schools and libraries seem to be the only institutions for which such support is now politically obtainable; they may take on larger significance by opening up access for families as well as the children themselves.

The schools need not only cheap connectivity, but also low-cost access to content. Currently, the Web provides free access to enormous amounts of information, but many sources are likely to be available only on a fee basis as commercial transactions become customary. To be sure, governmental sources, many nonprofit organizations, and schools themselves will continue to offer publications and other resources for free. So will companies interested in fostering long-term business. But many journals and other sources will be available only at a price, and students will lose access to such sources unless there are affordable site licenses and other arrangements for schools. The development of online libraries providing free access to work in the public domain and low-cost access to copyrighted material of educational value should be a priority for both public and philanthropic support. What Carnegie did a hundred years ago can now be accomplished more efficiently, for the entire world.

s the cost of computers declines, schools will likely move from computer laboratories to desktop computers distributed through classrooms and then, in a further stage, toward more mobile forms of computing. Voice activation will often obviate the need to sit down at a keyboard; wireless will liberate the networked computer from its place on the desktop. Increasingly, students and teachers may scarcely even think of computing as a distinctive activity. Drawing an analogy with electricity, Marc Weiser of Xerox PARC (Palo Alto Research Center) suggests that a truly powerful technology "disappears" from awareness and that the computer of the future will assume diverse forms (tablets, pads, badges, whiteboards) and be a ubiquitous but taken-for-granted part of the built environment. Weiser envisions computers becoming so cheap that some would be left around like scratch pads, the very opposite of the "personal" computer. Thus, a computerized classroom in the future might not have students sitting at keyboards and monitors; it could be a classroom where computing was both ubiquitous and incidental, allowing students freedom to play and work with one another while using the technology's extraordinary capacities.

But, of course, none of this will answer the truly important questions about learning. Here Postman and the other skeptics are right. Ultimately, the qualities of education that we care most about are not technological; they are matters of educational philosophy and practice and in turn depend on broader moral and political judgments. In thinking about education, we ought not to be preoccupied with computers at all, and if the technological transition is successful, we will not be. Because of all they make possible, we must make computers part of education. Then they should "disappear."