Social Impacts of the Transition

by R. Stephen Berry,
Department of Chemistry and the James Franck Institute
The University of Chicago

This chapter deals with some of the social changes that the transition from paper may cause from the perspective of a scientist. The discussion is in five sections: broad impacts not specific to the sciences but relevant to them; impacts, largely favorable, on the way scientific research is conducted; impacts on the institutions and institutional procedures in which science is conducted; some potentially undesirable effects; and, finally, effects on inter-institutional and especially international activities.

The advent of computers and the electronic disposition of information creates a paradigm of how technology changes with the change of generations. The generation or two of forerunners, inventors and first developers of computers consist primarily of only a tiny elite for whom the new medium seems the obvious new force in the developed world. Two generations after them comes a cadre for whom the medium is an obvious, inevitable part of the world, as much taken for granted as telephones and automobiles. The grandchildren of the pioneers of computers learn to type on computer keyboards before they learn to write with a pencil. This level of assimilation has some of the most obvious and pervasive social ramifications associated with the new medium, and is the first topic of this chapter.

Another development requiring new social adaptations is the necessity of reconciling open and protected access to on-line materials. Defenders of open access argue for the importance of activities in the public good; those defending protected access argue that we must protect intellectual property rights of individuals. This conflict tends to be argued and eventually settled in a legal venue in the United States, but any legal resolution should reflect the way the society resolves its conflicts over social equity. This topic is included in the third section, which also reviews the potentially undesirable or problematic effects of the transition. The last section addresses interactions of individuals in different institutions, particularly among individuals in different nations. The interpersonal contacts enabled by networks have consequences far beyond their manifest content.

The Transition of Generations: "I can show you how, Mommy!"

The first aspect of the transition is the change from computers being special tools of the "nerd elite" to their being an integral part of the daily experience of many, soon including most young people in the developed world. Today, in typical research groups in the research-oriented universities of the developed world, there is a general expectation that the graduate students and undergraduates, and possibly the postdoctoral associates, are the computer experts, not the senior faculty who are the nominal leaders of those groups. A common household pattern has the young teenager showing parents how to use email and the World Wide Web. Can we think of any prior situation of our society that would be comparable? Movements have occurred that captivated youth, but not of a sort that parents wanted and even needed to learn and that were generally considered benefactions. Although adolescents and occasionally children could, once in a while, take an early automobile out for a spin, the first cars were the toys of a mature generation. When our society assimilated the telephone, radio and television, it was with the enthusiastic participation by almost all the generations able to use them.

Most children's first experience with computers is with computer games. Three-year-olds learn to manipulate a "mouse and clicker" to play interactively with animated screen images. Many children learn letters and numbers from computers. By age 5, many children now read text on screens. It would not be surprising or unlikely if computers turn out to be effective stimulators of literacy, far surpassing any other mechanical tools we have ever tried. It will be a challenge, perhaps a subject for doctoral theses in psychology and education, to find how computers affect the capabilities of children to read, write and do elementary mathematics. The interactive character of computer games makes them more captivating to many children than television, because they provide active, not passive, entertainment. It would be no surprise if, in even 5 or 10 years, computers become far more popular than noninteractive television shows, videotapes and videodisks. After all, the designers of computer games are still learning how to make their products more and more engaging; shows on television of whatever kind can probably enhance their attractiveness only through their content or through devices such as the now-discarded broadcast lotteries that involve viewer or listener involvement. It will continue to be an even greater challenge for designers and vendors to make computer games that both fascinate and teach. Indeed, some games we consider effective will probably seem crude and naive in five, ten or fifteen years, when the new game designers are people who grew up with the early varieties we have now.

In another very few years, today's computer-wise youth will move into positions of responsibility. When that occurs, we can expect a transition in attitudes toward computers and electronically-available information. The normal expectation of any new employee will include these, at very least: communication within a firm will include a large portion of email traffic, standard in some but by no means all firms already; records and almost any other information the person needs will be accessible by computer; and all archives will be electronic, regularly backed up and kept in protective, redundant form.

The power of computer-aided design already allows the architect or industrial designer to test many possibilities, where previously the same person could study only a very limited number of possibilities. The scientist already uses computer programs to design apparatus and experiments, e.g., optics, whether for light or for particle beams, for the special needs of each experiment. A subsequent stage of this evolution may be that we will use powerful, completely automatic optimization methods to assure that any moderately complex task will be done well. We could do this routinely for anything from planning a vacation trip to designing a house or a retirement plan. We can already do this a little with commercial programs that compute taxes. Outside the workplace as well as within it, most financial transactions will be done electronically; credit cards will be replaced, at least in part, by automatic debiting of bank accounts. These are changes that we, who grew up with the earliest generations of computers, can envision now. We can expect entirely new concepts (and their implementation) from the generations who, since infancy, have taken computers and the Internet as normal parts of daily life. For example, perhaps multisite telecommunication links might become common forms of casual communication, to be used not just for special events like teleconferences and electronic conferences.

Educational tools

We can expect that the attractiveness and sophistication of all kinds of computer activities will improve steadily for a long time, as the computer-literate generations see ways to change things that were not at all apparent to their predecessors. This will surely apply to the full range of activities, from "pure entertainment" games through coordination skills to logic-based and knowledge-based games, to full-fledged, extended educational programs. We can imagine, for example, an interactive program to learn the language and culture of another people, through modules that actively correct the learner through rewards and punishments. It is possible, and optimists believe inevitable, that we shall see educational programs such as these that will enhance the learning process far beyond the contemporary classroom experience.

Until now, simulations have been important aspects of computerized training—distinct from education; examples are driver training, flight and spacecraft simulation, and battlefield simulations of armored-vehicle warfare. In the context of education, while some very highly-praised computerized learning systems have appeared, few have been more than supplements to traditional methods, and fewer still have been incorporated into curricula. However, this certainly does not mean that nothing will come along to change things more substantially. It may mean only that such changes are unlikely until the program developers and the teachers who will select those programs come from a generation that has grown up with computers as part of their lives. Interactive simulation programs could become integral tools of the science learning process, allowing students to see the consequences of different answers or approaches to problems. Still further ahead will be computer activities that lead the student into an active role in innovative thinking, by providing open-ended questions that the student must pursue by making them into more precise, answerable questions and then working toward answers. This stage is something that students, particularly pre-college students, rarely encounter, partly because of the standardization of classroom curricula, partly because so few teachers are both prepared and able, in today's classroom, to pursue such challenges. When those pupils have their own computers and expect to spend time working alone with them, it will be natural to provide them with such challenges.

The Conduct of Science: The Evolving Roles of Computers and Computation in Scientific Research

The computer's impact on the processes and progress of science has several facets.

Computers influence how scientists communicate with each other and even with their machines. Second, they influence how scientists do what they do. Third, scientists must adapt what they do to constraints based on exogenous concerns about computers and their use. The communication of results, i.e., the future of distribution and publication of scientific information, is the topic of several other chapters. Here we shall simply assume that scientific communication is likely to take the form of some kind of "Giant Archive," with capacity for searching for virtually any desideratum, for linking from any item to related items, for comparing content of different items, and probably much more. The interfacing of scientific-scholarly-educational needs and practices with those of the commercial and entertainment worlds also forms a topic addressed in other chapters of this work. Here we focus primarily on the practices and processes of scientists, and how we might see these evolve as we integrate computers more and more into those processes.

One computer-driven change in the sociology of science that has begun quietly and accelerated is the way we use computation in the process of doing science. There are several manifestations of this change: one is in what scientists accept as a solution to a conceptual problem. For many years, "a solution" meant an analytical solution to the equations representing the problem, while a computed result was considered a particular case, an example, an illustration or a specific application of the concept or theory—and not a general solution to the problem. Now, with storable, friendly programs that can be used by people with no interest or ability in devising or even modifying those programs, a general algorithmic solution is just about as acceptable as a general solution as an analytic solution of the same set of equations and corresponding model problem. We have simply broadened our concept of what constitutes a general solution. An interesting corollary to this evolution is the experience that some scientists have already had, of discovering that it is faster and easier to solve some problems—that is, to solve the equations that represent them—by computing the solutions numerically from the original equations than by solving the equations analytically and then evaluating those solutions. This writer and his collaborators had just this experience in the 1980's, with an integral equation that could be evaluated by quadrature or solved exactly, to yield an elliptic function. The quadrature was far faster than evaluation of the elliptic function.

A second manifestation is the changed concept of "primary data." Many years ago, scale readings from primary sensors such as thermometers and voltmeters were replaced by chart recordings from amplifiers or digital-to-analog converters, removing the experimenter a step or two from what had been the primary data. Now, sensor signals go through many kinds of processing before the experimenter sees them as highly analyzed data. Sometimes an automated apparatus collects data taken under several different conditions and supplies those data to a computer that adds, subtracts, rescales and otherwise manipulates the data to eliminate or minimize "noise" and then delivers a "signal" to the experimenter. Yet we still consider the first data seen by the experimenter as "the primary data." There are many examples: the processed images from a telescope, the spectra obtained by Fourier transformation of time-varying signals, the events recorded as multiple-coincidence events when a molecule or nucleus breaks into many fragments. (In fact, in high-energy, particle physics experiments, it is still standard practice to save the vast files of information from the primary detectors.)

A third manifestation of computer-driven change in the scientific workplace is the integration of computation and analysis in the generation of solutions to scientific questions. For many years after computers came into use in chemistry and physics, one "did" analysis or one "did" computation. Of course many people did both, but as separable activities. The theoretical model came from analysis, which could then be expressed in a computer program and used to generate numerical solutions to the problem. This has evolved into a much closer integration of steps carried out on a computer with steps carried out using pen and paper. One advance that stimulated this integration is symbolic computation, which allows one to use a computer to carry out mathematical manipulations that were, heretofore, lengthy, error-prone and sometimes irrelevant to the originality and insight brought by the researcher to the problem.

Another advance is the expanded capability of computations, a notable example being animated simulations, to reveal physical relationships that become vital guides for constructing theoretical models. The simulations sometimes show physical relationships that the investigator might never have imagined without their guidance. Other chapters deal with the content we can expect or imagine in the computerized information of twenty-five years hence, but we nonetheless can add some speculations here. In the "Giant Archive," the hyperlinked database of all nonproprietary scientific information that we imagine, there will not only be text, tables, graphs, pictures and animations; we can expect holograms, still and animated, and forms of representation that allow us to envision phenomena in several dimensions. The simplest we can imagine will use time, color (hue), intensity and grayness (chroma) to indicate "distance" in dimensions other than those shown in a conventional projection from three dimensions. We can also expect to see manipulable images that are "constructed" in perhaps 6 or 10 dimensions and processed to show an image projected down to three dimensions. We will be able to move our viewpoint with a suitable set of controls, perhaps several "joysticks," through the full, multidimensional space and see the projection of the object into our selected three dimensions as we move. It would be a straightforward exercise now to construct such an image manipulator for a simple, 4-dimensional object such as a tesseract.

What does this suggest for future changes in how we will do science? We can expect more and more reliable simulation, and with that, an increasing acceptance of simulation as a valid tool. In time, not in two or three years but probably in less than twenty, scientists will accept some kinds of simulations as tests that are as valid as laboratory experiments. We can already see the forerunner of this in the occasional example of a theoretical result that turns out to be more accurate than a highly precise experimental counterpart, just as we have already seen an era when some properties of atoms and simple molecules could be computed far more accurately than they could be measured, properties such as the electron affinity of the hydrogen atom. The desire for valid, acceptable simulations will probably stimulate studies in how to evaluate and improve the robustness of models, both in specific scientific areas and in a more general computational-mathematical context.

As we find ways to conceptualize objects and phenomena in spaces of dimension greater than 3, we can expect the thinking of scientists to expand correspondingly. The first inklings of conceptions of a physics with more than one time dimension are just now appearing. It is amusing to wonder what influence the tools of modern computation might have had in helping to stimulate this exceedingly speculative direction of theoretical exploration.

Another possible consequence may be a diminished emphasis on analytic mathematics in the training and expected skills of scientists. Traditionally, physicists and physical chemists expect to have mastery of a fairly standard core of mathematics, in calculus, differential equations, series methods and group theory, at the very least. Now, if they can evaluate complicated integrals and sum series and solve differential equations by instructing a computer to run a suitable program, students of chemistry, biology and even physics may feel less need to learn how to do these things from their mathematics colleagues. A parallel might be drawn with the way graduate programs in the sciences have reduced and eliminated foreign-language requirements as the need diminished to have a working knowledge of French, German and other languages apart from English. Problematic aspects of this issue are discussed in the next subsection.

Scientific Communication and Literature

We can expect increasingly open access to the literature through electronic archives, including much of what is now available in libraries. Electronic publication will at first supplement conventional publication of printed materials, and then will begin to supplant it in many of its roles. Despite the title of this study, we cannot expect paper publication to disappear from two of its roles for a very long time. One is the personal copy of an article or journal or book that a reader can carry to read in any environment. Computers that are as easy to carry and read as a magazine or book may well become commercially available within the lifetimes of people now living. (Some groups are creating prototypes even now.) These computers may also be capable of linking, by some form of high-bandwidth wireless communication, to data archives that will make a single computer capable of functioning like an enormous library of published material—with the power of creating links at will from one published piece to others. But, even though experiments are underway now to create something like this, these personal library devices will require significant infrastructure to empower them to do all this. Inertia and cost may make the development of such devices slow, particularly if we specify that they should be cheap enough that a household can buy several. The conservative projection here is that we will receive our "journals" by Internet but print paper copies of the articles we want to study carefully. That is, the printing of articles will shift from the publisher to the reader. This pronouncement, however, must carry a caution; as very light, portable computers with large, thin, flexible screens come into the market at prices working scientists can afford, and especially as wireless access to the "Giant Archive" develops, it may be that we will no longer prefer to read paper versions of new scientific articles. The supposition that many, perhaps most scientists will still want paper copies for study is predicated on the guess that those moderately-priced computers with book-like or journal-like portability and legibility will not be on the market for 20 years. That supposition is nothing more than a guess, and if that is wrong, then the entire prediction must be changed accordingly.

The other function we can expect paper publication to continue to fulfill is archival storage. Good paper lasts many centuries, at least. (Tragically, we know that high-acid paper hardly lasts decades!) Everyone familiar with the evolving world of electronic communication and computers knows the three kinds of fragility of modern data storage: the medium itself may be impermanent, however well-treated; the capacity to read any particular medium may disappear; and electronically-stored data are vulnerable to contamination and wipe-out, inadvertent or otherwise. The consequences are discussed below. Even so, known, tested methods can enable us to cope with all three, and those methods, like sustaining medications, will suffice to let us go on using electronic information storage, so long as we continue the medication.

Regarding impermanence, we do not yet know how permanent some of the electronic media are, notably CD-ROMs made to be archival; we unfortunately do know that some of the media, such as the large reels of magnetic tape, have such short lives that it is costly and even hazardous to use them for archival purposes. It was standard operating procedure, for many years, for government agencies handling very large data sets to copy old tapes onto new on a regular basis, typically biennial or triennial. Otherwise the data on those tapes would have been lost because the tapes simply degrade. Optical disk storage, such as videodisks and CD-ROMs provide, has reduced this problem qualitatively, but we have little information about whether a videodisk can be expected to last 50 years, or to be readable by devices available fifty years after the disk is written.

This brings us to the problem of technological obsolescence. We have not yet learned to institutionalize the maintenance of capabilities to read obsolete but relatively recent forms of stored information. It is already difficult in some places to find microfilm and microfiche readers, yet many doctoral theses are readily available only in one of those forms. People will insist on having paper copies as ultimate archival forms for a long time.

Third, protecting electronically-stored archival data from contamination or destruction can be done by systematic backup, combined with geographically-distributed redundancy and storage of the backups off-line, in protected locations away from the computer sites. Just as with impermanence and technological obsolescence, avoiding this kind of problem requires systematic human intervention. Presumably the requisite intervention will become institutionalized, creating a class of information-protectors who will serve a function akin to one of the many that librarians now carry out.

An amusing possible consequence of the transition to electronic libraries might be the disappearance of the personal library, the room or office lined with shelves of books and journals. Perhaps only collectors will maintain these. Perhaps the bookshelves will be replaced with electronic screens covering the walls, on which the homeowner can choose to display art, dramatic entertainment or whatever else he or she fancies. (We can also expect that collectors of computers will begin to appear soon, looking for early models. There are already computer museums, and computer sections of technology museums have existed for years.)

In the same breath, we can describe ways we will assimilate computer-based communication into casual, daily communication. email will be as ubiquitous as telephone messages, particularly if the sender has a choice of email or voice mail, with its interminable monologues and endless forked paths. Moreover, although both email and faxes can be forged, email can carry material with signatures and the contents can be protected by encryption.

The Institutions Where Science Goes On

The institutions that sustain scientific research—the universities, the Federal laboratories of various kinds and the research laboratories of industry—will evolve as more and more scientific information and information exchange become electronic. Within universities, the libraries are most obvious to be changed; this subject is discussed elsewhere within and beyond this volume. Other, more subtle changes also may occur in universities. These include the implications of electronic publication for recognition, promotion and tenure; the ease of access to information across disciplinary lines; the vast possibilities for electronic modes of education, a topic treated in this volume by N. Kestner; and the ease of ties among researchers in different institutions, much less different parts of the same institution. The final section of this essay examines some aspects of the ease of communication among institutions; here we look only at possible effects of such communication on the home institutions.

Recognition, promotion and tenure in science come ultimately from the extent to which an individual’s contributions influence and stimulate the thinking of others. The usual pathways for disseminating those contributions have been publication in recognized media, presentation at meetings and in seminars and colloquia, and sometimes personal discussions. The "recognized media" in science have, for two centuries, been scientific journals, in which an imprimatur of some level of validity is a necessary condition for publication. The early system of validation by an editor has been almost entirely replaced by the now-standard system of anonymous refereeing, with the referees selected by editors from their lists of people presumed to be demonstrably competent in the fields for which their expertise is called. It is important to recognize that the refereeing process, which operates largely on a volunteer basis, presents a rather low threshold of validation. It is sometimes supposed by nonscientists that the refereeing process should guarantee the correctness of a paper. This is a misinterpretation of the process. Rather, refereeing is supposed to guarantee that the paper is free enough of egregious errors, and is presented clearly enough, to be a valid topic for scientific discussion, scrutiny and criticism, as well as something other scientists may use to advance further. Some papers, after all, present speculations and proposals for trying new directions, with enough justification to make the speculations plausible. Some of these turn out to be correct, and others, simply not. Other papers appear occasionally with very subtle errors, or with assumptions that seemed correct at the time they were published but that later proved wrong. These papers all pass the screening of the referee system, and some of them are extremely imaginative, so much that they influence subsequent thought even though some aspects of them prove incorrect.

The advent of electronic publication has stimulated questions about the present refereeing system and the role it plays in evaluating faculty scientists and scientists in other institutions. Three alternatives to conventional refereeing are 1) open publication, e.g. by posting on one’s own Web page or its future replacement, or in unrefereed electronic archives; 2) refereeing by open comments by identified authors, posted in the medium in which the original article is presented, e.g., as sometimes occurs with articles submitted to the Los Alamos electronic archive "xxx.lanl.gov", and 3) certification by a panel of people presumed to have appropriate expertise, e.g., as proposed recently by a group of academic administrators led by Charles Phelps of University of Rochester (see R. Wilson, "Provosts Push a Radical Plan to Change the Way Faculty Research is Evaluated," Chronicle of Higher Education, 26 June 1998). But regardless of medium, it is extremely unlikely that the fundamental procedures of evaluation for scientists will ever change. The essence of such evaluations is accumulating a body of information about what other scientists think of the ideas and the work of the person in question: the members of the individual’s home institution study the work and form their own informed views, and, at the same time, ask for opinions of others elsewhere who are capable of giving expert critical judgments. The evidence that any of these people use will inevitably consist of anything that they have used or even just read previously and anything that comes to them when they are called to make their evaluations. If some of the work has been published in unrefereed or "very tolerant" media, this may be noticed, but will not necessarily lead the evaluator to denigrate the work; rather, it may put a bit of extra burden on the evaluator if he or she asks why the work was not published in a medium with more stringent standards. In short, this writer believes that the question of electronic publication influencing evaluation, promotion and tenure is essentially a non-issue, at most a small transient one that will evaporate as electronic publication spreads.

There may, over a period of decades, be a different kind of change in scientific institutions created by electronic communication and electronic information. Specific to institutions that exchange their scientific results and ideas readily, this possible change is a weakening of ties of individuals to their home institutions in exchange for stronger ties to the communities of their fields of endeavor. This has already been a trend in fields such as experimental high-energy physics, in which large inter-institutional collaborations have strengthened ties within the collaborating group and have kept researchers for extended periods at sites remote from their home institutions, and hence out of day-to-day contact with colleagues in those institutions. The ease of inter-institutional collaboration by electronic means removes some of the physical displacement but offers strong inducement to choose as collaborators the people one can most readily identify as appropriate, on the basis of characteristics such as expertise in a needed area. This has always been a significant factor, but in the past, there has also been motivation to collaborate with colleagues in one’s own institution, for both convenience and collegiality. The consequence has, in many instances, led to creation of very strong centers of expertise as consequences of evolving, converging interests of people who daily discuss topics of intense interest to them. Such self-generating, localized groups may come into existence less frequently as scientists easily find their collaborators elsewhere, rather than down the hall.

Untoward Consequences of the Transition: "What's long division, Daddy?"

With all the potential positive consequences for computer-literate generations, there are also potential negatives that we cannot overlook. The best we can do, from the perspective of this study, is to examine them, determine which are real problems and which are merely our reactions to doing things in new ways, sensitize ourselves to the real potentialities, try to shape our future to avoid or minimize those problems, and learn to recognize and evaluate new potential problems as they come along.

One possible course that seems negative to someone in a transitional generation is that virtual reality may become as satisfying as reality, as a means of entertainment in the largest sense. It seems necessary that most people who would like to know what it is like to fly a plane would satisfy themselves with a simulation. Should it be considered a good thing that one will be able to climb mountains in virtual reality? to visit distant cities? to talk with people in foreign countries?

We have already addressed the possibility that symbolic mathematics programs may make the learning of much traditional mathematics obsolete. If computers can generate or store the results of many traditional skills, may we be in danger of losing those skills? Will young chemists and physicists know how to carry out a Fourier transformation analytically, or to diagonalize a matrix, or to solve a linear, ordinary differential equation? Will the development of mathematics become even more separated from its applications, as a result of the availability of "black box" computer programs that provide solutions to virtually any well-posed set of equations? Already, construction and programming of algorithms have become professional skills in their own right. Can we expect another computer-centered profession to evolve, the development of new methods to carry out symbolic mathematics on a computer? The computer may influence career patterns in many fields outside the sciences, of course. Computer art is already an established form of expression. If a computer can be taught to do Japanese ink-brush calligraphy, will this art disappear? There is a danger that such arts and crafts will become arcane, practiced by only a few, highly-skilled experts with no coterie of devoted amateurs around them any more. At the same time, there may well appear new varieties of art-oriented computation.

The growth of electronic media may well increase specialization, but this would be analogous to the new specializations spawned by every new technology. We can expect more and more separation of strata in the population, corresponding to the computer-illiterate, the computer-comfortable, the computer-facile and the computer-knowledgeable. That is, there will remain those who choose not to use electronic means of expression; there will be many who use computers as most people use automobiles, at ease with all the normal controls and rules of operation but unprepared to repair any but the most minor problems and certainly in no way concerned with adding innovations; there will be those who know enough of how the programs and systems work to make their own adaptations, something like being able to "hot-rod" a car; and then there will be those who know the software or the hardware so well that they can manipulate and advance them—the best of the hackers, as well as the software and hardware engineers. There are now computer facility maintenance and repair firms, like heating and air-conditioning firms, that will come on call to carry out repairs, as well as to design and install new systems. Already, many firms and academic units have specialists on their staffs for this purpose. Without them, the computer-comfortable but unknowledgeable majority would be helpless in the face of any serious system failure. Moreover, this points out the vulnerability we create by assimilating electronic technology into our way of doing research, scholarship and teaching, as well as business and entertainment. It is one more, quite extreme example of increasing the complexity of the society by creating a new kind of interdependence on one another's specialized skill and knowledge.

With this interdependence comes a need to keep the skilled technologists up to date in the technology. Obsolescence sets in faster with computer systems than with any other human activity. The computer salesman, repair man and consultant are probably strongly motivated to remain as current as possible, at least to the part of the market they serve. Their employers are surely at least as strongly motivated to keep their staffs current. But the field moves so fast that the suppliers who need such people to be in contact with their customers and clients, or independent entrepreneurs ready to exploit the opportunity, will be trying to offer training programs to see that their agents do maintain currency. How will university programs designed to teach students to use computers cope with technologies that advance as fast as new courses are offered? Can academics learn to educate students to adapt to changing methods, instead of or in addition to teaching students to use the methods and tools of the recent past? Can we teach such things?

One issue related to maintaining skills is the low-probability, heavy-consequence problem of protecting against disaster. Maintaining redundant storage sites, whether true mirrors or independently-structured (but compatible), is the natural way to protect against loss of information from a local disaster. What about disasters that affect many places, such as a nuclear war? Even here, redundant sites will provide considerable protection. It may be, if we again go through a tense period comparable to the Cold War, that we will put some of these sites into protected locations, with protected power supplies. But we will need to maintain the capability to service the machines that enable us to read the information. This second-order preservation will pose a more difficult problem. One natural recourse will be to keep paper copies of the valued documents, but these may be at least as vulnerable as the far more condensed formats of CD-ROMs, and will obviously not be nearly so convenient to duplicate. It may be easier to find ways to preserve the necessary technological skills than to preserve many libraries.

One of the social aspects of the fast evolution of electronic media is the near-term issue of finding a way to balance free and open access to material available electronically against the protection of private intellectual property. This subject has grown into an acute current concern at the time these essays are being written. Because treaties and laws have been proposed that offer strong assurances of protection of private intellectual property, some believe they threaten freedoms of access, based on arguments of public good, to which the scientific and educational communities have become accustomed and depend. Finding a balance will certainly occupy a good many people.

We can try to look beyond the immediate problem of resolving this conflict in the context of current electronic media, to conjecture what the situation might be in another ten or twenty years. At that time, it is likely that a person motivated primarily by a desire to disseminate ideas and materials—as is a scientist when she or he publishes a scientific paper—will be able to do so electronically, not only as a "journal article" that, very likely, goes through peer review, but also as a "book". Both of these may be made accessible only to those who pay a price set by whoever has put the material onto the network. This means that self-edited, self-published scholarly books may compete with conventional published books in hard copy (but not necessarily in hard cover!). An author already can put a book up on the World Wide Web, with enough material accessible for free browsing to enable a potential purchaser to make a rational decision, after which the reader can gain access to the full text upon payment of a fee. Perhaps there will also be printers that will bind the text into a book format, or, as mentioned previously, computers that handle like books, so the text can be read as we now read book texts. Whether the Web will offer means to market such "books" as well as professional publishers can is an open question. If not, then texts oriented toward university courses, particularly large undergraduate courses, may still be produced and distributed by commercial publishers. If the Web turns out to be an effective marketing medium, then authors of the future might well bypass commercial publishers in disseminating their books—unless, of course, the publishers were to offer other inducements such as skilled editing or artistic work, well beyond what the author would be prepared to provide.

However one looks at the situation, the role of the conventional publisher is very threatened, and the publishing industry must be prepared to find creative ways to serve authors and readers, if it is to survive for more than ten or twenty years, in any form even remotely like its present one. Scientists and scholars often ask now whether middlemen will continue to have a natural niche in scholarly publication. It is possible to interpret the contemporary steps by some publishers toward increased restrictions as a "hold the fort" reaction, which is less productive than a more creative adaptation. At present, some publishers of scientific journals have adopted pricing policies for the electronic, on-line versions of their journals that have stirred intense resistance from some academic institutions, particularly from the libraries. One possible outcome is that scholarly "journals" will disappear from commercial publishing houses and remain only as the products of the professional organizations whose members have strong vested interests in the continuation of these media. Another is that commercial publishers will find new paradigms for their pricing. A particularly unlikely one seems to be evolution of scholarly publication to little else but unrefereed self-publication. This model seems to be the lurking fear of some of the most severe skeptics of automatic e-print archive distribution. However, reviewing prior to publication is so firmly established in the sciences, to establish a minimum threshold that new material must pass to be suitable for general discussion in the community, that it seems unthinkable that the community would surrender that. Rather, it is more likely that communication will evolve to use both modes--some form of the e-print archive for distribution and the refereed journal for certification and, we hope, for other forms of added value, such as links that the e-print archive might not offer.

International Ramifications: "How's the political mood in Sofia today, Ivan?"

One social aspect of electronic communication of which virtually every American scientist is aware is the ease of communication with colleagues almost everywhere in the world. Already, this has tied people together socially as well as professionally. Electronic conferences are no longer rare. Teleconferencing, at least at low resolution, is no longer expensive. Wireless telecommunication and earth-circling, high-speed, fiber-optic cables are already part of the foreseeable future. We can project from these to imagine the world of personal communication in another two decades, and conjecture what implications the situation will have. Scientists, like many other computer users, spend substantial periods each day communicating by email with others in their discipline. email has become, for many, their preferred mode of informal communication, terser and far easier to use than voice mail, easier to read and cheaper than fax, accessible from virtually any part of the globe, and obviously advantageous over telephone, even with answering machines, because one can send full messages whether or not the recipient is there to receive them. By sending attachments to email, one can easily send long documents and images.

While email and other graphic communication modes will indubitably persist, we can look a little further ahead and expect audio and video modes to become moderately common as well. The graphic modes will remain because they require minimal preparation and minimal bandwidth. In addition audio and video email are almost sure to become common, in the form of files or packets sent and stored, much as textual email now is. Live, real-time audio and video will also appear as computer media; in fact, direct audio communication by computer-to-computer link is with us already, as is live, real-time text communication, in the form of programs such as "Talk". The change will be a) further expansion of video links and b) audio and video message packets that can be examined at the receiver's will. It will probably be much less than 20 years until we can expect to find a computer in every hotel room, giving the occupant access to the Internet, just as we now expect to find a telephone. Many hotel telephones already have sockets for cords to computer modems, but we don't have to bring our own telephones along when we travel, even though it is sometimes useful to take a cellular phone. By the same token, we can expect to have the option of leaving our computers at home and to be provided with Internet access wherever we stay. Likewise, the time will come when we will have wireless "telephones" with every computer, which will allow us to connect to the Internet or its successors from any spot on earth—or possibly even from the moon!

One impact of email communication already, and one that will inevitably become stronger and stronger with expanded use and modes of communication, is the cohesion among members of the communicating groups, and the consequent ready flow of all kinds of information between nations. This, in turn, will have a greater and greater stabilizing effect on international relations. It will simply become more and more difficult for anything to stay secret or unknown to the outside world. Information will flow readily in all directions, and claims and rumors will be verifiable within the time of an email turnaround. It will become more and more common for people to "know" one another via email and e-video independent of or long before their personal encounter. Especially when e-video becomes as commonly used as email is now, people's circles of acquaintances and friends will, for many, be far wider and far more cosmopolitan than at present—even for scientists now active in that already international community. (Will we ever reach the point of Internet marriages consummated only by artificial insemination?)

One optimistic conjecture one might make with some confidence is that computer communication by simple email might become cheaper and more widely available internationally than (voice) telephone communication. This could become a stimulus for literacy exceeding anything schools can now provide. It is predicated on the notion that simple, inexpensive computers and the requisite networks, whether wireless or glass fiber, perhaps capable just of narrow-band, email exchanges, could become cheaper than conventional telephones, whose bandwidth handles extended calls with the full bandwidth of voice.

There is an opportunity for developing nations, particularly for computer-literate scientists in developing nations, to take advantage of the pattern of evolution of computers and computer prices, to leapfrog into the computerized world of electronic communication. Foreign aid funds could be spent for friendly, up-to-date computers and networks to enter this world with none of the baggage of prior generations of slower, less powerful and less friendly machines. Whether this will be recognized by those managing the foreign aid programs of the developed nations is an open question; in some countries, the international relations community has traditionally been among the most scientifically illiterate and innumerate parts of the educated population, so they can hardly be likely to lead such a move toward foreign aid for computers. However, it is possible that if one or two nations realize the value of such a program, the other developed nations will follow soon. And from that, all the benefits described above will naturally ensue. In any event, we can say with considerable assurance that computer links to library and journal archives will be far cheaper for developing nations than maintaining libraries equivalent to those accessible by Internet.

Conclusion: Epilogue

The foregoing discussion is a set of imagined adaptations and consequences of the transition scientists are going through today and will continue to go through for the foreseeable future. As yet, we cannot imagine the state of computers and computer communication when it reaches a stage of "mature development," when its growth slows to the pace of the prosaic components of our technological economy, such as transportation and building construction. Now we can think of so many ways that computers can evolve that any kind of saturation is far over the horizon. Each new speculation stirs new awareness of what might be done, and, remarkably, the optimistic possibilities seem more numerous than the ill or untoward consequences. There are some, but opportunities offered by computers must surely stand as presenting one of the brightest outlooks we can find as we look among the many areas in which our world is changing.