The fundamental changes in scholarly communication may not be the most spectacular effect of the availability of cheap computing and very cheap world-wide networking, but it will have the most profound impact on the people and organisations we serve. The concepts governing the nature and speed of these changes go to the heart of our profession, to the future prospects of our profession - and therefore of our own futures. In such a fast-changing, networked world we need to be quick on our feet in making decisions about services for our users because opportunities arise and disperse very quickly. This paper will discuss some of the economic, technical and political issues involved in answering the questions so liberally posed in the title, though philosophical and semiotic discussions about the nature of scientific communication will be left to those better able to do them justice.

One aspect of the changing face of publishing is quite certain: the expenditure patterns of publishing organisations is changing drastically. This section of the paper looks at the nature and extent of those changes, forecasts that they will result in very different types of organisational structure and outlook, and in publishers using different strategies to attract a viable level of income. The underlying assumption is that over time the printed journal will disappear completely, replaced by some electronic, network-dwelling entity. This seems on the face of it a wild assumption, but who would have forecast the ubiquity of the Web only five years ago? In practice, some general-purpose journals such as the BMJ and the Lancet will survive for the foreseeable future in something like their present form, but the specialist journal is likely to go the way of the vinyl record and the Apple Mac.

In plotting the expenditure implications of a move to electronic dissemination of research, I have taken as a model a fictitious middle-sized journal publishing company run on a not-for-profit basis and producing about 30 medical titles from a turnover of $33m/year. The purpose of the model, which uses what information diligent research (electronic and otherwise) has been able to glean, is to determine what proportion of total expenditure goes in the irreducible essentials of editorial management and peer review. From this estimate is drawn some inferences about the size and nature of the residual organisation necessary to produce the equivalent quality and quantity of information in electronic-only format.

Harnad provides what he says is a conservative estimate that such a switch would cut costs by 70% but this can be further reduced substantially (a) by factoring in an allowance for the way the nature of the publisher's organisation would change ("small is beautiful" savings); by (b) the simplified revenue-generation picture; and © the huge and continuing reduction in the cost of computing and networking.

The overall expenditure analysis is set out in Figure 1, and the potential cutting of costs to around 10% of the original figure is achieved thus:

Of the total annual expenditure of $33m, $10·5 is spent on printing and $7·5m on postage. These figures relate to the direct external costs of these processes - ie the total bills paid to the printers for producing the journals and to the PTT for delivering them to subscribers.

From the remaining $15m can be taken $5m worth of associated distribution costs, these being the internal costs of maintaining subscription and mailing lists, putting journals into envelopes, maintaining the stock warehouse and so on.

This leaves an operation turning over $10m a year, from which can be further deducted advertising and marketing costs of $4m: the electronic production is a low-overhead business and the "journals" have more-or-less zero marginal distribution costs, so there is no need in a non-profit environment to run expensive departments to chase after lucrative display advertising and bread-and-butter classified advertising for jobs. Further, since the organisation no longer needs to sell thousands of individually small subscriptions to libraries and individuals and need no longer be touchy about copyright protection and trade-association protectionism, there is no need for marketing and international relations departments.

Having transformed the organisation from a $33m flagship of the publishing industry to a de-glamourised $6m scholarly approval factory, we can further trim the overheads by eliminating the prestige offices. Cutting staff numbers down well below 100 also means that bureaucratic and procedural overheads (for example a Personnel department) can be shed, saving a further $2·5m.

All of which leaves us with a $3·5 organisation in modest provincial accommodation, serving the needs of its professional community by producing as much high-quality, peer-reviewed information as the $33m behemoth with which we started. A Jane Austen organisation rather Universal Studios. In practice it would serve its community better because it would not have to turn away for reasons of space material which had passed its peer review, and its output would be available from millions of workstations rather than thousands (or even hundreds) of libraries.

There are, admittedly, some elements of both the conjuring trick and of utopian wish-fulfilment in this picking apart of the evil print publisher. The figures are based on approximations and assumptions, although there has been an attempt to err on the side of caution. There is no element for buying and maintaining the computer systems, and for the development capability such an organisation would need to maintain. It also makes big assumptions about how such a brave new world would find the money to keep itself going, which are made more explicit and then justified in the next section of this paper.

On the other hand, the costs of the day-to-day business of managing a peer-review operation should not be over-estimated. For many years the BMA library has run annual book and film competitions which have had at their heart a rigorous assessment process which is closely analogous to that run by the best scholarly journals. Each year we arrange for some 180 videos, 250 books and 25 multimedia productions to be judged by both their intended target audience and by professionals able to make informed judgements of the accuracy, currency, completeness and balance of their content. This is by no means a simplified version of the editorial judgement. To give but three examples it covers the whole range of health and medical subjects, both competitions involve a variety of technology to access all the entries, and it is a multi-stage process overseen in careful detail by the elected members of the BMA's Board of Science. Including all overheads it costs around $15,000 a year.

The radically different type of organisation postulated above has, in turn, radically changed income requirements. No longer does its cost base require to be underpinned by expensive-to-attract advertisers, and with only $3·5m to find each year, the expensive, uncertain and inefficient sale of single subscriptions to single titles is no longer the only practicable method of keeping the finances in balance. The whole cost of operating can be met from one or two strategic "sales", for example to the higher education funding body of the country in question, or to university or library consortia. The reduced level of outlay even makes it become feasible for there to be a general return to the Enlightenment model (which still exists in some instances today, of course) of professional bodies taking on responsibility for communicating news and innovation and to the general world of scholarship. We can expect different sectors of the publishing industry to react differently and at different speeds to these changes.

Commercial publishers have the simplest of corporate motivations. They will find it hard to adjust their operations and their aspirations from the highly-geared Big Capitalism of the present-day industry to the no-frills environment in which they will increasingly be forced to compete. If librarians and others with similar motivations (and similar levels of pay) can manage the editorial process adequately - and we can - and integrate it with the related tasks of indexing and bibliographic control, then it is hard to see capitalism remaining competitive in a cottage-industry milieu. Certainly it has no moral right to do so, as the cycle in Figure 2 shows only two clearly - and as a plot of journal prices against general inflation would show equally clearly. From a strong, short-term position int he market, many commercial publishers are seeking to sustain themselves by manipulating the development of electronic products to make them in many ways less accessible than print, rather than more: for example by restrictions on fair use copying, using annual licensing to ensure users do not enjoy the perpetual access to back-runs which we take for granted with print. Worst of all, they seek to hobble delivery systems so that they deliver a dismal and calculated subset of the technology's inherent potential.

With learned society and university publishers the position is by no means so clear cut. The overall aim of the BMJ Publishing Group, for example, is to disseminate high quality scientific information to the medical community, and to earn enough money to enable them to do this effectively. For the professional presses and their parent organisations to survive in the harsh economic climate of the 1980s and 90s, however, they have had to behave more and more like their commercial cousins. Now, when the best of them have mastered this new modus operandi, the whole paradigm of scholarly communication is changing. And they will have to change with it. Typically such presses are subject to a measure of democratic control over strategy, and an increasingly information- and network-literate membership will demand a proper opening up of full-strength electronic "journal" systems. They will expect their professional bodies to seize the chance to free their research communities from their indentures to commercial copyright owners. Professional presses will adapt to the new low-cost, unglamorous world, or die. It is our job as librarians and professional to make sure that they adapt because we need them to much to let nature take its course.

Self-publishing by authors has become a genuine force in scholarship. Almost all the research for this paper used Net resources made available by the academic community in one way or another. Self-publishing authors simply want to be read, appreciated as significant and cited. They are prepared to spend their employers' money freely to achieve these ends and it is probably our professional duty to assist them in this. It is the energy and the base-line accessibility arising from self-publishing that will drag us all into the new environment of print-free dissemination of knowledge.

Accessibility is too big a technical and political topic to be dealt with comprehensively in the context of this paper, but there are three considerations of particular relevance to librarians that I would like to touch on here:

When considering the breadth of distribution of Internet resources, we need to take care to avoid being too library-centric in our attitude. The days are numbered of the concept of the library as exclusively a literal "place". People will still come to where we work, but the new emphasis has to be on our going to where they work.

Standards and interfaces are important issues for ease of use. But the crucial elements are not ISO, professionally-driven standards - the crucial standards are so-called industry standards such as Netscape. As a profession we can safely stop fiddling with MARC format variants and get on with making sure our users know how to use bookmarks and formulate Advanced searches on AltaVista.

In an anarchic, distributed publishing environment, our sophistication and still in organising information will be crucial, and securing funding to allow us to apply them is of paramount importance.

Again there is no space to do more than touch on a few key issues (readers looking for a good summary should consult Odlyzko. For the next few years networked "journals" will not be a complete alternative to print in the sense that when we want to read something with serious attention we will continue to print it out first. Until battery and (especially) display technology improves, then our reading devices will not be able to pass the Read In The Bath test with which UK popular journalists seem so obsessed. Our readers will have to like and feel at home with their pocket readers - even the crustiest humanities dons. The new media have to fulfil all the functions of print and not just the informing functions, though it is hard to see how network output will ever manage to stuff a lecturer's office so full of evidence of erudition that no-one can sit down. Finally, they have to look cheaper than print, at least as far as administrators and policy-makers are concerned. This is, of course, more important than whether or not they are cheaper.

There are real dangers that still have to be circumvented. Only time will give users a sense of security and confidence over peer authentication of results, and allow a significance hierarchy to emerge (the equivalent of "it must be good, it was in the BMJ"). In the pioneering fields of mathematical, physical and computing research this is already happening. Some means needs to be found of rendering immune from accidental or deliberate tampering the "reference" text of results; and on a related issue we librarians have to tackle the problems of archiving and of preserving the chains of thought which have previously existed with scientists' personal papers. Finally, there is the perhaps insoluble problem of the Year Zero Syndrome familiar from Medline: whereby papers published before the technological watershed are unfairly ignored because people are too lazy to search laboriously for them, and more recent papers are unjustifiably cited because they are too accessible and have alluring abstracts.

It seems certain that all our scholarly print media are in the early stages of terminal decline. The particular virtues of the book will ensure that this decline is steady and spread over a considerable period of time [Figure 3]. With periodicals the picture is likely to be very different. Few users or working librarians can find any warm place in their hearts for the traditional scholarly serials, perhaps as a result of years of gouging price increases and getting our clothes filthy while working in stack shelving. Once all the conditions set out in this paper are met, there will be a catastrophic decline in the number of titles published. When precisely the collapse shown in Figure 4 will occur is still unclear, but probably 10-15 years. Until then, we will see scholarly communication mutating rapidly to take advantage of the new possibilities: video and sound, real international collaboration, iterative methodologies replacing conventional concepts of "authorship" of research. Librarianship, publishing and IT will gradually become a single profession, and as librarians we have to ensure that we are alert and quick-moving enough to ensure that our contribution to this process is both incorporated and valued.