81 REVIEW AND SYNTHESIS B. C. Brookes 1. What is science? During our discussions it became apparent that we used the term science in different and confusing ways. What science? First consider pure science. Science is the body of knowledge about the world we live in which has been found by the activity we call research. Note that in considering the nature of science or research we are in no way concerned with the formalisms or techniques we have discussed. These are merely means to an end - a set of man-made tools from which we each make our own selections when we do research. When Francis Bacon formulated the Method of Induction 350 years ago he started a philosophical debate about research which has not yet been concluded. The method of induction requires the enquirer patiently to cumulate as much data as possible about the phenomenon under study until the enquirer finds it possible to make some generalization or formulate some explanatory theory about the phenomenon. The Method of Induction is much used in everyday life in much the same way as Moliere's Monsieur Jourdain found that he had used prose. But Bacon was writing in the early 1600fs before modern science had emerged. The philosophers' search for the secrets of scientific method was made more difficult by the astonishing success of Newton's theory of gravitation (1687). It seemed that the theory of gravitation was true. Even the great philosopher Kant (1787), who was sceptically critical about many other matters, believed that the Newtonian theory was true. So Kant led later philosophers to consider how the Method of Induction could lead to Absolute Truth. This problem was studied for more than 100 years - until Einstein, with a new theory, explained some of the baffling anomalies that 82. Newton's theory could not explain. proved that Newton's theory was not In so doing, Einstein absolutely true. He Karl Popper (1959) made a decisive break-through. {vide pointed out that no scientific theory can claim to be True the history of science) but is merely the best we have All scientific theories, however strongly in the Popper also pointed at any one time. established, remain always open to falsification light of later evidence and thought. out that, in any case, Newton had not derived his theory by the Method of Induction and that there is an asymmetry in the logic of the verification and falsiification. "swans": to disprove To prove that "all swans are white" one would have to observe all that "all swans are white", it would be enough to produce merely one non-white "swan". Though the logic of Popper's argument is sound, the falisification of a well-established theory is not, in general, achieved quite as simply as he claimed. It is as important to defend older theories against new heresies as to consider what advantages the new heresies may offer. The debate has been continued more recently by Kuhn, Lakatos and Feyerabend. In opposition to all other philosophers of science, who have been seeking for 'law and order' in science, Feyerabend argues that there is no method at all - that "anything goes". But he also admits that he is being provocative. My own view is that the scientific method involves some 'law and order', ie some discipline. In the natural sciences a phenomenon is studied, observations are made, data are collected only until an idea, a 'hunch', a generalization, a hypothesis arises. How? It does not matter. Nor can rules be given. If the hypothesis looks good, it should be tested destructively before publication (though referees help in this work). If the hypothesis is published, publication in a scientific journal carries the implicit challenge to all it may concern: refute this if you can! If it is 83. not refuted, it is criticized, modified and punched into better shape until, in turn, it is knocked out by some better theory. There are disciplines to be learned. have a distinctive rhetoric. Scientific papers To be taken seriously, the contributor has to use the current terminology of the subject and demonstrate that he knows its recent theoretical history. reasoning. He has to recognize what constitutes relevant He has constantly to refer to the phenomenon observational evidence and the various forms of fallacious under study and to compare his description of what he sees with the descriptions of the same phenomenon given by others. 2. Pure v applied science and research There has been no adequate analysis of applied it often emerges as hardware. research form: partly because much of it is not published in document The main difference between pure and applied research is that, in applied something "work". PURE RESEARCH Search for truth and unders tanding Must be published and invite public criticism Usually long-term Seeks generality Must take risks: hypothetical science, the search is not motivated by the A comparison: APPLIED RESEARCH Search for something which works Need not be published (except as a patent). Usually short-term ad hoc Seeks particular objectives desire for understanding and truth but by the need to make Must "play safe": positivistic Otherwise, both forms of research may call on the same formalisms and use similar working techniques. 3, Synthesis (a) The research levels If the above distinctions can be accepted, at least provisionally, then how do they apply to information science? The unchallenged research areas of information science centre on the operation of IR systems, the needs of users, the organisation and dissemination of information and the various technical processes involved in such work. In all these areas research is already separable into pure and applied. The Forum did not call for papers on particular systems, ie on applied example. research, but several papers on pure research were presented, of which Heine's model of IR systems was just one So I conclude that information science has now discernibly separated out into at least two levels of research in which problems which arise directly from the traditional operational areas of the subject are studied. There was also some evidence that a more fundamental level of empirical research may be arising around the central concept of the nature of information. These two levels of pure research will, in due course, feed back ideas and results into the operational level. Thus information science now has the research pattern of any other science; (b) it has arrived as a science. The need for coherence The pure research levels are still very thinly manned and the few active individuals are widely scattered. It does seem, however, that the pure research topics are almost ready to be linked into a coherent corpus of theory so that an integrated theory of information science, identifiable The fact that the research and autonomous, will soon emerge. a science: ideas are not yet coherent does not make our subject less of even theoretical physics is not coherent yet. oo The operating level is very busily occupied at present with the extension and improvement of information systems and has little interest in pure research. be applied. On the other hand, pure research has so far fed back little that could There is a need for more interaction between the research levels. (c) Related subjects Other papers presented at the Forum, for example, that by Samuelson, showed information science interacting with related subjects. science, This kind of interaction, too, with geneis also important for the development of ral systems theory, decision theory, cybernetics, computer information science now that it is strong enough to retain its identity. At present, information science is still a net importer of ideas from these other disciplines and it will not be well regarded elsewhere until it has ideas and theories to export to others. (d) Integrable topics One of the main possible integrations could embrace the mathematical/statistical models of IR systems with the analyses of dissemination and other explorations of the use of scientific literature. One of the needs is to link these processes with the time dimension and to generalize them into a unified general theory of human communication, largely statistical, which Goffman has already initiated. The time dimension is needed because though information can now be transmitted almost instantaneously, social acceptance of new ideas takes time even in the pure sciences. A second area which calls for closer interaction centres around various linguistic matters with which information science is involved. Theoretical linguistics has so far been wholly unhelpful in application to information work but in a more closely integrated information science those doing research on linguistic problems would have access to linguistic data of a new kind. The problems of indexing, abstracting, and compaction of titles, etc. offer a new kind of 86. relationship between language and its use and therefore an opportunity to study language in a new empirical way. (e) Shannon information theory The paper presented by Lynch reminded us that Shannon information theory is relevant to some of our problems. In the design of information systems we are concerned with many different kinds of input, many different kinds of user and many different types of search question. All this implies that the system parameters can only be statistical and that Shannon theory should therefore have further applications. For example, in the dissemination of information there scatter is a conflict between what appears to be a natural pact information relevant to any given theme. resting to note that the entropy bution of sources is gaussian. of both sources and users and the need to organize and comIt is inte(in the Shannon sense) of Gaussian distributions The cumulative a system of given mean "power" is a maximum when the distrioccur in the Swets model of IR systems. entropy. form of the Bradford law also has the form of a Shannon At this statistical level of the analysis of information systems and processes the Shannon theory may yet become one of the integrating formalisms we shall need. (f) The concept of information itself, There is also a developing concept of information mental level of research. In the scientific the study of which might become the focus of a more fundaexploration of information one would expect empirical detailed studies of individual 'users 1 before and after they have received some information. than quantity In the analysis of this kind of process rather Such work offers the a different approach is needed in which structure has to be considered. prospect of a new theory of semantics, based on structural relationships of a new kind, being established from within information science. ideas. Belkin has started work on these (g) Social aspects At the present time information science is still mostly concerned with the developments of information systems and networks for scientists. But the field of application It widens as new data-bases and networks are established. therefore becomes increasingly necessary to consider the social implications of the impending 'information revolution1 : Leimkuhler discussed the implications of a wider dissemination of information for the firm which (from a theoretical point of view) offers a clearly-delineated entity interacting with its environment - an "open system11 - and so provides a test-bed for observational study. Here the main objective of the firm is to ensure its survival in an increasingly competitive environment. Kochen's WISE project is concerned with the wider dissemination of information to society at large. And Brookes outlined an evolutionary theory in which man could be regarded as the latest stage in a series of animal species evolving towards an ever increasing information-processing capacity in a struggle for survival on Earth. The unique step that man has taken, through records, reprography (wide sense) and the computer, has been to provide himself with an exosomatic information storage and processing capacity of unlimited scope - a kind of social brain which, in principle, can be shared by all mankind. He saw this as likely to be increasingly concerned with the problem of ensuring the survival of the human species - either on Earth or on some other planet. The continued development of information services is going to raise issues of increasing social importance which need series and early consideration. What is it all for? (h) The place of the discipline My answer to the problem considered by Wersig: What is information science, and which is implied in my comments above, does not conform exactly with any of the solutions "broad", "medium" or "narrow" - which he propounds in his excellent analysis. It seems to me that Wersig's three solutions can be regarded not as mutually exclusive but as representing phases in an evolution from the narrow front from which information science emerged to a broader front as the subject develops. In fact, the whole corpus of active information scientists is always likely to constitute a group accommodating a wide range of individual interests and attitudes, so that all three of Wersig's could co-exist. "solutions In this early developmental phase of the subject a diversity of approach could be creative, as Wersig suggests. The issues that separate us are not crucial and will be decided by history rather than by debate. 4. Main Conclusions with the (a) The shape of an integrated information science other empirical science is now discernible. operational and research levels characteristic of any (b) The central topics of the sciences are information organisation, dissemination and retrieval - the topics unchallenged by any other discipline. (c) The pure research levels are still thinly manned and the individuals concerned are too widely scattered for the critical interaction which is needed for ' t a k e - o f f . 5. Recommendations (a) We need no longer examine ourselves introspectively. The next Forum on research could more usefully concern itself wholly with specific research topics and their integration. (b) Those with academic responsibilities need to work towards a research-oriented integration of the subject. recruits to the pure research levels. We need ) Theoretical research in information science is still marked by a tendency to 'play safe1 and to be positivistic because it has grown only slowly out of applied research. So it is still marked by timidity. It could now afford to be more boldly speculative, intellectually exciting and. therefore more attractive to intelligent and ambitious students.