david firnberg,editors, ,computers, management and information (1973) george allen and unwin,london...

2
730 BOOK REVIEWS They probably do not locate function in this sense of properties of the model Once the model is known, though, neurology should gain by having at last a decent theoretical basis. It should be possible to relate function to structurewhen we know what functions the structures perform. But for this we need to have an adequate model--a statement of what the functions are, which are carried out for behaviour and intelligence by the components of organisms. As a naughty generalization, one might say that each kind of brain scientist tries to "reduce" all brain function to processes of the kind he investigates. This is just as true for adherents of artificial intelligence---only they hope to reduce brain function to computer-processes (software rather than hardware) most of which does not (yet) exist. There are those who feel this to be like writing science from the standpoint of science fiction--and trying to "reduce" brains to imaginary science fiction robots. But surely all creative science has its fictional aspects, and no scientific models exist in the.sense that structures exist. This is true, for example, of astronomy: both lumps of matter (stars, etc.) and "models of the universe" figure in astronomy. The only difference is that though astronomers may pontificate like gods they do not set out to create universes: they are content with the one we have. This is not so for brain modellers: they want to create new kinds of brain. This book is not a blueprint for brain-building--none yet exists but it is probably the best primer, linking actual with potentially possible brains that has so far been written. The book is illustrated with bold black and white diagrams. It has notes and references for further reading and an author and subject index. R. L. GREGORY University of Bristol, U.K. Computers, Management and Information By David Firnberg. 1973 London: George Allen and Unwin, 136 pp. £3.00. It sometimes seems, to this reviewer at least, that the most pernicious contribution to the "information explosion" has been the veritable avalanche of books on management information systems. Tricker's annotated bibliography, published five years ago, listed some 196 books on the subject and, in the interviewing period, the pile of fat and expensive tomes on "integrated" systems, "total" systems and every other imaginable kind of system has grown ever higher. Most of these are, no doubt, of absorbing interest to someone, somewhere--although one is sometimes inclined to wonder if anyone ever really reads them. Yet, in spite of this embarrassment of riches (or at least of megawords), there still remains a mysterious gulf between the managers who are ostensibly the people for whom information systems are designed, and the computer specialists who provide these systems. At worst, this gulf manifests itself in badly-designed, inefficient and inordinately expensive MIS; at best it is seen (cf. the 1968 McKinsey report on computer utilization) as an unwillingness on the part of management to use the computer effectively as an aid to decision-making. Either way, the gap between the manager and the systems analyst is one of the great lost opportunities of the age. Mr Firnberg's short book is an attempt at bridging the gulf. It is written for the layman and aimed at the professional manager who will have to come to terms,

Upload: john-naughton

Post on 15-Jul-2016

212 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: David Firnberg,Editors, ,Computers, Management and Information (1973) George Allen and Unwin,London 136 pp. £3.00

730 BOOK REVIEWS

They probably do not locate function in this sense of properties of the model Once the model is known, though, neurology should gain by having at last a decent theoretical basis. It should be possible to relate function to structurewhen we know what functions the structures perform. But for this we need to have an adequate mode l - -a statement of what the functions are, which are carried out for behaviour and intelligence by the components of organisms.

As a naughty generalization, one might say that each kind of brain scientist tries to "reduce" all brain function to processes of the kind he investigates. This is just as true for adherents of artificial intelligence---only they hope to reduce brain function to computer-processes (software rather than hardware) most of which does not (yet) exist. There are those who feel this to be like writing science from the standpoint of science f ict ion--and trying to "reduce" brains to imaginary science fiction robots. But surely all creative science has its fictional aspects, and no scientific models exist in the.sense that structures exist. This is true, for example, of astronomy: both lumps of matter (stars, etc.) and "models of the universe" figure in astronomy. The only difference is that though astronomers may pontificate like gods they do not set out to create universes: they are content with the one we have. This is not so for brain modellers: they want to create new kinds of brain. This book is not a blueprint for brain-building--none yet exists but it is probably the best primer, linking actual with potentially possible brains that has so far been written.

The book is illustrated with bold black and white diagrams. It has notes and references for further reading and an author and subject index.

R. L. GREGORY University of Bristol, U.K.

Computers, Management and Information By David Firnberg. 1973 London: George Allen and Unwin, 136 pp. £3.00.

It sometimes seems, to this reviewer at least, that the most pernicious contribution to the "information explosion" has been the veritable avalanche of books on management information systems. Tricker's annotated bibliography, published five years ago, listed some 196 books on the subject and, in the interviewing period, the pile of fat and expensive tomes on "integrated" systems, " to ta l" systems and every other imaginable kind of system has grown ever higher. Most of these are, no doubt, of absorbing interest to someone, somewhere--although one is sometimes inclined to wonder if anyone ever really reads them.

Yet, in spite of this embarrassment of riches (or at least of megawords), there still remains a mysterious gulf between the managers who are ostensibly the people for whom information systems are designed, and the computer specialists who provide these systems. At worst, this gulf manifests itself in badly-designed, inefficient and inordinately expensive MIS; at best it is seen (cf. the 1968 McKinsey report on computer utilization) as an unwillingness on the part of management to use the computer effectively as an aid to decision-making. Either way, the gap between the manager and the systems analyst is one of the great lost opportunities of the age.

Mr Firnberg's short book is an attempt at bridging the gulf. It is written for the layman and aimed at the professional manager who will have to come to terms,

Page 2: David Firnberg,Editors, ,Computers, Management and Information (1973) George Allen and Unwin,London 136 pp. £3.00

BOOK REVIEWS 731

sooner or later, with the computer. The identification of this specific audience dictates the tone and the treatment of the topic--br isk and practical, with a mini- mum of theory and a refreshing absence of jargon.

There are six chapters. The first is a short and largely superfluous preamble entitled "Science and Society". Chapter 2 sets out to discuss the concept and nature of data and to describe (with the aid of an excellent example based on personnel records) the principles of marshalling data into a data base. To some extent, the treatment in this chapter can be described as low-level theoretical. The next three chapters then discuss the most important constraint which reality imposes on all data-processing activities--the need to cope with incessant change. Chapter 3 contains a brief resume of the history of data processing and discusses identification and coding. Chapter 4 introduces the "human component" and its effects on the design, implementation and operation of computerized information systems. Chapter 5 discusses the topic of management information in terms of the "information sandwich" in which every manager finds himself--caught between the information he is obliged to impart and the information he needs to collect for decision- making purposes. With regard to the latter, Mr Firnberg introduces the reader gently to the concept of terminal access for managers via the use of standardized information request forms. Chap te r 6 is taken up with two quite detailed case studies one of which gives a blow-by-blow account of the operation of an on-line information system which enables managers to search for company information by means of a Teletype-VDU link. It 's all very interesting and seems even quite useful. But, going on past experience, one would have to talk to the users before passing judgment.

JOHN NAUGHTON The Open University, U.K.

References

TRICKER, R. I. (1969). Management Information Systems: an Annotated Bibliography. General Educational Trust of the Institute of Chartered Accountants in England and Wales.

Foundations of the Theory of Learning Systems By Ya. Z. Tsypkin . 1973. New York , L o n d o n : Academic Press. 205+x i i i pp.

$15.00.

As in his previous text on a similar subject (1971), the dominant message of Ya. Z. Tsypkin's latest book on learning systems is the unification of learning algorithms under the banner of Stochastic Approximation (S.A.). And there can be no doubt that he presents, yet again, a persuasive argument to support such a unified approach to learning system design.

But it would be erroneous to assume that this new book presents anything like the current state of the art in learning systems theory: in many ways, it merely attempts to reinforce the message of the earlier book by bombarding an already beleaguered reader with stochastic approximation algorithms of, it seems, every sort, shape and size. We indicated in our own review (1972) of the previous book by Tsypkin that the reader would, perhaps, be rather bored by the end of the book with the constant, almost numbing reference to the stochastic approximation type of algorithm. One can only say that, if we were right then in our diagnosis, the