The use of history in IS research: an opportunity missed?

Abstract

The article is shaped by two regularily repeated cliches. The first is History is bunk. Henry Ford's well known saying has two implications: (1) that what purports to be history is more often than not inaccurate if not a downright lie, and (2) that we have nothing to learn from history as modern innovations make the past irrelevant. The second cliche is We will heed the lessons we have learned from past disasters. How often do we hear that claim with respect to information systems (IS) failures? Again there are two implications: (1) that history repeats itself, and that if we learn how prior mistakes were made we can avoid the same mistakes being repeated, and (2) that we have the capability to analyse the past with sufficient accuracy that we can identify all the problems that led to the mistakes being made. In this article, I will argue that the historiography of IS is important to understanding IS and its evolution through time, and that understanding even the most transformative, revolutionary, innovations benefits from the study of the historical context. Henry Ford's viewpoint is far too prevalent, and in my view damaging to IS research. The argument will be supported by a number of examples.

Appendix

Frank Land responded by e-mail to six questions from Dan Power, DSSResources.com editor, about his past involvement with computerized decision support systems (DSS) and his current perspective on the issues that need to be addressed.

Q1: How did you get interested in computerized decision support?

Land's Response: Decision Support has an ancient history. Decision makers have always surrounded themselves with specialist staff to provide information as a crucial aid to decision making. In the army, for example, the decision support function was provided by the adjutant.

We can perhaps distinguish two kinds of DSS which we might term Traditional and Modern. Is there also a Post-Modern type?

Traditional DSS are the historic kind, though today still as important as ever – the decision makers being supported by a range of formal and often informal information and knowledge providers. These may be people, like the adjutant or accountant with formal support roles or informal like the business rival over a game of golf. Or they can be artefacts, formal, like an official report requested by the decision makers, or informal like a newspaper report seen by the decision maker at just the right moment. As is often the case serendipity plays an important role in reaching decisions.

Modern DSS are largely reliant on formal models whose expression and evaluation depends on computer technology. They rely to a considerable extent on mathematical modelling and simulation techniques. Many of the ideas stem from the decision sciences and operational research and were first developed in the run-up to the Second World War as part of the war effort.

My own involvement arose out of my first employment with J.Lyons & Co. in 1952.

J.Lyons & Co, were the largest and best organised company in the UK food trade – restaurants and hotels, food manufacturing including bakery products, confectionary, tea and coffee, and specialist caterers for events such as the annual Wimbledon Tennis Tournament, and the Royal Garden Parties, had established a Systems Research Office in the early 1930s.

In Lyons the management structure was, in a sense, based on decision support. Each functional unit – for example, the bread and cake bakery – had at its head a member of the Board. A liaison unit served that function providing detailed information on each days trading via a set of cost accounts. The head of each unit was directly responsible to the Board member. He/she were responsible for reporting variances and providing explanations of any variances discovered from the cost accounts to the manager in whose area of responsibility the variance occurred. In addition, the head of each unit was required to work out answers to questions from senior managers of the functional area served of a ‘what if’ nature. For example, what would be the impact of changing the production mix to increase the production of swiss rolls, or to replace raw material ‘a’ by raw material ‘b’. In practice, they spent much of their time working on these problems and providing the required information for the decision makers. The kind of questions might be of a local operational nature or much more concerned with matters relating to company strategy.

The system had been designed and implemented by one of the true pioneers of Decision Support – JRM Simmons, a Director of J. Lyons, recruited by the company in the early 1920s directly from Cambridge University where he had graduated as the top mathematician of his year. It was John Simmons who had persuaded the Lyons Board to build their own digital computer, Lyons Electronic Office (LEO) to support the business in 1947. His book ‘LEO and the Managers (Simmons, 1962) sets out his ideas and shows their development in the computer age.

Thus Lyons had, before the advent of computers, a well developed and effective decision support mechanism though Simmons recognised that computers would play a crucial role in making an effective system even more effective.

Another pioneer was David Caminer who had joined Lyons as a management trainee in the 1930s. On returning from war service David became manager of the Systems Research Office established by Simmons in 1932. David was made head of systems and programming when the decision to build the LEO computer was made. He played a crucial role in the design of most of the early computer applications for the Lyons business. It was perhaps natural for him to see the role of computers at Lyons as supporting the work of the liaison staff. Hence, nearly all early applications dating back to the early 1950s and subsequently incorporated decision support elements. There were numerous examples ranging from the system which helped the managers of the chain of Lyons tea shops in placing their daily orders on the factories and suppliers, to the Bakery Rounds application which printed an order form for each customer the bakery salesman called on, listing the items ordered in previous calls, as a reminder of that customers preferences.

I joined the Lyons computer team in 1953. After graduating from the London School of Economics (LSE) my first job in industry in 1952 was with Lyons working in one of the liaison units described above. As a result I absorbed the Lyons way of working and the way they had developed an organisation capable of supporting management in both its strategic and day-to-day operational decision making. When I became part of the Lyons computer team in 1953 these ideas were already deeply ingrained in my thinking.

Q2: What do you consider your major contribution to helping support decision makers using computers? Why?

Land's Response: As part of the LEO team at Lyons I was responsible for the implementation of a number of computer based applications, at first exclusively for Lyons, and later, when LEO became a subsidiary manufacturing and selling the LEO range of computers, for a number of industrial clients. The applications included a system for the ice cream business, which advised ice cream retailers how to fill their cabinets based on weather forecasts and the systems knowledge of each customer's ice cream sales history. This system was devised with the help of the Lyons Operational Research team and, looking at it in retrospect, was a step from Traditional DSS to Modern DSS. Another system I was responsible for implementing was the Tea Blending Programme, which supported the tea mangers in determining the best mix of blends to schedule each week based on tea prices and forecast demand. The system was in use, I believe for nearly 30 years.

Later (1967), I was recruited by the LSE to set up teaching and research in systems analysis. About 1970/71 the UK National Computing Centre set up a research project into evaluating the costs and benefits of computer-based information systems. Three of the researchers, Enid Mumford (Manchester Business School), John Hawgood (Durham University) and I (LSE) became interested in developing a tool which could be used by managers to choose between alternative views of what systems requirements really were and alternative methods of meeting the requirements. We developed a Decision Support System called BASYC based on the notions of multi-objective, multi-criteria decision making to be used for that purpose. An important insight gained from experiments with our system with savings banks was that the system enabled a group of decision makers to thoroughly explore the decision space and in doing so to surface often hidden assumptions. The process involved in using the DSS was as important as the numbers produced by the DSS (Land, 1975; Hawgood and Land, 1977).

I subsequently became interested in Executive Information Systems (EIS) and whilst at the London Business School developed an executive course in which EIS was demonstrated with course members role playing senior executives faced with choices on which direction to take.

Q3: What were your motivations for working in this area?

Land's Response: Two archetypical positions had emerged with the growing power of computers and management science. The first, positivistic in its philosophy, has a strong belief in the power of science to model economic and business behaviour. Those who followed this line believed that decision making was best taken out of the hands of fallible human actors and computer armed with management models were the appropriate tools for this. In the 1950s, for example, Bob Deem, a management scientist working for BP, persuaded the company to let him develop a comprehensive computer system which would automate the scheduling of refinery production. Despite the ultimate failure of the system the underlying belief still has wide credence.

The second archetype has its origin in the social sciences. Amongst its tenets is the conviction that the behaviour of a system involving human actors is non-deterministic and emergent. Further, it is argued that the success of such systems requires the active engagement of its stakeholders. This would enable the Sociotechnical system to capture their knowledge, lead to further learning and provide motivation. Hence the role of the computer is to act as an assistant to, rather than as a replacement, of the human participant.

My interest was not in DSS per se, but in developing a repertoire of approaches and tools fitting in with my interest in a Sociotechnical view of Information Systems. DSS and in particular GDSS provided a mechanism for utilising the Sociotechnical precepts.

Q4: Who were your important collaborators and what was their contribution?

Land's Response: Whilst at Lyons and LEO the main collaborators where the managers of the functional units – such as the managers of the tea factory and, of course, my seniors and in particular David Caminer.

My move from industry back to the LSE led to a much greater study of the systems literature. I was influences by Steven Alter's book on DSS which gave a name to some of the ideas I had carried tacitly from my days with LEO, and enabled me to articulate them more clearly.

But the greatest influence was my collaboration with Enid Mumford and John Hawgood. This led directly to our work with the savings bank. More importantly it helped me to find a rationale for the views I had adopted intuitively from my 16 years working with LEO.

My interest in evaluation, fired by the project noted above, was continued later working with David Target (London Business School and Imperial College, London)) and Barbara Farbey (LSE and University College, London). The partnership developed a real synergy resulting in a book and a number of papers based on our joint research with industrial partners.

Another important influence was (and is) Professor Lawrence Phillips Visiting Professor of Decision Science at the LSE (see http://www.lawrencephillips.net/). Larry is another pioneer in this area. He introduced the ‘Pod’ an environment for group decision making using a variety of aids to help arrive at difficult decisions in situations where radically different solutions are initially advocated. He has repeatedly demonstrated the power of his approach.

But it is impossible to list all the people with whom I collaborated or who contributed to my understanding and learning. Sometimes a conversation over coffee with a colleague was as influential as reading a paper or a book.

Q5: What are your major conclusions from your experiences with computerized decision support?

Land's Response: The best DSS are those which provide clear explanations of the rationale behind the alternatives offered up for consideration and permit the decision makers to explore the decision space and to bring to the surface underlying assumptions and hidden conflicts. But to make the process work it needs a facilitator with an understanding of group behaviour as well as of the way the DSS is constructed.

Without the assistance of a facilitator Managers sometimes find it difficult to follow the underlying logic of the DSS leading either to the dismissal of the DSS or to the blind acceptance of the recommendations without a full understanding of the implications of the choices made. However, at their best, when designed jointly with the decision makers, they can be highly successful.

A DSS which is simply parachuted into the decision situation has little chance of being adopted. Ideally the DSS is the outcome of collaboration between the decision makers and systems designers. The way the DSS is deployed is highly dependent on the working style of individual or group decision makers. The point is illustrated in the 1986 Ph.D. thesis of Richard Baskerville when my student at the LSE. The DSS was designed to support the activities of the Admiral of the US Navy in charge of its London Office. The very successful system designed to suit the officer in charge was sidelined when he was replaced by an officer with a very different working style (Baskerville and Land, 2004).

Q6: What are the issues associated with decision support that we still need to address?

Land's Response: Note the importance of keeping the logic in line with changing conditions in a turbulent world. Too often decision makers, not fully understanding the underlying logic, rely on a model embedded in the DSS which has ceased to reflect the changed world. Designers, on the other hand, often do not ensure the mechanisms are provided for the rapid and easy updating of the models underlying the DSS.

The importance of the informal systems which run though most organisations. These often are more information rich than formal systems, which are restricted in the information they can gather. The importance of informal systems and their role in decision making is often neglected by systems designers.

However, developments in the use of the internet such as Web 2.0 and the ideas behind the open source movement are permitting the informal to infiltrate computer-based systems.

Perhaps most importantly we need to further improve our understanding of how decisions are made and the role played by non-instrumental issues such ‘office’ politics, human relations and intelligence.

DSS References

Baskerville, R.L. and Land, F.F. (2004), Socially Self-Destructing Systems, in C. Avgerou, C. Ciborra and F.F. Land (eds.) The Social Study of Information and Communications Technology: Innovation, actors and contexts, Oxford: Oxford University Press, pp. 263–285.

Hawgood, J. and Land, F.F. (1977), The BASYC Approach to Planning, Evaluation and Designing Computer-based Systems, paper presented at Informationsforum: Die Wirtschaftlichkeit von Informations – und Kommunikations-systemen, Cologne, December.

Land, F.F. (1975), Evaluation of Systems Goals in Determining a Design Strategy for a Computer-based Information System, Computer Journal 19.

Simmons, J.M.M. (1962), LEO and the Managers, London: Macdonald.