There is a commonly held misconception about computers underlying current software engineering methodologies. The misconception is that computers process "information". Information processing implies the following three steps: "getting input", "processing this input following an algorithm", and "delivering an output". Input and output are conceived as variables (in the sense used in programming languages) which contain the "actual information". If "actual information" is taken to mean "data prepared for specific purposes" there is nothing wrong with this view. But then it would be more appropriate to use the term "data processing". However, the term "information", as used in our everyday language, has a very different meaning. It is this meaning of "information" that is socially relevant. Humans deal with information. We have to take great care that we do not transfer the restricted sense of information processing (namely data processing) to characterize human behavior and human expertise. Conceiving computers as data processing machines provides a realistic framework for assessing their capabilities. We will demonstrate in what ways the capabilities of computers and humans differ.
The problem in using the information processing metaphor is that the contrast between computers and humans tends to somehow "fade away". This becomes seriously problematic if the same expression is used for the understanding of both the computer's capability and of the human expertise. If the expression "information processing" is used for the description of both computers and humans, then both are understood essentially as data processing systems. This is precisely the view underlying expert systems (see below). But the view of humans as information processing systems is widely held in the society at large, not only within the expert systems community. We do not think that the "information processing" metaphor is a convenient description of expertise. Consequently, we also have to revise our concepts of human expertise. The aim is the same: to improve the effectiveness of the design process. The idea is simple: if we have a better understanding of the differences between computers and humans we can define the tasks that can and should be performed by computers, and we can sort out those that should remain under human control.
As an alternative we propose a description of human behavior which can be subsumed under the term "situatedness". We will argue that if we describe humans as "situated agents" rather than information processing systems we will be able to better accentuate the distinctions between human expertise and the data processing capabilities of computers. We have borrowed the term "situated agents" or "situated actors" from Lucy Suchman's seminal book on "situated actions" (Suchman, 1987).
We will proceed in three steps. Firstly, we will reconsider the computer's data processing capabilities by capitalizing on the difference between data and information. In what follows, we will use the more general term representation instead of data. Data are specific instances of representations. The distinction between representation and information has to be made with reference to our socio-cultural world. This sets the scene of the second step. Here, our goal is to emphasize that our expertise is strongly based on our ability to communicate. If we are not told what a red traffic light means, we are not able to use it as a decision aid when crossing a street. This phenomenon will be discussed in terms of so-called "socially shared knowledge". Finally, to emphasize the individual skills of humans, we will briefly talk about the term "situatedness". We will begin by discussing the need for change in the field of "software engineering". We propose the title "work place design by means of computer technology" instead of "software engineering" because the meaning of the term "software engineering" is too restricted.
One way or another these fundamental problems all relate to the distinction between real world vs. virtual world (e.g. Pfeifer & Rademarkers, 1991). This issue in turn pertains to the question of how rationalistic descriptions relate to real world phenomena (Winograd & Flores, 1988). If we acknowledge the indefinite richness and the dynamic nature of the real world in contrast to the predefined static nature of a virtual world, and if we question the suitability of rationalistic explanations to explain the phenomena of mind and consciousness, we begin to realize that we have to take the fundamental problems listed above seriously into account.
Müller, M. & Pfeifer, R. (in press). SITUATED DESIGN IN A LARGE PAPER MILL: Developing effective computer systems supporting knowledge intensive work. Will appear as a chapter in: Khosrowpour, M. & Liebowitz, J. (eds.): Cases on Information Technology Management in Modern Organizations. The book is scheduled to be published by Idea Group Publishing in Winter 1997.
A short list of Publications you can find here.
[HOME]