The information systems literature recognises that the term ‘information system’ is a broad one and throws up a number of different definitions. These definitions fall loosely into two categories: those that have computerised technology artifacts at the centre of the system and those where systems are not necessarily computer technology dependent. Definitions in the former category typically describe an information system as being ‘any organised combination of people, hardware, software, communications networks and data resources that collects, transforms and disseminates information in an organisation’ (O'Brien, 2003) or ‘people, data/information, procedures, software, hardware, communications’ (Benson and Standing, 2002). Those who subscribe to the second position include similar elements but without the idea of technology being an essential component. Stair and Reynolds (2003), for example, define an IS as a ‘set of interrelated components that collect, manipulate and disseminate data and information to provide a feedback mechanism to meet an objective’. Similarly, according to Laudon and Laudon (2006), an IS is ‘to support decision making and control in an organisation’. Most other approaches to information systems theory propose that all information systems exhibit certain basic features, notably input, processing (which produces output) and feedback. The input stage involves ‘the gathering and capturing of raw data’ (Stair and Reynolds, 2003), the output involves ‘producing useful information’ (Stair and Reynolds, 2003) from the processed input, and processing involves, ‘converting or transforming data into useful outputs’ (Stair and Reynolds, 2003). Finally, feedback ‘is output that is used to make changes to input or processing activities’ (Stair and Reynolds, 2003). In this way of thinking, a system that includes the interrelated components that perform these tasks is perceived as an information system. A common view also adds the idea of self containment in that users work with the system but without direct utilisation of the real world to which the system refers. Rather, they work through a model or abstract representation of the real world (Wand and Weber, 1995). So, for example a system participant can make a decision about action through reference to an inventory database and does not need to look at the stock on the shelf. Within the literature there are also definitions that stress the importance of overall goals and define information systems as systems that provide the impetus for activity (e.g. Goldkuhl and Agerfalk, 1988).

In comparing different types of systems such as traditional computerised information systems and routine, manual systems, implementation details are necessarily different. Consequently, in examining these definitions, we argue that it is valid to ignore those differences or variations that are concerned with implementation. However, four essential concepts remain common to other definitions: input and output, processing (which produces the output) and feedback.

Data inputs are the facts that are gathered and captured in the system. There are many definitions of data in the literature but they are all essentially about the projection, or communication of facts. As examples, data has been said to be ‘facts, concepts or derivatives in a form that can be communicated and interpreted’ (Galland, 1982) and a ‘representation of facts, concepts or instructions in a formalised manner suitable for communication’ (Hicks, 1993). Other definitions also include the idea of facts being available for processing (Laudon and Laudon, 2006; Maddison, 1989; Martin and Powell, 1992).

Processing is often considered to be the way data is manipulated, developed or built upon in some way that transforms it to create meaningful information. This is what is contained in the idea of a system itself. Systems theory, an umbrella theory within which information systems theory fits, considers the idea of transformation, which describes the structure of change in natural systems (Land, 1973). Land uses the term ‘transformation’ to describe how participants in a system negotiate meaning. Faced with signals from their environment, they define and redefine what to do next, repeating successful approaches. Thus, transformation incorporates the notion of processing.

When raw facts are transformed by system processing, output is produced that signals or communicates to participants in the system. Essentially, what defines output in its role as an information systems component is its indicative status, that it signals or projects itself to be acted upon in some way that has value to systems participants. It is the output signalling to users that leads to action. That is, participants react to processed facts and take action for as long as this approach is perceived to be goal attaining.

Feedback occurs when a user responds to the output in such a way that the system input is altered. In traditional information systems this may involve deliberation. In routine, manual systems, where there is a reactive response to the output, feedback can ensue without the cognitive activity that deliberation entails. Rather a routine response occurs, which has been learnt from earlier experiences. Either way the output triggers a response in the system participant as a guide and precursor to the feedback activity.

Where there is action it can be presumed that it has resulted as a response to output. It is this action and its effects on system inputs that is significant and keeps the system functioning. So while feedback is the traditional term used its significance in a system results from the action it promotes. The primacy of action in information system is reinforced within semiotics where it is claimed that ‘Information systems should be conceived as … systems intended for action’ (Goldkuhl and Agerfalk, 2000). Information systems are also seen to ‘exist to support directly those taking the action which results from the formed intentions’ (Checkland and Howell, 1998).

This focus on ‘those taking the action’ is an important element of feedback in an information system. It is suggested that information systems have a ‘social dimension’ with problems they try to solve being ‘people centred’ and involving human participation (Benson and Standing, 2002). Also, many definitions of information systems include people and the procedures followed by people, as essential components (Benson and Standing, 2002; Boddy et al., 2005; Stair and Reynolds, 2003). This involvement of people in information systems feedback and activity is crucial.

In general systems theory, all systems include input, processing output and feedback, which creates a relatedness between the parts (von Bertalanffy, 1972; Donde and Huber, 1987). Complex systems such as control systems (for example a thermostat) include these elements. These systems are not, however, regarded as information systems since they do not involve a human or intentional agent who recognises the informational value of the output. In the systems literature there is much discussion of control systems, such as the flyball governor, a system developed in the late 18th century to automatically maintain steam engine speed despite changes in loads and steam supply. This system involves two balls connected to a shaft. The balls rotate in response to the steam supply, causing it to cut off the steam with increased speed and to open the valve as the shaft velocity slows. While this is an example of a system with input, processing producing output, and feedback, it is not an information system without a user to recognise the informational value of changes in the steam supply. Information systems provide information to someone or something; they are not just self-operating control mechanisms. In an information system, human involvement is evident. In traditional information systems the distinction between systems in general and information systems is more obvious because in the latter the informational aspects are separated off into the computer, which provides output reports or summaries or calculations for the user to respond to. In routine, manual systems many of the artifacts in these systems (such as magnets on a whiteboard) are not obviously sources of output information until they are placed in a certain way or set in a particular physical context. In these contexts, participants perceive the informational value of the artifacts and make a response to this value. This response requires some human perception of information content. Even where the action is routinised, it will have eventuated from a routine learnt through an earlier process where information was provided that imbued the routine with value for the human participant.

Considering all of these issues, we propose that, in deciding whether or not a system is an information system, the following test should be applied:

In deciding whether or not routine, manual systems are information systems, through examining them alongside traditional systems, we extract examples of elements from both types systems to populate Table 1 below. In this table, the four common elements — fact, transformation, signal, and action — that we claim to be characteristic of all information systems, are provided in the second row. In the rest of this paper, traditional and routine manual systems will be examined in order to populate rows four and five, which are those aspects of the two systems that manifest themselves as facts, transformation, signals, and action. Row five will contain examples of this manifestation as they are uncovered.