This document is under construction

Co-ordinating Representations: the Flow of Work in Civil Engineering

Mark Perry

Department of Computer Science and Information Systems,
Brunel University,
Uxbridge, Middlesex,
UB8 3PH, UK.
Tel: (+44) 1895 274000 x2136
Email: Mark.Perry@brunel.ac.uk

Abstract

CSCW has long been involved in developing an understanding of the role of artefacts and the representational status that they carry with them, enabling the co-ordination of distributed work activities. Despite this interest, little headway has been made as to how this can occur because of the huge range of influencing factors that give these representations meaning - covering aspects as varied as individual cognition, social and organisational (or cultural) influences. This paper examines how a construction company uses representations, both formally and informally to organise collaboration amongst teams of workers incorporating all of these influences, contrasting the role of organisation with the media used in representational transformation. The implications of this study for design are discussed in terms of the co-ordinating role that representations can play.

Keywords: Civil engineering, CSCW, collaboration, representations.

1 Introduction to Construction

The construction industry has traditionally been slow to adapt to technological change due to the lack of technology robust enough to survive the rigours of the environment and low expectations of the work force by management. Recently, technological innovation has made hardware robust and portable enough for use on site and developments in interface design have made computers more accessible to users. However, it was felt that it was important to understand more about the process of construction so that the tools developed would be appropriate to the problems that civil engineers face on the ground. We need to understand the work itself before we can judge the effects of technology upon it.

An ethnographic study of civil engineering was conducted: a team working on a bridge building project was studied and using interviews, document collection and an ethnographically informed observation, data was collected about the group, the task, the tools used, and the organisational context.

The team was organised into distinct roles, determined by training and experience and was made up of (by seniority) the team leader, the senior engineer, middle rank engineers, junior engineers, foremen, gangers and labour. Additionally, quantity surveyors were involved in the financial management of the construction. Down to the level of foremen, these were all co-located in a open-plan office.

The work of construction involves the combination and transformation of mathematically derived CAD drawings (detailed plans) and schedules (dates for task completions) into simple and unambiguous instructions for the labour force. The initial instructions are computed by the senior engineer and passed on to the junior engineers who pass this on in more specific terms to the foremen who distribute it on to the gangers, who work on site supervising the labour. Thus the complex representations are converted into a simpler set of sketches, representing only the required information, performed at the times specified by the schedule, and finally, down to the simple spoken instructions to the workers on the site. At each stage, work is decomposed into more manageable and simpler units. The team leader and middle rank engineers act in a support role, engaged in liaison, procurement, troubleshooting and in an advisory capacity for the other team members.

2 Abstracting Construction - Planning and Design

The initial state of the team involves a set of specifications that must be transformed through planning and actions into constructions that conform to these plans (the end state). In a sense, the process is broadly computational with information being processed by the tools and people in the group and transmitted to other information processing units:

[diagram]

What occurs in the process is that the complexity of plans decreases over time, from a highly abstract set of complex instructions into simpler instructions set in natural language and diagramatically. As the information moves through the chain of command, it changes from a formal state into more fluidly interpreted descriptions, as plans are implemented and representations become interpreted as actions. What was once abstract has now become 'concrete', in more than one way. There is a stage where use of the representation is suspended and environmental constraints take over - the representation cannot be the only means of planning a construction because eventually reality intrudes onto conceptual design. The 'wants and needs' of the designers are locally interpreted by workers reacting to the conditions that they find on site and contingencies encountered that could not have been predicted (Suchman, 1987).

This 'downward' flow of information is the way that civil engineering is often perceived. However, construction work is not all top down - problems arise with the local conditions and the implementations of the plans, which can occur at all stages in the process. Feedback procedures are required to pass this information back up the 'chain of command' so they can be acted upon and the plans adapted. How these disruptions are co-ordinated is vital because the mechanisms through which these occur are poorly understood and are generally organised informally. An information management mechanism that is designed for construction workers will need to support these processes, so it is central that they are understood and can be incorporated into a model of collaboration.

Planning and design are not completely distinct from the process of implementation [see note 1]. As the design moves towards its implementation phase, design does not cease; rather it continues at a more situated level, where environmental constraints must be considered and adapted to, where previously these had been more abstract and theoretical. The process of design has therefore been 'opened up' to examination by implementation. Problems are filtered back through the organisational structure so that designers (should) receive appropriate information and adapt their plans accordingly. The design activity may take place in any of the hierarchical levels - construction workers solving a minor problem, team engineers exploiting their local knowledge and their engineering skills to fashion a more advanced solution or the conceptual designers amending the plans for a complex problem. Deciding who deals with problems is a socially organised phenomena - but successful resolution is usually dependent on this - too simple problems going too far up the hierarchy could cause a loss of contextual information or information bottlenecks, and complex problems being solved by too junior personnel could lead to catastrophic failure as the knock on effects made structures unstable.

3 Representational Media and Function

The table below details the forms of communication used between the team members:

[diagram]

NB. Shaded areas were not investigated extensively.

One of the most important features of the communication process above is general perceptual monitoring (Heath and Luff, 1991) - through using representations that are available to all, a shared understanding of problems can be developed and propagated naturally up the hierarchy. These representations are necessary to incorporate into any communications system that hopes to build on this understanding. Communications also tend to be grounded in the media of the representation: conversations are related to drawing numbers, revision numbers, schedules for particular weeks and details relating to paragraphs in the original contract. These representations provide a focus for construction and provide a context for the work.

4 Representations to Co-ordinate Work

Co-ordinating representations are representations that facilitate the structuring of activities in collaborative work, that is, work requiring people to co-ordinate their activities to achieve a shared goal. These representations might exist in a number of different artefacts, generated, modified and transmitted between people, such as drawings, letters, forms, post-it notes, speech, body language and so on. Within particular situations, certain representational forms may be more appropriate in co-ordinating activities because they are unambiguous, can be quickly interpreted and processed, and passed on to the next user of that information. Rogers (1993) describes one form of co-ordinating representations as 'mediating mechanisms', and describes how a whiteboard is used to co-ordinate the behaviour of co-workers, although the board does not comprise of the work itself. Brown (1995) [note 2] describes artefacts-in-use as 'objects of collaboration': artefacts used in work processes both as mediums for getting the work done and co-ordinating that work. These enable collaboration to naturally arise from the artefacts used in work itself.

The media used to communicate between collaborating workers is central to the way that communication mechanisms can operate, as much as the organisation within which they exist. One person's output is another person's input - making these easily compatible will save time, effort and reduce the need for explanatory communication, which can be as confusing as it is illuminating.

Whilst the objects of co-ordination direct the general direction of design, mechanisms to organise the micro-structure of design activity exist to bring together the actual co-ordination of events not explicitly proscribed in the plans. These might be problems such as ambiguous, inaccurate or inappropriate plans, plans using unavailable materials, plans aborted due to weather conditions, industrial disputes, other management problems, and so on. These need to be communicated back up the chain of command, to be negotiated with the other stakeholders whom the decisions affect. The downward planning information channels cannot deal with feedback of this kind because these communications are usually structured by demand rather than planning. Whilst some formal channels do exist, such as regular meetings or problem report forms, they are often in the circumstances (e.g. too slow) or are not known to the participants (who may be inexperienced, or even illiterate).

Where formal channels are inappropriate, people use the information that they have in the environment around them to communicate, jotting notes onto drawings and leaving them on the recipients desks, pinning letters to them on notice boards and talking loudly enough so that anyone attuned to particular words might overhear them. As with the plans, they are objects of co-ordination: representations that mediate collaborative activity. Users tend to chose the medium of communication that best matches the message given their situational requirements; flexibility of media is central to this.

5 Conclusion

Social and cultural organisation allows the distribution of labour so that people can specialise in particular forms of work. It also allows them to communicate in a way that avoids duplication of effort. Thus, in a disorganised group, who does what, the methods used, the order of completion and so on, are complicated issues. Imagine a group of seven year olds being given a job: make a meal. The result is chaos, as all go about things in their own way. Organisation allows humans to concentrate their abilities in an efficient way and allocate their resources most effectively. Through structuring the organisation and the form of the representations used, work can be distributed over a team so that unnecessary duplication is avoided, individual goals are set and a single output can be agreed upon. Empowering individuals to do this with technology will not solve any of the systems' problems any more than teaching the children to cook. As any restauranteur will tell you, there is more to being a chef than cooking: the answer lies in preparation and organisation.

Distributed cognition is one means of analysing settings: Hutchins (1995) describes the socio-cognitive system of navigation - information propagates itself through the navigation system in different representational forms, from the pelorus to the bearing readers, to the bearing log recorder via speech, thence to the bearing log, to the helmsman and finally marked onto the chart. Construction can be described similarly; however, it is not as methodologically formalised or regulated and therefore require a great deal more communication to assess quality, accounting and safety control. Construction systems are not so much formal systems, as informal ones, and as a consequence require more maintenance and supervision.

Representations are used to co-ordinate work so that each individual knows where to get their inputs, what to do with their outputs and has the ability to understand the information that they receive so that they can act upon it. Simply concentrating on the work of the individual cannot reveal details of the system, although this is the area that most research has concentrated on, with the development of CAD, accounting and schedule planning tools. We must look beyond the study of interaction with a single tool and consider how different objects are used in conjunction with each other. The reality of construction is that each person is but a cog in the system: understanding the properties of the cog cannot not reveal the properties of the machine. What brings the metaphorical cogs in contact together at the correct speeds and ratios are the representations used as inputs and outputs, and organisational structure involves placing the cogs in the appropriate order to do work.

References

Brown, B. (1995) Transition report.
Heath, C. & Luff, P. (1991) Collaborative activity and technological design: task coordination in London Underground control rooms. Proceedings of ECSCWÕ91, Amsterdam, 65-80.
Hutchins, E. (1995) Cognition in the Wild. Bradford: MIT Press.
Robinson, M. (1993) Design for unanticipated use..... Proceedings of ECSCWÕ93, Milan. Kluwer: Netherlands.
Rogers, Y. (1993) Coordinating computer-mediated work. Computer Supported Cooperative Work, 1, 295-315.
Suchman, L. (1987) Plans and Situated Actions. Cambridge: CUP.

1 This is more likely with physically situated work than it is to be with intellectual work, such as software engineering, although there are parallels - such as new technology breakthroughs making the impossible possible halfway through projects, in copywrite disputes, or personnel loss.
2 Drawing from work on common artefacts by Robinson (1991)