This section examines some organisational and managerial implications of distributed research communities. It concurs with the argument of Orlikowski and Yates (1994); that existing organisational literature has not systematically addressed the relationship between new communication technologies and changes in work and interaction. Much management and organisation research has taken place, either in formal organisations with relatively stable structures, or it has taken place in groups of people, where face-to-face behavioural effects are the focus of analysis.
Distributed research communities can be characterised as complex hybrids of formal organisations, on the one hand, and groups on the other hand with their own peculiarities. For example, whereas they could be seen as 'groups', in the sense that they have a limited, defined membership (Alderfer, 1977), they communicate relatively little face-to-face, hence much behavioural theory derived from research on groups is likely to lose some of its validity. The frequency of face-to-face contact between members corresponds probably more to what is found in large, formal organisations. They have also 'inherited' characteristics from formal organisations, such as a certain structure, formal power relations, resources, budgets, etc. Where they distinguish themselves from formal organisations, however, is at the level of organisational boundaries and membership characteristics. Whereas formal organisations exhibit physical boundaries (the confines of a building, etc.), the boundaries of distributed communities may be defined by electronic membership, or defined by intellectual/social cohesion. Also, whereas membership characteristics of formal organisations tend to be exclusive (e.g. "you cannot work for Coca Cola while at the same time be a staff member of Pepsi Cola") and contractual, membership in a research community may be one of several types of organisational attachments of researchers. The latter is a characteristic which distributed research communities tend to share with groups. These distinctions are general and will not be true in every case. They are, however, valid for a great deal of research in the organisation and management area (Nohria & Berkley, 1994).
With increased use of computer technology and needs for improved responsiveness in rapidly changing markets, a growing number of formal organisations are moving towards structures that resemble more distributed communities (Management Today, 1992) than the traditional 'Fordist' types of structures. Hence, the distinction may gradually become less clear. However, distinct shifts towards community-like structures in formal organisations are not yet there and may take a long time to materialise.
Although distributed communities may in some respects be seen as hybrids of the two above-mentioned structures, it is unlikely that the studies of distributed communities will yield any kind of 'average' data between formal structures and groups. The use of electronic media as the main medium of communication alone is likely to produce insights to which neither of the other two types of structures lend themselves. It is useful, however, for the sake of understanding, to use perspectives which are commonly used in the organisation and management field. Therefore, in the following discussion, particular attention is paid to the issues of organisational structure and management/leadership, which have been central concerns to organisation and management researchers and practitioners, and which are also important for the success of managing and participating in distributed research communities.
Hughes and Christie (1994) suggest that research communities:
" . . . may be seen as alternative forms of organisation, or alternative forms of relationships between actors, to markets, and to hierarchies or fixed organisational structures." (p. 2)
They do not pursue the types of differences that may exist.
It may be suggested that four characteristics distinguish distributed communities from other forms of organisation and these are reviewed below.
Whereas a formal organisation tends to be, by definition, a problem-solving enterprise, such as production of goods or services, a distributed research community may exist primarily for the sake of exchanging knowledge between members. Section 4.1 of this Report refers to Hughes and Christie's typology of interdependence between members, which includes communities in which members' projects are independent of one another. This means that some communities may be regarded as organisations in which the operational interdependence of members is low, i.e. to borrow Carl Weick's term, a high degree of "looseness of coupling" (Weick, 1976).
Members of distributed research communities tend not to belong exclusively to the community, but also belong 'somewhere else'. For example, although they may adhere to a community, they tend be employed by a parent organisation. They may also be members of several communities at the same time. This suggests that their membership of the community may be volatile, and largely based on what they perceive as being the benefits of being a member. A further consequence is that whereas in formal organisations members may abide by rules that they do not necessarily agree with, in a distributed research community, that may result in leaving the community, or remaining a passive member. This has significant effects on the ways in which members' motivation might be understood and catered for in the community.
Distributed research communities tend, by definition, to be temporary organisations. Although the duration may run from a few weeks to several years, their duration is normally definite, and for a large number of them, pre-determined. There may be no pre-existing 'organisational memory', codes of conduct or established communication patterns which dictate how members should fulfil their functions or relate to one another. Such patterns need to be created by each community in a short span of time.
Whereas in formal organisational structures, electronic communication is an auxiliary medium to face-to-face interaction, distributed research communities use electronic communication as their main medium, with face-to-face interaction being auxiliary. This has a number of implications. For example, the effects of communicating through an electronic medium need to be understood, such as the difficulty of making certain types of decisions referred to in Section 4.1. Also, the possibility that face-to-face interaction takes on another role has to be allowed for in distributed communities.
The list includes, incidentally, points very similar to those suggested by Nohria and Berkley (1994, p. 115). It is by no means exhaustive, but it raises a sufficient number of peculiarities of distributed research communities warranting special study of how aspects of management and organisation are dealt with in such organisations.
The three communities studied by the Project constitute a limited sample and it is therefore appropriate to signal a warning to the reader that the discussion which follows may be limited in scope and validity. Nevertheless, the findings are thought to be significant, as they have been identified from a wide background of knowledge in organisation and management theory and practice.
Broadly speaking, debate about organisational structure reflects two opposing views, which by Burns and Stalker (1966) have been termed respectively 'mechanistic' structures and 'organic' structures. Although since then, a great number of variations have been proposed (Mintzberg, 1979; Handy, 1989; Quinn, 1992), the two basic categories suggested by Burns and Stalker have stood their test of time.
Mechanistic structures are largely functional, i.e. with formal reporting lines, clear distinction of responsibilities, fixed goals and centralised decision-making. The main emphasis tends to be on order and predictability. In organic structures, on the other hand, the structure is recognised as being more chaotic, with partly unclear reporting lines, overlapping responsibilities, participative decision-making and evolving goals. Management in organic organisations tends to emphasise personal initiative and organisational adaptability. Although each organisational type has its advocates, each type of structure has its relative merits, depending on the type of task it is intended to achieve.
In order to simulate similar differences in structure, experiments have been carried out on groups where member interaction has been structured, either in the shape of a 'circle' or as a 'star'. The circle represents peer-to-peer networks and the star represents hierarchical networks. The parallels with 'organic' and 'mechanistic' structures respectively are evident. Findings from experiments and analyses concur on a major point; that whereas hierarchical networks (stars) are faster and more controlled, peer-to-peer (circle) networks are more creative and adaptable to change (Leavitt, 1972).
Observations of groups interacting through electronic communication are similar. Finholt, Sproull and Kiesler (1990) argue that a 'spoke and web' pattern (i.e. a star shaped, mechanistic structure) is useful for relatively simple tasks, where a central person is the most competent member of the group and can collect all the information necessary for the task. On the other hand, a 'circle pattern' (i.e. circle shape, organic) represents 'extreme decentralisation', and is suitable when competence is evenly distributed in the group and when information is too complex or too burdensome for a central person to handle.
Such findings are instructive for understanding semi-permanent structures, such as companies or universities, who have relatively consistent structures over long periods of time. However, as suggested above, distributed research communities have finite life-spans, over which the nature of the task may change rapidly. In Case Study 1, for example, the project group went through different phases requiring different management structures over a time-span of about one year (Section 4.1). If, as Stanchev (1995) argues, research groups may swing between extremes õ from highly centralised to groups of absolutely independent researchers õ a series of issues present themselves, such as how should the group arrive at the decision that a change of structure is appropriate? It seems that the ability to change between two fundamentally different types of structures may be a requisite which distributed research communities have to satisfy, but which permanent structures do not have to live up to, to the same extent.
In this section, Stanchev's terminology of 'full-graph' structure and 'star' structure are used to distinguish between the two main (extreme) types of structures already discussed.
The Case Study 1 project exhibited an evolutive structure, which may be considered in three phases. In Phase 1 (the start-up period), much of the decision-making power was with the Project Coordinator and some Senior Researchers. This was a natural condition, because they had most knowledge about the background and context of the project. In Phase 2, which emerged after two months, the Fellows were formally given a large share of decision-making power, as it was recognised that they exchanged, and hence represented, a major part of the knowledge which was evolving in the project. In Phase 3, which became apparent about eight months into the project, the Project Coordinator started initiating a series of operational decisions, which were also followed up. Hence, using the 'star'/'full-graph' terminology, the community started with a 'star' type of structure, moved to a 'full-graph' structure, to then return to a 'star' type structure.
Considering the description from Section 4.1 of how the task of the community changed over the three phases, a relationship (such as given in Table 5.4) may be seen.
Table 5.4 The evolving structure in Case Study 1
Phase 1 Phase 2 Phase 3
Nature of task · Initial structuring · Building shared
· Operational
· Introducing Fellows understanding
· Conceptual development
Type of structure 'Star' 'Full-graph' 'Star'
It is important to note that the analysis of the structure as reflected in Table 5.4 was done in retrospect and is based on an impression of how it was generally perceived. There were exceptions. For example, one member thought that although the structure at times appeared to be of a 'full-graph' type, there was an implicit, informal structure which replaced the formal one:
"Some people are implicitly in charge. Sometimes there is discussion between a Fellow and the Project Coordinator as being the two key leaders; also because they have a closer face-to-face relationship."
There was disagreement about the type of structure which was most appropriate. Some advocated a more centralised structure:
"I don't know exactly why (the Project Coordinator) hasn't taken a leadership role"
Others thought that too a rigid structure was not necessarily appropriate:
"The second meeting felt different, (the Project Coordinator) was more clearly anxious to get things shifting, he was anxious to get agreements and gave the impression that he felt that time had been wasted. I don't agree with this. It is just one way of approaching research."
The Case Study 2 project had one person, not the Project Coordinator, chairing most of the meetings. The person taking this role changed but initially was someone recognised as the specialist in the concepts and in methodology which were central to the project but with which none of the others were very familiar. This person came to take on a central, conceptual guidance role. Hence, in relation to a 'star' structure, the role of that person may be termed a 'conceptual node'.
At a later stage of the project, when the focus moved to operationalising the new shared concepts, the 'chair' became a member of the team who was close to the implementation context. During the times when most of the 'routine' work of the project was being undertaken between face-to-face meetings, this 'chair' acted as the moderator of the computer conference in which the negotiation was taking place. This responsibility rotated on a weekly cycle between the four institutions involved. The initial 'guru' remained strongly visible in the community but took a lower, still central, profile as 'consultant' to assist in resolving conceptual ambiguities.
This suggests that in a community, the role of a hierarchical decision-maker is not the only 'node' of the group, but that other sources of influence, such as being recognised as an expert, may equally well be perceived of as a 'node'.
In the Case Study 3 community, it seemed that a regular structure was in place from the beginning, where members had formal hierarchical roles. This observation transpires from the way they described their use of the telephone for making decisions on urgent matters, where a member would call upon his (or her) superior to facilitate quick decision-making. Thus, the structure of this community resembles a more 'mechanistic' type than the other two communities that were studied. There are two factors which account for this.
Firstly, the need to make a large number of operational decisions within tight time limits, which would naturally demand that a decision-making structure was in place. Secondly, the clarity of the goal (i.e. to build a physical device together), which was not questioned would seem to favour a tighter, more mechanistic structure than the other two communities in particular, where the understanding of the goal or methodology needed to be reviewed in order to be able to work effectively together.
It is possible to compare the three communities beside one another, as is done in the table below.
Table 5.5 Comparison of the different structures of the communities in the study
Community Nature of work Goal Structure
VMDL Changing from Defined overall but with Evolving between 'star'
conceptual, through flexibility offering the chance and
'full-graph' and back;
methodological, to of negotiation between Changing roles
over
operational; members, evolving; three different phases.
European Project Innovative concepts Overall relatively Different
types of 'nodes'
Community requiring reformulation to unambiguous but with
(administrative and
meet key responsibilities methodological and political conceptual).
of team members; uncertainties;
A community of Highly operational, Given from the start, Stable
, defined hierarchical
Physicists shared understanding unambiguous, unchanging; roles.
between members.
As might be expected, Table 5.5 illustrates that the structure of communities take different forms. However, it is difficult to say whether the communities in Case Study 1 took on the forms naturally in response to the tasks they were meant to perform, or whether there were other factors influencing the form they adopted.
The data from Case Study 2 suggest that the ideas of 'star' and 'full-graph' are in practice more nuanced than what research generally suggests. For example, a community may have one or several 'conceptual nodes' (i.e. recognised specialists), while having a full-graph decision structure. It may also be important to consider that although a structure is explicitly defined in a certain manner, members may perceive it as functioning implicitly quite differently.
The data from Case Study 1 suggest that not only was there divergence between members as to what should be the appropriate structure, but also about the kind of structure the community actually had. This is not a novel finding in the studies of organisations, as the idea that different members may perceive similar organisational phenomena quite differently (Weber, 1964), is well established. However, considering the often-limited life-span of distributed communities and, by consequence the importance of 'getting things right, together, on time', it seems that the importance of having a shared perception among members of the nature of the structure is crucial. From this, it might be inferred that it is important for members to try and agree what organisational structure might be most appropriate, not just at the beginning, but at successive stages during the life of the community.
Because the nature of the community's task changes over the life of the community, the required type of structure will also change. This means that the structure cannot be assumed to remain constant with the important implication that there needs to be an on-going analysis of the existing structure, in which the community members are involved. In a community with a short lifetime, a slippage of some weeks in determining the appropriate structure might effect the output of the community considerably. In contrast to more permanent structures, where much longer delays may be allowed before the structure is changed, this shows the importance of sensitivity to needs for rapid structural change in volatile, distributed research communities.
In this section some reflections are made about leadership and management aspects of the communities studied. There will be a particular focus on the following two questions:
· how were leadership roles created and maintained in the communities?
· how were management decisions made?
The reason for distinguishing between leadership and management is that they represent conceptually two quite different phenomena. The term 'leader' may be taken as a role which is not formally defined, but which is (often implicitly) attributed to someone by other members of a community on the grounds that the person generates commitment towards certain goals among members. 'Management' on the other hand, should be seen as more of a formally allocated role in which a person is primarily responsible for making sure that the goals of the organisation are achieved. Maybe for this reason, 'management' tends to be seen as a more functional, operational role, and leadership is seen more as being symbolic in nature. Probably for the same reason, 'leaders' are seen more as generators of culture in a group or in an organisation (Bennis, 1989).
Of course, this distinction does not exclude the possibility that ''managers' can also be seen as 'leaders', which will be the case when the personal qualities of a manager are instrumental in generating commitment among members.
This distinction may seem unnecessarily academic, which it might be if the object of our study were large, formal organisations with given management structures. However, as pointed out above, distributed research communities distinguish themselves from formal organisations, partly by the commitment of members to be guided by personal goals and social norms. Therefore, management in distributed research communities is likely to rely much more on consensus than management in large, formal organisations. Not only does this concur with a number of observations, such as those made in voluntary organisations by Rothschild-Whitt (1979) and predictions of 'knowledge' organisations (Handy, 1989), but interviews in the VMDL case study communities point in the same direction. A statement made by a member of the one community underscores the point quite well:
"Making decisions is difficult in science - it is not right to cut through - we have to work it out."
In this section, the data are drawn mostly from the VMDL community rather than from the other two which were studied. The reason is that the proximity to the VMDL community over a long period allowed a deeper study to be conducted.
Influence (or leadership) in a community follows from the adoption (explicit or implicit) of an overall structure for the activity. As discussed in section 5.3, the role and perception of role of individual members of the community is a key dimension in establishing influence.
Most, if not all, research communities are created due to the initiative of one or more researchers. These initiators naturally take on an leadership role as the work gets underway. However, whereas this position usually remains in place in 'star' structured communities, the opportunity for changes in leadership in a community with a 'full-graph' structure is open to individual initiative or a collective agreement to share responsibilities. In either case, the leadership role will only be accepted by the community if and when the potential leader demonstrates certain qualities. These may be related to academic expertise, personality or simply the willingness or freedom to devote more effort to the work than others in the team. However, whatever the qualities of a team member, the overriding factor is the trust and integrity of the person involved: it must be clear that the team's success is the prime motive for taking on the leadership role.
It has been suggested that the use of electronic communication media removes (or minimises) hierarchical structures in communities, but research reported by Mantovani (1994) suggests that this is a myth and existing structures are reinforced rather than changed by the use of the new channels of communication.
In Case Study 1, it may be seen that leadership was largely a function of possessing particular knowledge which was of importance to attaining the goals of the community. As reported earlier, the influence shifted over the three phases (Table 5.1.1), and that it was in the middle phase that influence was greatly exerted by the Fellows. It was mentioned in section 4.1 that the decision was made to give more influence to the Fellows because the volume of knowledge specific to the project was largely concentrated among those Fellows. It is suggested that it is beneficial to the community to allow authority to members who, at a given time, possessed important knowledge while at the same time acknowledging the insights that experience brings to a task.
A similar situation occurred during the work in the community of Case Study 2, but for different reasons. In this case, pressure of time prevented a key team member from participating fully during an important, formative stage of the work. Informal leadership was assumed by a team member (a peer) who was able to commit more time to the work and this created a situation which took some time to resolve. It is suggested that the exposure of this assumed leadership to the whole team through the computer conference exacerbated the problem. There was no malice intended and it is difficult to propose ways of avoiding such situations in 'open' research communities.
However, these examples point to the importance of social processes in the distributed communities. The simultaneous presence of knowledge and decision-making authority among a limited number of members can lead to a feeling of exclusion among other members, who were not equally involved in community activities. Whereas this feeling may not always be made explicit, it can be seen in messages to the computer conferences. For example:
"The Fellows are part of the team; they do not define the team and should not exclude others on the team. Their views are not the only views of the Project."
An interesting question to ask is: how do such situations arise in which part of the community comes to feel excluded?
One cause seems to be that it is difficult for members of a distributed community to always be 'in step'. Whilst it is natural and positive for the achievement of the collective goals, for individuals or sub-groups to 'surge ahead', there is a real danger that those bringing up the rear become disillusioned and even antagonistic, perhaps holding the view that the 'surge' is in the wrong direction or is 'reinventing the wheel'.
One case study illustrated the need for a sub-group with supposedly 'more appropriate' knowledge not only to exercise this knowledge, but to realise that they have developed a shared language which makes this group increasingly difficult to penetrate for other members of the team. This situation may be particularly difficult as the shared understanding is likely to be tacit and, therefore, hidden even to the members who have it. In one particular case, a visual model was created during an informal gathering: attempts to share this with the rest of the research team through text-based electronic conferencing failed as the concepts and language associated with them were fundamentally visual.
If such problems are to be minimised, it is necessary that those who come to share more implicit knowledge than others, and practice this knowledge in moving the community forward, take extra time and use appropriate media to explain to other members about their plans and their activities. It is essential that they become aware of the 'exclusive' knowledge they share and take steps to ensure that this does not create a barrier to other members' participation.
The 'other side of the coin' is that a sub-group which has developed 'privileged knowledge' may get over defensive about their body of knowledge, particularly if they have spent considerable time and effort in creating it and obtaining internal agreement about it.
The issue of initiative taking and leadership raises again the issue of trust in the community. It seems clear that in the absence of trust between members, fear of others taking the community in a direction which one would not have chosen is likely to exacerbate conflict and put the quality of the community's work at risk. Distributed communities seem more at risk than formal organisations, partly due to the fact that so much of the communication is transient and takes place in the absence of social cues, allowing more room for speculation and lack of trust in other members' intentions. Sabel (1990) points interestingly to the volatility of trust in networks, arguing that they it may oscillate considerably over time and between members. Fornengo (1988) found, from studying telematic networks in companies like Benetton and Fiat, that the more stable patterns of trust existed between members who were relatively interdependent and who had developed long-standing relationships. Thus, trust is not just volatile, it may also be highly localised between certain members. Overall, to be trusted by members of the community requires a specific effort to demonstrate a commitment to the collective goals.
Leadership roles may thus be created, not only by members possessing particular knowledge, but also by the way they behave. In one of the case study communities, for example, leadership roles were perceived as being quite unclear. It was mentioned above how one member was seen as a kind of 'conceptual node' and later another member emerged as a kind of manager.
One of the members described it like this;
"That was the problem. We needed a manager, and we didn't have one."
But then one emerged:
"(She) was very much present, without having a specific managerial role. She was always there. She was not overpowering. She was friend, easy, coordinating without people noticing that that was what she was doing."
It seems that the role that person took on was useful to the community, and that it was also socially desirable. The ability to create a friendly and sometimes informal atmosphere by senior members of the community was perceived as important.
"The personal stuff, such as (the Project Coordinator) mentioning that 'we have snow in Boege!', made us loosen up a bit."
The VMDL case studies (together with some of the existing research literature) indicate that different leadership roles may serve different purposes, and they may emerge more-or-less naturally in communities, particularly when a structure is not rigidly defined from the start. The case study communities studied which had a more evolutive, 'full-graph' structure, encouraged the evolution of roles.
The data presented in Chapter 4 suggest that management decisions were easier made face-to-face or by telephone than through electronic communication. It is also suggested above that decision-making will probably have to be more consensual in distributed research communities than in more traditional structures. If we assume that these two statements hold true for a great many communities which rely primarily on electronic communication, we are faced with the dilemma of how to make managerial decisions in distributed communities when, on the one hand decisions have to made consensually, and on the other hand, the medium of communication does not lend itself to making decisions.
In an interview, the Coordinator of one of the communities was asked whether he could force decisions in the computer conference if he wanted to. His answer suggests that the dilemma is non-trivial:
"That is a good question. I am not sure."
It is worth noting that he had the formal authority to make decisions, and he did make decisions face-to-face, but that authority seemed somewhat eroded when electronic communication was used. It is also not sure if agreement between members is sufficient for decisions to be made. In the Case Study 1 computer conference, for example it was evident at different times that the feeling was that 'all we have to do is make a decision', still, decisions took time before they were made.
DeLisi (1990, p. 83) argues that in organisational decision-making:
"the highest form for executive operation is at the intersection of knowledge access, face-to-face encounter and group decision-making."
Against these criteria, electronic communication suffers two serious weaknesses. Firstly, the immediacy of feedback is not there for people to interact while communicating. Members of the Case Study 3 community converged on this weakness of the use of e-mail, pointing to the lack of interactivity during reasoning in electronic communication. Secondly, the social cues are missing, which in face-to-face situations offer instant images of people's reactions.
Still, decisions are made electronically. Some research even suggests that processes of decision-making are not substantially different between electronically connected groups and face-to-face groups (Weisband, 1992). Among the data from the Case Study 2, there is an example of a relatively important decision which was made electronically, whereas in Case Study 1 no such report was made, and in the Case Study 3 community e-mail was purposely avoided for decision-making.
Rather than discuss whether or not electronic communication is suitable for decision-making, it seems potentially more useful to explore conditions under which electronic decision-making is possible. Although there is limited data from this small project, it is suggested that further work is warranted in this direction.
The first condition may be labelled 'managerial legitimacy'. Certain members may, for certain issues, be generally regarded as legitimate decision-makers by the other members having demonstrated their commitment to the community goals. It is quite possible that given a certain legitimacy, appropriate members may make decisions which other members abide by and support, although they may not necessarily be in agreement with what is proposed.
The second condition relates to what is conceptually possible, given that in networked communities the community is manifest only in the text of the conference (Hodgson and Fox, 1995). It may be labelled the 'constraint of shared meaning'. Decisions may be made if the issue at stake is well within a shared understanding of what it is about. For example, the trouble of making decisions in the first phase of one case study project seemed to stem from lack of shared repertoire of meaning among members. Once a subgroup were able to compose physical models in informal face-to-face sessions, decisions could be made with more ease between them (but not necessarily across the whole team).
The third condition relates to the collectively perceived need for making a decision. It is likely that the probability of being able to decide through electronic communication is higher if there is general discontent with the present situation among central members. Whereas this point is not new in organisation studies (Spector, 1989), the Case Study 2 data point towards a similar explanation. A decision to take turns to moderate a debate was made in the electronic conference after it had been recognised that the existing system did not work satisfactorily.
This chapter has tried to synthesise the data collected in the VMDL case studies. Of course, different foci for the analysis could have been made but the Project has attempted to capitalise upon the disciplinary interests and skills of its members and, at the same time, draw in depth from the data which was collected.
The CSCW field is still quite new and much needs to be done in developing an understanding of the use of communications media in the role of support for distributed research communities. The final chapter of this report attempts again to bring out the key findings of this short, but critically naturalistic, study.