Appendix B. Organisational Memory

Since February, I have participated in the COST CoTech working group CIS-ORGMEM (on Common Information Spaces and Organisational Memory). While this has turned out to be basically outside the scope of my main PhD topic of evaluation - which has led to my scaling down my involvement in the group - I have produced two outputs during the year as part of this group, which are worth mentioning here. Firstly, I prepared a bibliography, currently of over 350 items concerning organisational memory; and secondly, I wrote a short position paper for the first meeting of the group, in Limerick. This is given here for the sake of fullness (and also because the case considered is that in the paper that introduces the PETRA framework); the bibliography is too long and beyond the scope of this report.

Introduction: the realist model

I shall suggest in this brief paper that the prevailing view of organisational memory has been a realist one, and that this is not an adequate capturing of the richness of the nature of group interactions. I shall suggest an alternative view, that organisational memory is an ad-hoc and intersubjective creation, and illustrate this by considering an example of this kind of development in a technological context. To begin, let us consider a typical description of organisational memory from the realist perspective:
By organisational memory I mean the record of an organization that is embodied in a set of documents and artifacts. Note that collective memory (i.e. the pooled memory of individuals) is excluded from this definition. (Conklin, 1993, p.561)
There are two important aspects to this view: firstly, it is based on a strictly objective ontology: organisational memory consists entirely of the organisation's various documents and artefacts; secondly, organisations are viewed from a distinctly biological/cybernetic metaphor (cf. the title of Stafford Beer's (1981) book, "Brain of the Firm").

On the first point, of course some organisational memory will be held in documents (and also, as Conklin points out, in the processes that lead to the creation of those documents - the design rationale if you will). It is also true that this is the easiest part of the organisational memory to model. However, to suggest that the only part of organisational memory are these entities seems to miss a lot, as I discuss later.

Ackerman (1994) also takes a rather limited view of organisational memory, suggesting that it is the knowledge about local experts on particular topics (in his case, concerned with X-Windows). That makes some sense where the knowledge is objective ('how do I write a Motif widget?') and can even be encapsulated in a set of Usenet-style 'frequently asked questions'; it makes less sense where the knowledge is more subjective or discursive ('what use is ethnomethodology in studying organisations?'). In the latter case, different answers will be obtained from different people - there will be no one local expert.

It is also worth noting briefly the biological metaphor that seems to be behind the very use of the word "memory". Companies simply don't have brains and to imply they do is to risk being caught up in general systems theory, and as Checkland (1981) has discussed, this is a view that fails to capture many of the most interesting parts of the nature of an organisational system. While this can be put down to over-flowery language - Checkland (1990) makes the same point about the use of the word "memory" to denote computer storage - it is worth remembering (sic) that drawing any greater meaning than metaphorical from this is likely to be dangerous.

An alternative perspective

The above critique of existing views of organisational memory is not meant to suggest that existing work is bad: merely that it is not enough. I wish to add to this rather objectivist view of memory a similarly rather subjectivist view. I would suggest that to ask the question "what does an organisation remember" is exactly the same as to ask "what does an organisation know" - which in turn is the same as asking "how can any group of individuals be said to have any knowledge in common?"

This is the question that since Hegel has been the fundamental problem of sociology - the problem of intersubjectivity. This is described by Heritage (1984:54) thus: "how can two or more actors share common experiences of the natural and social world and, relatedly, how can they communicate about them?". This has been a much-discussed question - the answer ethnomethodology gives is that it is socially-constructed on an ad-hoc basis, constantly changing from time to time (and not, for example, the Platonic pre-existing group-view-on-an-issue that it is taken to be by traditional sociology, and indeed it seems by some organisational memory theorists).

If we take the ethnomethodology view of intersubjectivity as our view of organisational memory, we end up seeing that such memory is socially constructed on an ad-hoc basis. Below, I outline how this might occur for what we might call (stretching the metaphor to it limits) short-term organisational memory. It should be stressed here that the same processes apply to 'long-term' organisational memory: what an organisation remembers next year is the amalgam of all the things its members know this year, some of it as reified into documents and the like, some of it purely existing in the collective minds of the members of the organisation.

To clarify this discussion somewhat, consider a university department. The membership of that department is a fairly fluid thing: staff come and go, research students have a limited time in the department, shorter times still are spent there by Masters and undergraduate students doing dissertation work. Each of these people has a set of knowledge, beliefs and understandings from their reading, from talking to and learning from others outside the department, from interactions within the department. Over a given period (say a year) the various individual understandings of the department's members will change, and its collective understanding will thus change, as the amalgam of the individuals' understandings and as the combination of particular individuals' understandings in discussion (informally or in more formal settings designed to develop the shared understanding, such as departmental seminars).

From this, we can see the fluidity and ad-hoc nature of the department's organisational knowledge from time to time; and hence of its organisational memory as that knowledge becomes past knowledge. The more objective side of organisational memory is also to be seen here: the members of the department produce books, papers for conferences and journals, internal reports, theses and dissertations, and so on. Some of the memory of the department is indeed held in these documents, just as some it is held in the statement that "Bill's the expert on X here" - except that Bill may have forgotten about the relevant bit of X, or Bill's student may be more expert in X than he is, or Jill who joined the department last week in fact may have written the seminal book on X.

Technological mediation of intersubjectivity

To show what I mean by the way a shared understanding develops in a group, and to move towards considering where technology comes into things, I will discuss briefly below a semi-experimental study (qualitative but semi-situated, according to the taxonomy of Plowman et al., 1995) of a group using a collaborative writing tool, ShrEdit. This study was conducted as part of an evaluation of that tool, and is discussed further in Ross et al. (in press).

The group consisted of three MSc students at Sussex University. We assigned them a task, to prepare a revision answer to an exam question for a forthcoming exam all three were to take (this was a 'real' task in that it was something they would have done anyway). Two separate questions were answered under different conditions: firstly with them answering the question in a face-to-face setting, secondly in three separate locations, linked by ShrEdit and a telephone conference call.

One of the aims of this study was to consider the question as to how the development of shared understanding is affected by whether the method of co-ordination is a social one or a technological one (Rogers and Ellis, 1994). There is an fundamental difference between the first part of the study, where representational media are essentially social (the participants were all sitting around a table talking and writing, with the only shared artefacts being drawings and writing on paper), and the second, where representational media were essentially technological (the only connection between the participants was a telephone and the use of ShrEdit).

The following text from Ross et al (in press) describes the situation and the way the repair of a breakdown in shared understanding occurs:

We considered as part of this the interrelatedness - what Plowman (1994) has called the interfunctionality - of talk and writing in the mediation of shared understanding: do they work together? This raises a further question: does writing on its own (via ShrEdit) mediate shared understanding in the way that talk alone does, or is it merely an adjunct to the understanding? We suggest that co-writing via ShrEdit acts as a technological mediator of shared understanding, and furthermore, that it acts as the primary mediator of shared understanding. Our evidence for the existence of such technologically-mediated shared understanding is not any explicit reference to it in the transcript, but rather the existence of breakdowns in understanding that the participants notice in their ShrEdit based co-writing. We describe these as repairs to 'shared mis-understandings' that progressively happen in the writing process. The following example shows such a mis-understanding repair:

C: We've got a problem with um if you look the line just above um point 1and what you've got (.) is halfway along, threshold step function er and that's wrong
N: Ri::ght. Tum tum tum tum - something
C: It should be "relies on a easily differentiable"
S: Yeah, I mean [ it's not easily taken
C: [ threshold function
C: Yeah - a step function isn't and it can't be used =
S: = Right, sorry
C: And I explain that later on in my part

Hence, technologically-mediated shared understanding exists in a tacit form in ShrEdit, and is primarily manifested in breakdown/repair situations. That is, one of the roles of social coordination is as a repair mechanism for the failures engendered by technological mediation. Thus shared understanding is partially embodied within the document. As one participant writes a section, they establish their individual understanding of it; as others read that, they add to the shared understanding by constructing their individual understandings. Breakdown occurs when one participant's understanding differs with what another has written, necessitating group discussion and the construction of a more meaningful shared understanding.


Appendix A / Contents / References

Go to the Evaluation of Cooperative Systems home page


Cooperative Systems Engineering Group | Computing Department | Lancaster University
Magnus Ramage 10 October 1995