To begin with, both concepts suffer from being over-reified, considered through a rather blinkered objectivist epistemology, and formalised in too simplistic a manner. In short, they're rather bigger than they're given credit for.
Organisational memory consists of two main types, of which only the first is sufficiently recognised by system designers. The first concerns those artefacts of cooperation that can be pointed to as explicit, 'hard' indicators of what has been happening. These include products on the one hand (documents as much as manufactured goods); on the other, records of collaboration and ideas, particularly minutes of meetings and items such as frequently answered questions (FAQ) lists that record common knowledge on a particular topic. These kinds of objects certainly represent the stored knowledge (the memory) of an organisation, or group, or project; and they are fairly easy to store, maintain and index (although see below for discussion of problems with those tasks). Thus some researchers and software developers equate them with the organisation's memory.
Yet they are not all that there is. There are many items of knowledge that are present in an organisation qua entity (i.e. belonging to one or more members in their roles as members rather than as ordinary individuals) that cannot be stored thus. Knowledge of where a particular person is, of who is an expert on a particular topic, of the politics of the group and its outside relations - these are all knowledge around in an organisation that is necessary for its smooth functioning. If such knowledge is present in the group collectively, or even in its key individuals (not necessarily in terms of hierarchies - the 'Mavis phenomenon' identified by our own sociologists, of the key individual who knows everything, usually a secretary or administrator, being an example) then it could be said to be organisational knowledge. And if it remains known, it passes into the organisation's memory.
"If it remains known" is a phrase that begs many questions. Not the least is how: by what mechanisms does knowledge stay in the organisation? Some will stay in the minds of its members (although if they leave, what then happens to it?). Other parts of the knowledge will become fixed, though in very different ways than the documents mentioned above: it will pass into the organisation's culture (for example myths like "we've always done it this way") and even into its formal processes (an example being the way safety processes get tighter after major disasters).
What regulates this passing on of knowledge from the organisation's short-term to long-term memory is worth considering. Three main factors will affect what becomes long-term: relevance, time and power. Some things will not be relevant to the organisation so will simply disappear (such as the football results). Others will only be important for a certain time (weather conditions, where people are) although some part of them may be retained in memory - "how we got to work during the last blizzard" or "don't you think Jim is out of the office rather a lot?". Perhaps the key factor is power, though: something may well only become an 'official memory' (e.g. instantiated in a formal process) if those in authority approve it; though folk memory of "the boss's injustices" (say) may also be powerful as a factor in the organisational culture.
Most of this paper so far has talked of organisational memory. However, it is our contention that what we describe above as the second kind of organisational memory is essentially cooperative awareness. Who is around; what they are doing; what they have done recently - these are the very things users of CSCW systems need to be aware of. In this sense, organisational memory could be thought of as the history of past awareness information.
Another way of considering links between cooperative awareness and organisational memory is to look at the problems faced in trying to make them into computer systems. We hold these to be four-fold, concerning capture, retrieval, interpretation and purpose.
Firstly, there is the question of capture. Here we see issues of what to store in the system - at what level of detail, from whose viewpoint, what kinds of information to store and so on. But there is also the issue of how the capture occurs - whether it has to be explicitly entered or whether it can be stored directly by the system.
On the one hand, if it is explicitly entered the questions of truthfulness, of completeness and of accountability present themselves. As we have mentioned before, what is recorded and what is not depends greatly on power and on interest; accounts are neither guaranteed to be objective nor, indeed, truthful. What is perceived as relevant and what is not, is also dependent on a wide variety of factors and explicit accounts are therefore unlikely to be complete. Finally, much of our knowledge about factors that are relevant to an organisational memory system is tacit and is therefore not readily accountable. Accounts of why a certain decision was reached, of what the organisational culture of our company is really like, etc., are more often than not rationalisations rather than truthful and objective accounts.
On the other hand, automatic information gathering through a computational system, while certainly objective at a trivial level, is limited by the fact that most interactions take place outside or at least around the system and not within it. Furthermore automatic information gathering is likely to result in an unmanageably large amount of data being recorded.
The next problem is to do with retrieval. A big risk of any system that shows what other people are doing and knowing (or have done and have known) is information overload. If you decide to capture a fine-grained level of information (every room an Active Badge wearer moves through in a day, say) then there will rapidly be an unmanageably large amount of information. Thus it needs to be filtered somehow. How should this occur? On an ad-hoc basis, like a database query? On a fixed basis, pre-determined by the programmers or system administrators? On a basis set up by the users in some sort of preference box? All these have benefits and flaws. The basic question here is: who decides what is useful?
To an extent these are technical questions. The third problem is a more fundamental one, though: having stored and retrieved the information, how it is interpreted. Both knowledge and awareness are subjective matters, and necessarily incomplete: one individual cannot know what everyone else is up to (unless there are only a small number of you, like Heath & Luff's London Underground controllers; or a larger group mediated by a single common artefact, like the air traffic controllers' flight strips).
There are also issues in the question of interpretation to do with context. The interpretation of anything depends on the context within which it is interpreted (that is, interpretation is a highly situated and subjective process) - just as one person will interpret something differently from another, so the same person in two different settings will interpret something differently. How do we recall enough of the context to the person reading the information months later (who may be the same as the original source or a different person)? What context needs to be stored - do we include the weather? How they were feeling that morning? What they'd had for breakfast, read that week or the argument they'd had last night with their partner? And if this is difficult for the same person, how much more so for another person trying to step into the shoes of the original holder of the awareness or the memory.
Of course this is the same problem as the interpretation of writing by people other than the original author, the study of which is known as hermeneutics. The basic answer from that discipline is that it can't be done: interpretation is always situated and subjective. And yet - what we also know (from ethnomethodology among other areas) is that human beings are quite astonishingly good at remembering things in quite intricate detail from quite small cues. Thus we suggest there is hope here if one views the purpose of the context to be found in the memory/awareness system as being a kind of memory-jogger rather than a definitive answer to everything.
Finally, the question needs to be considered: cui bono? Who benefits? For whom is the system working, who will use the information, who will act upon it? These are not obvious issues, to be tucked away under the blanket response of "the user": for there are many uses of such systems, from the relatively innocuous retrieval of one's own actions a few days ago to the Big Brother scenarios of managers timing how long you're in the loo or lawyers waiting to pounce on that one bad decision in the project.
The problems are many, the situations are similar, and both suffer from under-consideration: these are the common links between organisational memory and cooperative awareness.