Knowledge in complex domains

colliver's picture

I came across Dave Snowden's work at the tail end of my close involvement in knowledge management. He was amongst that pack of people talking about knowledge as a social process, but he made alot more sense that most. Snowden argues that the way knowledge is developed and used depends on the complexity in the system. Operating in complex systems is different to operating in simple systems – it requires collegial structures, not professional or bureaucratic structures.Snowden articulates critical differences between systems in four different states, along with implications for working with knowledge each state:[i]

Simple domain Few cause-effect relationships, simply connected, so that cause and effect connections are predictable and repeatable. Problems that are solved tend to stay solved, at least in the medium-term. Bureaucratic structures work as a way to organise action, because best practice can be built into procedures. The learning sequence is sense-categorise-respond.
Complicated domain Multiple causes and effects, where cause-effect relationships can’t be immediately recognised, but can be understood through research. Professional structures handle this domain best, using communities of practice to evolve and confirm best practice. The preferred learning sequence is sense-analyse-respond.
Complex domain Multiple causes and effects with complex feedback loops. Causes are only perceivable after the end, and work as explanations rather than predictions. Problems keep returning despite many attempts to resolve them, and solutions imported from elsewhere don’t work. A lot can be learned from mistakes. Collegial structures work best, and the learning sequence is probe-sense-respond.
Chaotic domain Multiple and turbulent cause-effect connections, so turbulent that there is no point in thinking about cause and effect. There is no point in trying to develop solutions: it’s best to act quickly to create stability. Any structure is of little help: just act-sense-respond.

Snowden’s prescriptions for working in complex domains resonate with Boisot’s views on knowledge management. Boisot also observed that knowledge behaves differently in different situations. When knowledge is codified and abstract, it can be managed in hierarchical structures, or through markets. But when it is diffused across a system, and highly concrete and un-codified, as it is in complex situations, people have to use their personal experience to make sense. In such situations, a clan structure is the most effective social structure for working with knowledge. Relationships in a clan are personal but non-hierarchical, underpinned by common values and beliefs, and shared goals develop as people work together.[ii] This fits readily with Wenger's work on communities of practice. which takes an understanding of the social processes that underpin such relationships much further.


[i] Snowden, D., 2002, Complex acts of knowing, Journal of Knowledge Management, special issue. This summary also draws on Bob Dick, “Complexity processes,” 2006 ALARPM Annual Conference, Brisbane.[ii] Boisot, M., 1999, Knowledge Assets, Oxford: Oxford University Press, pp 124-136.
Dianne.ALLEN's picture

Another connection .. another thank you

Two references to David Snowden in the same week, or near enough!

I must be re-immersing myself in the field and finding the edge again!

This time Ross, your summary of David's material is so helpful.  It rings with Ian Hughes' reference at ARLIST, where the discussion also includes somethign about 'generalisability'.   Ian passed on this connection and it has gone into my favorites (I'll have to dip into del.icio.us soon Marg!)

Dianne Allen KIAMA, NSW

Dianne.ALLEN's picture

Serendipity in complexity

Hi Ross,

Since upgrading the site I have had a number of items pop up again in my unread box!

Today, the pop-up is this post.  And the serendipity of this complex process (upgrading a computer-web-based portal, and keeping track of the past) now allows me to remark about knwledge domains, again.

When you first posted this entry, not only did I have the immediate connection with Ian Hughes' contributions elsewhere, I was taken back in memory, but inadequately, to another analysis of knowledge domains.   I needed then to chase the reference.  That took a day or so, and by then the connective tissue had dissolved.

The reference was to the knowledge domain analysis of Gibbons et al 1994

The context of the reference was a workshop on the nexus between teaching and research in the university environment.

My sense, at the time of making the connection with your post Ross, was that Snowden's analysis appeared to be more helpful in the practice situation, than was the Gibbons et al analysis.

Dianne Allen KIAMA, NSW

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.