Models are are useful until they are not

Pirámide de Kefren - Khafre's pyramidAlternately: All models are wrong, some are useful. George E.P. Box

I know I have written about the data-information-knowledge hierarchy in the past.  Many people still use it as a metaphor, but now it seems that it has outlived its usefulness.  I came across it again in a new-to-me blog from Phil Green at Inmagic, Social media is challenging notions of the DIKW hierarchy.  One of the biggest concerns in the hierarchical view of DIKW is that it reduces these human concepts to "objects" upon which computers can try to act.  And it reduces human behavior down to computation, which isn't the case.  This is also why it has been popular with the KM-as-technology lines of discussion.

The DIKW model is an easy way to describe the relationship between these concepts, but it breaks down when you put more critical thought into the question.  The whole idea that one can be drawn from the other becomes difficult when you consider the connections of experience, domains, language, culture, social networks and the rest of human behavior.  I left a comment on his blog that says as much too. 

Dave Snowden - who has been rather vociferous in his condemnation of the DIKW model - wrote recently that It's information to date we need, not DIKW.  And I like the simplified version that Eli Goldratt uses in The Haystack Syndrome: Information is the "answer to the question asked."  If you have no question, there is no context or meaning, thus no information.  And this handily avoids the question of knowledge, let alone wisdom.

As far as the proposed topic of Phil's post itself, there is one paragraph where he suggests the DIKW hierarchy metaphor is challenged by social media - and that he will be writing more in the future:

The DIKW model is a uniquely relevant topic as social technologies take hold and challenge not only the relationships between data, information, and knowledge within enterprise organizations, but also how information and knowledge is captured and transferred amongst your staff.

This is something that could lead down an interesting path.  Maybe it will help develop a new metaphor for "what is knowledge" that doesn't give us such a simplistic, computational view of the world.

p.s. I note that the title of this post was also used by Steve Major in May, but I swear I've heard versions of this before.

[Photo: "Pirámide de Kefren - Khafre's pyramid" by Xavier Fargas]

2 Comment(s)

Mike Sivertsen said:

All good points Jack and areas I've considered and written about in depth while completing a Masters degree in knowledge management in 2009. My Capstone paper and a subsequent presentation at a national conference in April 2010 (see below) explored David Snowden's Cynefin framework and discussed the utter failings of models in two important areas: the U.S. financial system and 'man-made global warming.' In both of these areas the complete failure of the predictive capability of the models should convince decision-makers to abandon models when working in a complex adaptive system (CAS) where the only model is the system itself (as Snowden has said). Unfortunately, models continue to look 'sophisticated' and 'smart-looking' and can easily become a surrogate for real thought or an ideological agenda. Acknowledging the inherent messiness of economic, ecological, or human states is uncomfortable as it means sharing decision-making power. Misplaced ordered system approaches in which models are used in a vain attempt to understand or predict a CAS results in ecological damage and human suffering. Conversely, understanding the dangers of entrained thinking and embracing the value of a CAS with weak constraints can improve systems - as indicated by events surrounding the Longitude Prize, Netflix Prize and the traders vs. Marines competition - all covered in my presentation.

Abstract and bio from the Lean Software and Systems Conference, April 2010

Scroll down to: "Cognitive Kanban: Improving Decisions in a Complex World," at this page for a richly annotated PDF in a "Beyond Bullet Points" format and an accompanying MP3 podcast.

Not that I was worried about modeling at this level, I wonder about the idea of small models of very specific elements of the larger system - rather than a massive model - could be more effective in these complex situations. I'm thinking of the basic model of flocking behavior: if you model one bird with some simple rules, the appearance of the V-shaped flying formation emerges from the models interacting together.

Leave a comment


Previous entry: Explicit work (#owork)

Next entry: Theory of Constraints as part of an integrated improvement process

Picture a steaming coffee cup. Better yet, grab one and have a read!

KJolt Memberships

Follow jackvinson on Twitter

View Jack Vinson's profile on LinkedIn