Knowledge Integration

Thanks to our scientific and technical achievements we've created a modern world of rapidly increasing complexity and vastly expanding new knowledge. We've created this over-abundance, and attempt to manage it, through specialization: the division of a larger and more complex whole into smaller and smaller specialized domains. While this tactic has been remarkably successful, and indeed necessary, there is a downside to all this specialization: solving complex problems requires that we have some grasp of the whole – to make good use of our specialized knowledge we need the broader perspective that our isolated expertise does not provide.

The ironic and historically unprecedented result of our specialist-driven success is that our ability to solve many problems no longer hinges so much on discovering new knowledge, but on making use of the knowledge we already have. Our new challenge is to start weaving knowledge back together again - weaving it into forms not only more organized, accurate and accessible, but also into forms that are more useful for solving the difficult problems we face.

Knowledge integration is the process of fitting our ideas – our theories of how-the-world-works – together into a coherent structure. That coherent structure, and the process of bringing knowledge together, has a number of critically important, yet under-appreciated, uses:

• As we expand the scope of our thinking we may come across just the idea, or combination of ideas, that enables progress on the seemingly intractable problems we face.
• 
As we reconcile conflicting ideas we can force into the open hidden assumptions and logical inconsistencies.
• 
As we synthesize diverse perspectives we can clarify our thinking and highlight areas of (in)coherence, (dis)agreement, or (un)certainty.
• As we connect ideas we can create a whole that is greater than the sum of its parts:

The goal of knowledge integration is to weave diverse ideas together into coherent networks; networks that are more robust and functional thanks to the cross-linking of ideas and that, in their synergistic combination, might enable an emergent level of intelligence in the face of complexity.

The success of the specialist reductionist approach depends on the system-of-interest (S) being both decomposable – separable into independent sub-systems that can be dealt with in isolation – and closed – insulated from outside influences. Real systems tend not to conform to this ideal and are usually much more complex – much more open and non-decomposable – than we like to assume. To deal successfully with these complex systems we must grapple with these interdependencies and outside influences.

Working with complex systems requires different methods and tools. In particular, it necessitates broad synoptic models to complement our specialized reductionist models. Specialized knowledge is necessary but not sufficient to solve complex problems. Specialized models, while perhaps successful in isolation, often fail in the complexity of the real world; not so much because they're wrong, but because they're incomplete – because they're disconnected from the other knowledge with which they must be consonant and from the wider systems with which they must co-operate.

We believe that we're handicapped in our ability to deal effectively with complex problems by an oversupply of specialized information and a deficit of broad unifying perspectives. Over time our knowledge gets ever better but also ever more fragmented: scattered among written texts, between different people, and across departmental, disciplinary, or cultural boundaries. To counteract this fragmentation, we need a new process –  a whole new discipline – that focuses on weaving knowledge back together into unified and useful perspectives.

Knowledge Integration & Visual Models

The cognitive affordances of visual models and their ability to explicitly show, in a single unified view, the relationships between a large number of diverse elements, makes them an indispensable part the knowledge integration process. Visual models have the unique ability to make explicit the integrated perspectives we require.

© copyright 2002-2004 Marshall Clemens – all rights reserved