Saturday, July 08, 2006

Tacit knowledge and tacitness as a dimension

One of the big challenges in knowledge management and training software agents to learn is the concept of tacit knowledge, which encompasses that which we "know" and which can influence our thinking, and behavior, but which is difficult or even impossible to conciously communicate to others. Rather than conceptualize it as a binary state, is or is not tacit, it may make a lot more sense to look at it as a dimension, a spectrum of difficulty of conceptualization and communication.

There are implicitly two aspects of tacit knowledge: our ability to conciously "think" about and contemplate knowledge, and our ability to convey or communicate that knowledge to others, including computational entities.

Here are some of the degrees or levels of tacitness that immediately occur to me:

  • What we in principle or theory could never "know" or communicate, possibly due to the "computational" limits of our brains and minds.
  • What is extremely difficult to "get our minds around" or articulate and then only with great effort or superior insight, but nonetheless can in theory be conceptualized and communicated.
  • What is relatively difficult to conceptualize or communicate.
  • What is relatively or moderately easy to "know" but much more difficult to communicate.
  • What we believe or are sure that we know, but have great difficulty communicating.
  • What we know, but can commincate only with individuals who "have been there" and already share a substantial amount of common knowldge or shared experience.
  • What can be communicated easily only between individuals with a common culture.
  • What can be communicated easily only within specific communities.
  • What requires a shared expertise.
  • What requires a shared world view.
  • What can be conceptualized and communicated within a genetic species.
  • What can be conceptualized and communicated with relative ease.
  • What other entities may already know and a few clues or cues are all that are needed to "convey" understanding.
  • What other entities already know and we simply need to reference.
  • What others already knew before we even told them, what we didn't need to tell them and should have known that fact
  • What we believe that others already know and believe need not be communicated.

To communicate with people, software agents (computational agents) will need to have capabilities for coping with these and other aspects of tacit knowledge. In fact, intelligent software agents will need these capabilities even to interact with other computational entities.

The eternal question is and will be: What do you know and how can I know that I know what you know? A deep understanding of tacit knowledge is essential to being able to answer that question.

-- Jack Krupansky