Christina's LIS Rant
Monday, June 30, 2008
SLA2008: KM at the Core, Facilitating Knowledge Sharing
This session was Monday at 9am and featured Dave Snowden - who is entertaining to listen to, even if he trashes my whole paradigm of research (harumph, if he can't tell anecdotal "evidence" from rigorous systematic qualitative research - that's because he doesn't know better, not because there isn't a difference!)
(podcast is available, and slides - apparently the same as from Limerick)
My notes were on paper - and this is 2 weeks later so...
KM (knowledge management) - is a theory or Weltanschauung with dysfunctional technology
Social Computing (social computing technologies - SCTs) - increasingly functional technology without theory or Weltanschauung
His goal is to build theory to use SCTs for KM (if there *can* be such a thing as KM) - given knowledge is volunteered, not conscripted. Lack of sharing: it's often not a matter of "knowledge is power" but the fear of abuse - so sharing happens with people who are trustworthy. By forcing knowledge sharing you get unusable stuff and you have to go back to the author.
Knowledge is contextual - you remember when triggered by a weak signal. The way people know something in the field is different from the way they discribe something. Standard deal: know more than we can tell, can tell more than we can write down.
3 groups of knowledge:
know by doing - muscle memory, only learn by experience, successfully transferred through apprenticeship (he doesn't like the term tacit, though)
stuff we can tell - communicated by story, pre-rehersal (so going over what you're going to do in a meeting before you do it), activate patterns (?)
stuff we can write down - severely limited because highly structured so "best practice" documents are expensive.
SCTs are working to support stuff we can tell, but there's no theory.
He likes Dervin's sensemaking (don't we all?), and using a complex systems approach - make sense of the world so we can act on it.
cogsci - the two predominant models - information processing and behavioralism are not supported by newer cognitive science research; therefore, the fundamentals of business research are misguided (at best)(!)
In an ordered system
- repeated relationships between cause and effect
- manufacturing > most km
- constraints on behavior - no degrees of freedom
- no innovation - but apply these methods in an unordered system and you'll have failure
Chaotic systems
- unconstrained - use statistics and probability theory
Complex adaptive systems
- system and agents co-evolve
- hindsight does not lead to foresight
- future is inherently uncertain
- very sensitive to starting conditions
- system level effects can be emergent
Flexible, negotiable boundaries
- use attractors to encourage good behavior and disruptors to discourage bad
- weak signal detection (to disrupt early) - good surveillance
initiate a system - with safe-fail experimentation - distributed cognition vice centralized cognition which has low resilience.
the Cynfin framework (no doubt TM)
"complex acts of knowing"
(lots of really illegible notes here - gee my handwriting is terrible)
probe - sense -respond, don't allow existing experts to dictate decisions based on best practices from historical data. "bounded applicability" need to increase dissent...
Narrative picks up more signals than analytical analysis (huh?). Don't confuse innovative with creative (yes, well, Jill and I also made that point)
Internal km systems - based on ordered system, formal proccesses with large chunks
wisdom of the crowds - distributed cognition - everybody must make a decision independent of everyone else - this is *not* a prediction market where you can see what decisions everyone else is making.
Cognitive differences help distributed cognition. Keep partial patterns - everything is fragmented - increase fragmentation - chunking and summarizing is too slow and loses important detail...
km has to be a request system, but you have to tell people what you know so they can find who to ask for information
and then my notes end... ¶ 3:58 PM| |cites (technorati) |
(podcast is available, and slides - apparently the same as from Limerick)
My notes were on paper - and this is 2 weeks later so...
KM (knowledge management) - is a theory or Weltanschauung with dysfunctional technology
Social Computing (social computing technologies - SCTs) - increasingly functional technology without theory or Weltanschauung
His goal is to build theory to use SCTs for KM (if there *can* be such a thing as KM) - given knowledge is volunteered, not conscripted. Lack of sharing: it's often not a matter of "knowledge is power" but the fear of abuse - so sharing happens with people who are trustworthy. By forcing knowledge sharing you get unusable stuff and you have to go back to the author.
Knowledge is contextual - you remember when triggered by a weak signal. The way people know something in the field is different from the way they discribe something. Standard deal: know more than we can tell, can tell more than we can write down.
3 groups of knowledge:
know by doing - muscle memory, only learn by experience, successfully transferred through apprenticeship (he doesn't like the term tacit, though)
stuff we can tell - communicated by story, pre-rehersal (so going over what you're going to do in a meeting before you do it), activate patterns (?)
stuff we can write down - severely limited because highly structured so "best practice" documents are expensive.
SCTs are working to support stuff we can tell, but there's no theory.
He likes Dervin's sensemaking (don't we all?), and using a complex systems approach - make sense of the world so we can act on it.
cogsci - the two predominant models - information processing and behavioralism are not supported by newer cognitive science research; therefore, the fundamentals of business research are misguided (at best)(!)
In an ordered system
- repeated relationships between cause and effect
- manufacturing > most km
- constraints on behavior - no degrees of freedom
- no innovation - but apply these methods in an unordered system and you'll have failure
Chaotic systems
- unconstrained - use statistics and probability theory
Complex adaptive systems
- system and agents co-evolve
- hindsight does not lead to foresight
- future is inherently uncertain
- very sensitive to starting conditions
- system level effects can be emergent
Flexible, negotiable boundaries
- use attractors to encourage good behavior and disruptors to discourage bad
- weak signal detection (to disrupt early) - good surveillance
initiate a system - with safe-fail experimentation - distributed cognition vice centralized cognition which has low resilience.
the Cynfin framework (no doubt TM)
"complex acts of knowing"
(lots of really illegible notes here - gee my handwriting is terrible)
probe - sense -respond, don't allow existing experts to dictate decisions based on best practices from historical data. "bounded applicability" need to increase dissent...
Narrative picks up more signals than analytical analysis (huh?). Don't confuse innovative with creative (yes, well, Jill and I also made that point)
Internal km systems - based on ordered system, formal proccesses with large chunks
wisdom of the crowds - distributed cognition - everybody must make a decision independent of everyone else - this is *not* a prediction market where you can see what decisions everyone else is making.
Cognitive differences help distributed cognition. Keep partial patterns - everything is fragmented - increase fragmentation - chunking and summarizing is too slow and loses important detail...
km has to be a request system, but you have to tell people what you know so they can find who to ask for information
and then my notes end... ¶ 3:58 PM| |cites (technorati) |
Comments:
Post a Comment
