Too modest to puff his own horn, Thomas Stewart takes on (the
failures of) knowledge management.
Tue, 15 Jan 2002 16:37:58 -0600
"Meltsner, Kenneth" wrote:
> Pretty much hits the nail on the head: explains why case-based help desk software isn't sufficient; why poorly organized email lists beat well-organized archives, etc.
Right on. Jibes well with some of the themes that have been keeping me up at night for a few years now: ontologies are brittle and categorization ephemeral; too much structure is a bad thing; raw search beats structured queries; views, not containers...
Here's the problem with knowledge management: most organizational and individual knowledge is unstructured, poorly organized, difficult or impossible to search *today*, and balkanized with respect to access. Today's KM solutions try to impose additional structure and relatively rigid organizational schemes on this information, which requires changes to the way people create that knowledge in the first place ---
and that means that it doesn't happen as an ongoing process. Further, when people go looking for stored knowledge, it tends to be in a very ad-hoc, associative fashion --- and it's often easier to simply try to find an expert and ask them, even though the information may be "out there" somewhere.
We've got the technology today to make this problem tractable.
* Artificial barriers to publication on the Web should be lowered / eliminated
* The distinction between filesystem and Web should blur
* We should admit the difference between structure and semantic value
* We should stop trying to build KM tools that require / emphasize structure
* We should recognize that often the metadata is implicit in the data
* We should build better tools for using unstructured data
* Associative access is key: similarity drives dynamic ontology
* Summarization is important
* Ontology is purely subjective
* We must embrace the inherently decentralized nature of stored knowledge
* Search trumps navigation
Just some thoughts. I think Tom's right out front in recognizing what will be a huge, growing need for true "knowledge management" rather than the poor and limited "KM" offerings of the past. The economics and trend dynamics of storage alone make this inevitable: despite the fact that storage capacity / $ is doubling every 12-18 months, *utilization as a percentage of total capacity remains essentially flat.*
This is a rather amazing fact: it doesn't really matter *how big* that drive you're about to buy is, odds are that in 12 months you will be at 45% utilization, at 24 months 65%, etc. Net effect: information accumulates at a rate proportional to the growth of storage capacity. Managing and using this infoglut, er, massive knowledge resource is both a critically necessary and (IMHO ;-) lucrative opportunity
space over the next several years.