Why centralised knowledge bases do not work

I am aware that the metaphor of knowledge animals is not a theory, it is just a 'wild idea'. Nevertheless, it helps me and my colleagues (among others Lilia )to come up with explanations for several successes and -so to say- less fortunate KM-interventions. The following explanations resulted from a short discussion on why centralised knowledge bases do not work (i.e. have limited value). Please feel free to add to this list of explantions or to add your examples of (un)successful KM-interventions.

Knowledge bases do not take into account that:
- people consult people even for documents as long as this is faster, easier (k-animals are lazy and want maximum output with minimal effort)
- people consult people even for documents, as quality and relvance etsimates are useful, and for this you need knowledge, thus people
- people mostly like snacking, and only sometimes nutritious meals like reports
- people need pure 'information scent' not made up traces
- people leave valuable traces anyway, but in their own territories, not somewhere else
- people people are proud and cherish their own spaces -territories- and the cost of shared spaces
- re-using knowledge of others (i.e., other territories) is hard, it is like battles in the frontier areas of territories; k-bases are not sufficient to support knowledge re-use
- ......