A hypothetical example.
An Airbus A-380 appears on the scene – bigger than any prior commercial jet. The FAA has to certify it. In certifying the plane, the FAA is also certifying that everything the FAA administers is safe in being used with the plane. We have a knowledge transfer topology like A-380 to FAA to airports, airlines (including those affected by the whale, but do not buy it) and air traffic control (ATC).
The aircraft meets FAA requirements for deplaning time in an emergency, but some airports find they cannot move 800 people out of harm’s way quickly enough, so ask for different deplaning plans. Some airports find their east west runway is too short, or their parallel runways are too close for turbulence from the big wings, so need to change their operation, which changes how ATC sees those airports, and ATC has its own issues with separation, which reflect back to the airports and the airlines.
Many foreign aviation administrations use FAA as the "gold standard", but some are unwilling to make costly changes to their airports. ATC uses foreign airports as an overflow resource in an emergency, so now needs to understand new limitations that might apply.
A large slab of new knowledge is brought into the FAA, testing the limits of the existing system, which had evolved for smaller craft. It is disseminated out to airports, airlines, ATC. Changes in practices at any of those organisations has to be synchronised – airports change something, tell ATC. ATC changes something, tell airlines and airports. Some knowledge has to be reflected back to Airbus. Foreign airports, airlines and air traffic control need to be advised of what FAA is doing, FAA needs to be advised of what they are doing (or intend to do, which is much harder to handle, as the knowledge is more abstruse). The knowledge goes around and around. FAA cannot wait to learn from a mistake, so it has to combine some physics, some technology, some operations into the knowledge it builds to handle the new craft. Once the knowledge settles, everybody can go back to, not exactly sleep, but not be too worried about huge transfers of knowledge.
A KMer has to not only know where to store the stuff, but handle the flow of knowledge, integrate new with old, have some idea of the dynamics – how long will the knowledge take to settle (a "learning organisation" is fine, but how quickly can the organisation learn significant new concepts in a dynamic environment, and has the Knowledge Architect directly addressed this problem). If the knowledge will take 9 months to settle, and the plane is in the FAA’s skies in 6, the risk of catastrophic failure is increased. In almost every case, IT will be snowed under and KM has to cover for them as well (even more knowledge is required at startup).
The example has been used because of the gravity of a mistake. It may be based on a misunderstanding, but some other example (the destruction of a major corporation, say) would link KM to the dynamics of knowledge just as strongly.
As a KMer, you may be uncomfortable in getting involved with the system dynamics, but if not you, who?