Nassim Taleb's Black Swan is also a great read on the instability (or in his words, fragility) of knowledge. A single observation is quite often enough to break what has been considered solid knowledge for years.

Eh, you also have to discern between individual bits of knowledge and systems knowledge when considering systems stability.

You can have complete knowledge of a component, but that may not be helpful when attempting to determine the affect of that component in a system where you have incomplete knowledge of other components.

Nonlinear systems that trigger at thresholds are a good example of this.

That's fair. But is it helpful to anyone to think about knowledge shifts on timescales that are longer than two lifetimes?