Overview: Subtitle says it all.
Also contains a simple guide to the most common types of mathematical curves relating to data. The 10 areas with the rules of thumb he advises are gaps (look for the majority), negativity (expect bad news), straight lines (they might bend), fear (calculate risk), size (get things in proportion), generalisation (question categories), density (slow change is still change), single perspective (get a tool box), blame (resit pointing a finger) and urgency (take small steps). This a a bald summary he makes a compelling case.
Relevance: It is necessary to know realistically how things are if one wants to make improvements, it may be better to reinforce success than go for something new and different. When tying something new do it on a small scale first. So thinking about this, for example we know coops work there are thousands of them.