Interesting analysis of success and failure of the dominant military dogma of waging war in Iraq. There is lack of explanatory power on the mechanisms of change however, and how there has been a change from a failing to a successful strategy. Indeed whether there is success in sight and what would that entail (i.e. whether a strategy exists that goes beyond an exit plan). It is not clear whether the US army has become less hierarchical, more reflective, has better feedback processes etc. The assertions made are typically ex-post facto identified.
A good point is raised on the importance of having a large number of variation in experimenting with different ideas and maintaining an unbiased selection of the most promising between them. Such gambles should challenge the survivability of an organisation/system/individual by being small. He calls this the Palchinsky principle (p.24-26).
He considers the finance & banking sector to need decoupling and solid barriers between sectors following the advice of John Kay for narrow banking licenses.
He also points to the literature of Klinger/Hausmann/Hidalgo/Barabasi on ‘product space’ through the comparison of trade data and the possibility of a different way of looking at comparative advantage based on simplicity or complexity of export base of a country. I have to look into this literature in more detail as it directly associates to my own interest in networks. Some interesting references to experiments as well.
He suggests that financial and managerial risk management should take a leaf out of industrial risk management. He makes the case with some rigour.
Weaknesses: His view of innovation and intellectual property. The difference between invention and innovation is confounded and generally ignored. The role of adoption in commercial or innovative success is not taken into account. The fact that the bulk of invention (and innovation) progresses in small incremental steps is ignored. The role of motivation and agency is only given cursory attention.
The most important limitation is that failure is not proven to be a condition of success (as claimed). Is the scope of failure just about learning? I will be reviewing the book by Paul Ormerod on this topic next to offer a relative comparison. Trial and error is not tantamount with the processes of acknowledgment and success. Indeed, success can favour agents randomly or most often through processes that have nothing to do with merit but through the adoption by (or coercion of) other agents. In that sense success could reflect variation in communication skills, or the acts of other agents beyond the horizon of observability of the focal agent and so on.
Recommended as a review of popular literature on processes of failure.