Based on a recommendation from Gunnar, I read The Starfish and the Spider by Ori Brafman and Rod Beckstrom. I spent all but the last couple chapters wishing that I were not reading it, but in the end it was worth it.

For most of the book, the core message is, “Look! Decentralized organizations that work!” Spiders die, starfish regenerate. Based on the reaction of the few intrigued people I talked to about the book, I’m sure this is a revelation to many, but since I’ve been interested in decentralized organizations since, oh, forever, this observation alone isn’t all that compelling. Certainly not enough to build an entire book around. They provide decent examples — the Apaches, Alcoholics Anonymous, Wikipedia (of course), Burning Man, P2P filesharing — but not a terribly nuanced examination of why decentralization works, or in what scenarios it can be successfully applied, or where it doesn’t work well.

Or so I thought. As I explained the book to my mother (one of the aforementioned intrigued people) I realized that they had provided an interesting analysis of factors that help decentralized organizations succeed in the face of increasingly centralized opposition. When facing a decentralized threat, whether it’s file sharing, terrorist cells, or botnets, one would do well to pay attention to the failures of centralized models. Becoming more centralized tends not to work.

I was surprised to find no mention of Dee Hock, Visa, and chaordic organizations, but that might stretch beyond the narrow confines of the authors’ intent.

In the final chapters, Brafman and Beckstrom at least begin to explore what I had hoped would be the meat of the book: merging decentralized organizational models with centralized ones. Or rather, using decentralized structures within a centralized organization. As with the rest of the book, there’s a rapid-fire series of examples, and a longer exploration of how this plays out in one company (GM). These are just a couple chapters in a short, easily read book, so I’m still a little disappointed by the depth of the analysis. But if you’ve got yourself a bus ride, you could do worse than to spend a little of that time in the last third of this book. If you are completely puzzed by the very idea of decentralized organizations, then you should definitely read it.

Now I’m working my way through the rest of Gunnar’s recommendations.