One can find insight in unexpected places. Team of Teams by General Stanley McChrystal et al is a book about dealing with organizational complexity, written by a general, which uses the war against Al Queda as a common thread. I was somewhat skeptical that I could find information here about software teams that I'd find immediately useful, regardless of how interesting it was. I was wrong. In addition to the advice I was expecting to read about team dynamics, I gained some insights that I thought would be immediately applicable to scaling Scrum Teams.
Fairly early into Team of Teams the authors explain that this is not a war story. While it's true that the common thread in the book is how General McChrystal worked to get the army more able to adapt to a decentralized, agile enemy, there are are also stories from commercial aviation, NASA, and corporate America. While there is a fair amount of military history in this book, there is also a discussion of the history of manufacturing process improvement, office space, and even personal stories about gardening.
The message in this book is that command and control structures don't work in complicated, information rich environments that deal with complex problems. He defines complicated as "having many parts" and complex as having many interactions. To get things done in this kind of environment you many need to cast aside what you are used to thinking of an efficient approach.
McChrystal draws a contrast between being resilient and being robust: The more you optimize a complexy system for a specific goal, the less resilient it becomes. Likewise efficiency isn't always better than adaptabilty: you can build a system that is good at doing things right, but is too inflexible to do the right thing. Redundancy and overlap can enable adaptability; an efficient system that does the wrong thing doesn't add value.
Had McChrystal been delivering this message in the context of a corporate enviroment, you might dismiss this as another instance of someone who is in an environment suitable for agile singing the praises of agility. That he's talking about battle command situations, the canonical top-down environment, gets your attention. As with Turn the Ship Around this provides evidence that an "agile" approach can work in more circumstances than you might guess.
At the core of the book is the philosophy that too much control of information and decision making slows down a group's ability to react to situations. The situations described in the book were ones where the risks of acting too slowly are higher than the risks of competent peope making judgment calls. Automomous decisions are usually good as long as they are visible and shared. While most business decisions don't have the same life or death implications of those made in a war zone, slow decision making can have a business cost, and potentially a morale cost if people feel undervalued and micromanaged.
McChrystal advises us that it's not enough to "empower" people in name only. Automony and delegation only work with a clear understanding of a common purpose. Communication channels and shared information are important to both sharing the common purpose and giving people the tools to act in a manner consistent with that purpose. One technique he used was to provide for open office spaces. McChrystal acknowledges that open office space only improves team productivity when everyone is working on related problems, and that open office space as a way to be "space efficient" is productivity inefficient. Video conferencing for his troops was also important, as his teams needed to communicate with others in remote places. As someone who works in a company with people in multiple locations, the stories of the challenges of setting up video conferencing sounded all too familiar. That they were successful in a low bandwidth war zone situation was encouraging.
The book also had some insights about scaling teams that had an implication for scaling Scrum. While it's appealing to think of projects at scale as hierarchical, each level of hierarchy hides information. That can be fine when everyone knows what other teams need to know and not know. But that kind of knowledge is rare. McChrystal's approach to scaling is that everyone needs to know someone on every team so that information will flow organically.
I found both inspiration for seeking ways to improve, and practical advice about how to structure teams and be an effective manager in this book. While not about agile per-se, I think that there are lessons here that apply to those trying encourage agile adoption in a non-agile situation, and also for those looking to scale agile.