I really like this idea - perhaps a proper balance to the feature creep that so often comes during just a few hours of brainstorming during a hackathon.
One way I could see it proceeding is making a mindmap of an organization - its functions, mechanisms, and aspirations, which are then tied to events and decisions which put this organizational 'image' in place. It could be a really great way of evaluation and gaining perspective.
Also this idea reminds me of a book I read some time ago and loved, Vehicles: Experiments in Synthetic Psychology by Valentino Braitenberg. The narrative walks through building small servo-motor "robots," and each successive chapter, the robots gain a little in complexity by simple decisions: deciding to cross some wires, creating new behaviors. Braitenberg then retrospectively tries to analyze the robots behaviours without using the knowledge of how they were built, and it proves really hard to adequately point to what components lead to certain behaviors. He's of course using the text to support neural net theory, which he's researching, but it reads really nicely - almost reminds me of Calvino. It might be a similar endeavor, and lend an interesting (aesthetic) format, to starting from complexity and tracing back.