I don't dispute that informal power structures exist even in organizations whose hierarchies are formally defined. However, a formal hierarchy serves as a baseline, and one can measure the level of political risk they're taking by noticing how much their actions are diverging from the formal hierarchy. If I bypass the chain of command, so to speak, and bring my concerns directly to a high-level executive, I can roughly judge who I'm going to be offending by taking that step. How does one do that in a holocracy? How do I even know who the "high-level executives" are? The holocracy literature completely hand-waves this aspect of the organization. Holocracy advocates say that you shouldn't worry; with time it becomes clear who the influencers are and who one should talk to for various matters. I don't agree with that at all. As anyone who remembers attending a public high school can tell you, humans will naturally form cliques, and it's not always obvious who the influencers are for a given clique. Moreover, as Paul Graham points out, hackers are much less interested in social dynamics (i.e. "politics" and "drama") than average people, and so are much more likely to make political miscalculations in a holocracy. It is for this reason that I consider holocracy to be a form of organization that is inherently hostile to hackers. It's a form of organization that seems custom-designed to hit us where we're weakest.
For me it's good to know that people still exist who can believe in the first. I also started out like that but the little experience I could gather from this world taught me otherwise.