It sounds like — and this is from far off, not from any discussions with folks at Medium — that those working this at Medium have come to a realization: boiling down a set of key principles to animate and structure a culture should precede the development of the rules that encode a model of governance based on those principles.
Isn’t that the right order?
But Holacracy is generally adopted tactically — here’s how roles work, how circles work, etc. — without much, or enough, of the cultural foundation being worked on. As a result, there can be a deep mismatch between cultural norms and expectations, and the channels of the governance model.
The principles listed show that Medium is asking the right sorts of questions, like these: How are decisions made? How do we balance the authority derived from greater and lesser degrees of ‘ownership’ in a company (or seniority, status, or popularity) with the immediacy of new ideas (or insights, data, or hunches). When is change the right next thing to do, as opposed to doing something the more established way? Who can commit how much of the company’s resources and when?
Let’s back up a bit. Holacracy is only one alternative to conventional management, and does not necessarily line up with the sort of deep culture that Medium — and other companies like Medium — need.
Some of the comments others have made here seem to take a binary stance: if you try Holacracy and reject it, that means either a/ you aren’t doing it right, or else b/ Holacracy is broken. The reality is — I feel — more nuanced.
I will follow the storyline at Medium — to the extent that they will let me — with great interest. I also suggest that those who are interested in these questions might read about SPADE, a decision-making framework developed by Gokul Rajaram, who oversees Caviar at Square, and Square colleague Jeff Kolovson (see Square Defangs Difficult Decisions with this System — Here’s How).
Read that, then ask yourself this: Maybe we don’t need a system as broad as Holacracy to democratize operations. Perhaps just fixing the decision-making process — and the way that it articulates with everything else — is enough.