Boards want speed and focus. Talent wants autonomy and voice. Many leaders try to square that circle by flattening the org and letting networks carry the load. Then decisions stall, invisible power brokers appear, and the company quietly rebuilds a chain of command. The real choice is not hierarchy versus network. It is formal power you can see versus informal power that runs the place anyway.
What self-management actually promises
Michael Lee and Amy Edmondson offer the clearest academic definition of a self-managing organization. It radically, formally, and systematically decentralizes authority across the whole firm, and it removes the manager–subordinate relationship as the basic building block of control (Lee and Edmondson, 2017). Their synthesis also notes why this agenda keeps surfacing. Traditional managerial hierarchy is good at reliable execution of known tasks in stable settings. It is less good when work is complex, crosses functions, and demands learning, where status differences can choke initiative and development (Lee and Edmondson, 2017).
That is a real promise. It is also bounded. Lee and Edmondson stress that most evidence here is theoretical and case-based rather than experimental, so strong claims about universal superiority are not supported (2017). The design choice is contingent. Leaders must match structure to the problem they are trying to solve.
When flat turns slow
Nicolai Foss and Peter Klein argue that so-called bossless firms often retain bosses in practice, only informally named. Under stable conditions a flat model can muddle through. In dynamic and time-sensitive markets, however, these firms tend to struggle with innovation, scale, and longevity, because decision rights become unclear and coordination costs rise (Foss and Klein, 2022). The hierarchy does not vanish. It migrates to those with status, information, or control of scarce resources.
Valve is a cautionary case in their analysis. The company championed a flat structure yet, by 2013, insiders described its creative state as “The Wilderness,” a period of drift. Even the acclaimed launch of Half-Life: Alyx required management intervention despite the bossless doctrine (Foss and Klein, 2022). The lesson is prosaic. When stakes or tempo rise, someone must decide, sequence, and enforce. If you have not defined who that is, the role finds a person anyway, only now without accountability or transparency.
A second line of evidence reinforces the point from a different angle. A strategy+business review of Why Managers Matter summarizes Foss and Klein’s view that near-bossless firms do not reliably outperform conventional ones, and that managers matter in concrete functions: designing, organizing, encouraging, and enforcing. The authors distinguish “Mark I” authority (hire, fire, instruct) from “Mark II” authority (lead, structure, build consensus) and argue the mix should vary by context rather than by ideology (strategy+business, 2022). That frame is useful because it gives leaders levers. You can choose where hard authority sits without defaulting to command-and-control everywhere.
Zappos, from doctrine to design change
Zappos is often invoked as a parable of self-management. In 2016 it was the largest company to implement Holacracy fully, a fact that fueled a polarized debate that Harvard Business Review described as hype for and against “bossless” management (Bernstein, Bunch, Canner and Lee, 2016). The arc since then is instructive. According to Quartz reporting, after a 2015 ultimatum to commit to Holacracy or accept a buyout, 18 percent of employees chose to leave. By 2017 the company pivoted away from pure Holacracy. It eliminated many of the system’s formal meetings, brought managers back, and introduced an internal market model in which teams operate like small businesses with their own P&Ls and trade services at prices, while keeping a circular hierarchy as an artifact (Quartz, 2020).
Taken together, those moves look less like renouncing autonomy and more like constitutionalizing it. Choice and networked teams remained, but decision rights and accountability for results became legible again.
Design networks with a constitution
The practical read is not to retreat to the old org chart. It is to fix the parts of hierarchy that block complex problem solving while specifying authority where time, risk, or interdependence demand it. A workable pattern emerges from the evidence.
- Define the problem class. If the work is novel and cross-functional, give teams formal discretion within scoped domains, which addresses the learning and motivation limits of traditional hierarchy (Lee and Edmondson, 2017).
- Make authority types explicit. Decide where “Mark I” authority sits for hiring, budgets, and irreversible bets, and where “Mark II” authority sits for convening, shaping interfaces, and building consensus. Publish this so no one needs to guess whom to ask or who can say yes (strategy+business, 2022).
- Install escalation rules for tempo. In dynamic contexts, unresolved decisions age badly. Set clear thresholds for when a role can decide unilaterally, when to seek advice, and when to escalate for a binding call, addressing the informal-boss problem flagged by Foss and Klein (2022).
- Tie autonomy to a result that can be priced or measured. Zappos’ shift to internal P&Ls and service transactions was an attempt to make accountability concrete while keeping networked teams, a move Quartz documents even as it reports the retreat from Holacracy’s mechanics (Quartz, 2020).
- Audit the shadow hierarchy. Map who actually gives approvals and who gates information. Where you see power aggregating informally, either formalize the role with obligations and transparency or redesign the workflow to break the bottleneck. This is the day-to-day challenge that sits behind Valve’s “Wilderness” episode in Foss and Klein’s account (2022).
The network is not the opposite of hierarchy. It is the terrain on which authority operates. Leaders who name where hard power lives, and why, free the rest of the system to act with confidence. The prize is not a flattering chart. It is faster decisions with fewer surprises, and a culture that knows who decides what, on what grounds, and to what end.
Sources
- Lee and Edmondson (2017). Self-managing organizations: Exploring the limits of less-hierarchical organizing. Defines SMOs and synthesizes when hierarchy helps or hinders, grounding the case for formal autonomy with clear limits. https://www.sciencedirect.com/science/article/abs/pii/S0191308517300059
- Foss and Klein (2022). Who's the boss? The persistence of entrepreneurial hierarchy in flat organizations. Argues that flat firms often grow informal hierarchies and struggle in dynamic contexts, illustrated by Valve’s “Wilderness” and decision needs. https://link.springer.com/article/10.1007/s41469-022-00132-8
- strategy+business (2022). Can bossless management work? Summarizes Foss and Klein’s view that managers matter, and distinguishes Mark I and Mark II authority as a contingent design choice. https://www.strategy-business.com/article/Can-bossless-management-work
- Bernstein, Bunch, Canner and Lee (2016). Beyond the Holacracy Hype. Frames the polarized debate on bossless management and notes Zappos as the largest full Holacracy adopter at the time. https://hbr.org/2016/07/beyond-the-holacracy-hype
- Quartz (2020). Zappos has quietly backed away from holacracy. Reports the 18% buyout, rollback of Holacracy’s mechanics, reintroduction of managers, and creation of an internal market with team P&Ls. https://qz.com/work/1776841/zappos-has-quietly-backed-away-from-holacracy