Return-to-office edicts, AI pilots, and rolling restructurings have revived a familiar promise: make the company flatter so decisions move faster. Then a product launch slips because two squads cannot agree on a dependency. A customer issue whipsaws between teams with no owner. The network stalls. The hard question surfaces. Who decides, and when?
Why networks stall at the moment of truth
Managers exist for a reason. The academic synthesis on self-managing organizations is clear about where classic hierarchy excels: reliable execution of known tasks in stable environments, with clear lines of accountability (Lee and Edmondson, 2017). The same synthesis warns that hierarchy tends to rigidify and dampen cross-boundary problem solving when the work is novel or complex, which is why leaders reach for self-management in the first place (Lee and Edmondson, 2017). But removing managers does not suspend the physics of coordination under pressure.
Evidence from firms that have tried to stay bossless shows the pattern. In dynamic and time-sensitive settings, flat organizations often struggle with innovation at scale and with timely decisions. People still look for authority, so formal hierarchies get replaced by informal ones rather than eliminated, which is harder to see and harder to hold accountable (Foss and Klein, 2022). Valve’s celebrated flat structure is a cautionary case. By 2013 employees described the firm’s creative state as “The Wilderness.” Foss and Klein report that Half-Life: Alyx required management intervention despite the stated bossless model, and that the absence of dedicated customer support and marketing functions left gaps the network did not fill (Foss and Klein, 2022). The mechanism is not mystery. Time-sensitive interdependencies demand clear, shared answers to two questions: who makes the trade-off, and on what clock.
Design a backbone, not a boss
The remedy is not to reinstall swaggering line bosses everywhere. It is to make the power you actually need explicit. The management roles that matter are concrete: design, organize, encourage, and enforce. Foss and Klein’s categorization is useful here, distinguishing authority that can hire, fire, and instruct from authority that sets structure, leads, and builds consensus. Both types exist in high-performing firms, and no single blueprint dominates across contexts (strategy+business, 2022). Treat these roles as a backbone the network can flex against, not a return to command-and-control.
Self-management, properly defined, is not the absence of structure. It is the radical, formal decentralization of authority across the organization, and its distinctive feature is the elimination of the manager–subordinate relationship. That design both demands and codifies local decision rights, escalation paths, and cross-team governance so work does not jam when priorities collide (Lee and Edmondson, 2017). Without that codification, teams improvise status hierarchies and unwritten vetoes. Decisions drift to the loudest voice in the chat.
A practical design move is to turn escalation into a service. Make time-boxed decision forums part of the operating system, owned by named roles with narrow mandates. Publish decision rights: which trade-offs a product trio owns, which a platform council owns, which go to an enterprise forum. Define the service levels for a yes or no. This is managerial authority of the second kind, structure-setting more than order-giving. It keeps the network fast by giving it a spine.
Zappos, the pivot that clarified power
The most famous experiment with a bossless blueprint showed how a backbone reappears when stakes rise. In 2016 Zappos was the largest company to implement holacracy in full, and the debate around it had already polarized between cheerleaders and skeptics (Bernstein et al., 2016). The next chapters are instructive for today’s leaders. In March 2015 CEO Tony Hsieh gave employees a stark choice to fully adopt holacracy or leave, and 18 percent opted for a buyout. Over time the firm stripped out the system’s bureaucratic meetings and brought managers back, leaving mostly the visual artifact of circular hierarchy. In 2017 Zappos replaced holacracy with an internal market where teams operated as small businesses with their own P&Ls and traded services at market prices, a switch motivated by tough business metrics and criticism that holacracy had grown inward-looking (Groth, 2020).
None of this reads as a failure of networks. It reads as a failure to formalize power where the network could not self-supply it. The internal market was a choice to re-specify decision rights and accountabilities in the language of prices and P&Ls, and to put a spine under the swarm. The return of managers was not a nostalgic relapse. It was a recognition that certain coordinating roles must be visible and empowered if speed and customer focus are to survive growth and shocks (Groth, 2020).
What to do Monday morning
Leaders face a practical design task, not an ideological one. Three tests will tell you if your network has a backbone:
- Time test: For cross-team trade-offs that affect customers or deadlines, who decides within 48 hours, and where is that written? If the answer is “it depends,” you have a shadow hierarchy. Use explicit Mark II authority to set structures and forums, and reserve Mark I authority for narrow, safety-critical calls (strategy+business, 2022).
- Ownership test: Which roles hold customer support, marketing, reliability, and security? If these are everybody’s job, they are nobody’s job. Self-management that works codifies local authority and escalation routes without reinstalling manager–subordinate dyads (Lee and Edmondson, 2017).
- Drift test: When disputes linger, who enforces the decision and on what SLA? If intervention requires heroics, formalize it. Informal bosses will emerge otherwise, and decisions will age out instead of landing (Foss and Klein, 2022).
The aim is not to choose between hierarchy and network. It is to couple a visible backbone to a flexible swarm so the organization can solve novel problems without freezing at the seams. The evidence suggests that bossless theater invites hidden power and delay, while self-management with formal decentralization and clear escalation keeps decisions moving when they matter most (Lee and Edmondson, 2017; Foss and Klein, 2022). That is a design choice, not a slogan.
Sources
- Lee and Edmondson (2017). Self-managing organizations: Exploring the limits of less-hierarchical organizing. Defines SMOs and argues hierarchy suits reliable execution while self-management needs formal decentralization to avoid rigidity. 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. Finds flat firms struggle under dynamic conditions and often revert to informal hierarchies, illustrated by Valve’s experience. 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 play functional roles, distinguishes types of authority, and stresses no universal best structure. https://www.strategy-business.com/article/Can-bossless-management-work
- Bernstein, Bunch, Canner and Lee (2016). Beyond the Holacracy Hype. Reports Zappos as the largest holacracy adopter at the time and frames the polarized debate to contextualize later shifts. https://hbr.org/2016/07/beyond-the-holacracy-hype
- Groth (2020). Zappos has quietly backed away from holacracy. Documents the 2015 ultimatum with 18% buyouts, the return of managers, and the 2017 shift to an internal market with team P&Ls. https://qz.com/work/1776841/zappos-has-quietly-backed-away-from-holacracy