The case against self-organization
Every account of decentralized organization has the same weakness: it is written by people who believe in it, citing companies that succeeded. Here are the strongest objections — including the ones Beta does not answer well.
Why this article exists
The literature on self-organization has a survivorship problem. The famous examples — Buurtzorg, Morning Star, FAVI, Handelsbanken, Haier — are the ones that worked. The companies that tried something similar and quietly went back are not written up, because nobody publishes a case study about a reorganization that fizzled.
That doesn't make the successes fake. It does mean the base rate is unknown, and anyone selling this approach — including us — owes you the counter-arguments in the same document as the promises.
The five objections that hold up
1. There is no such thing as a structureless organization
Jo Freeman's argument from 1972, written about the women's movement, applies directly: when you remove the formal structure, an informal one takes its place — unelected, unaccountable, and far harder to challenge, because officially it doesn't exist. Friendship networks become power networks. Newcomers can't find the door.
Where Beta answers it: cell structures are explicitly a different structure, not the absence of one — named cells, named roles, written agreements between them. Where it doesn't: that only works if the agreements are actually written down and visible. Where they aren't, Freeman's objection lands unchanged.
2. Peer pressure can be harsher than a boss
James Barker studied a company that replaced supervisors with self-managing teams and found control did not disappear — it moved inside the group and got tighter. Team members policed each other's hours more strictly than any manager had. Values hardened into unwritten rules that were impossible to appeal, because there was no longer anyone to appeal to. One worker's summary: I feel like a supervisor now, but I don't get paid for it.
Where Beta answers it: not really. Transparency and market coupling are meant to replace supervision, but they can just as easily amplify peer scrutiny. Honest position: cells are not immune to this, and a self-organized team can produce more overwork than a hierarchical one.
3. Careers, pay and status don't disappear just because positions do
When Zappos adopted Holacracy and offered anyone who disliked it a severance package, roughly 18 % of employees left. Rule bureaucracy and meeting overhead were part of it, but so was something more basic: without positions, there is no promotion, and it becomes unclear how anyone advances or what they are worth. The model was later quietly rolled back.
Where Beta answers it: partially — roles instead of positions, pay the person rather than the box. Where it doesn't: the question of how someone grows, and how compensation is decided without a ladder, is genuinely open. This is the single biggest unsolved problem in the whole field.
4. The model companies have bad years too
Buurtzorg is the strongest evidence for decentralized organization anywhere — and in 2013 it made a substantial loss, when payers refused to cover roughly 9.5 million euros of care the teams had delivered. Attempts to replicate the model abroad have largely failed. It also depends on a strong centre: a small back office, a purpose-built IT platform, and a founder whose role is hard to transfer.
Reading: this is not a refutation — Buurtzorg still delivers roughly 30 to 50 % fewer care hours at higher satisfaction. But "self-organization works" is too simple. It worked here, in this sector, with this founder and this platform.
5. Some of the theory is not science
Frederic Laloux's developmental stages — the colour scheme that culminates in "Teal" — have little empirical support, and the book's evidence is twelve hand-picked companies. Naming a stage does not demonstrate that organizations move through it, or that later stages are better. Read as description it is illuminating; read as proof it is not.
Where Beta answers it: the BetaCodex argues structurally rather than developmentally — right structure, not higher consciousness — which avoids this problem entirely. But: it borrows credibility from the same case studies, and those carry the same selection bias.
The objection from inside the field
The sharpest criticism doesn't come from defenders of hierarchy. It comes from people who agree that command-and-control is finished, and disagree about what follows.
A recipe is still a recipe
- Capability sticks to the capable person and cannot be transferred as a method.
- A transferable structural blueprint is therefore itself the "dead" kind of thinking it claims to replace.
- Internal markets without genuine choice are theatre, not markets.
- And bluntly: decentralization as dogma is romanticism — strategy and innovation belong in the centre.
There is no single target state
- No practice is universal; every answer is bounded by its context.
- Ordered, rule-based work is a legitimate domain, not a defect to be abolished.
- An idealized future state is the classic reason change programmes fail.
- Measure direction and speed from where you are — not distance from a model.
Both are worth more than any defence of the pyramid, because they accept the diagnosis and attack the cure. If you only read one criticism of Beta, make it one of these.
What follows from all this
A practical reading list of what to watch for, if you are doing this:
- Write the agreements down. Unwritten structure becomes invisible power. This is the cheapest protection against Freeman's objection.
- Watch for overwork, not just output. Rising hours in a self-organized team are a warning sign, not a sign of commitment.
- Answer the career question early. Before you remove positions, know what you will say when someone asks how they get ahead.
- Distrust your own case studies. Including the ones in this library.
What this means for BetaOS
Two design decisions in the product come directly from these objections. There is no maturity score — no percentage of how "Beta" an organization is — because that would measure distance from a target state, exactly what Snowden identifies as the failure mode. The governance view shows a pattern across cells, deliberately without a total.
And the structural views exist to make informal power visible rather than tidy. An organization chart that looks neat while the real decisions happen elsewhere is the problem Freeman describes. Showing all three structures side by side is uncomfortable on purpose.
What the product does not solve: the career question, and peer pressure. Those are not software problems, and claiming otherwise would be the kind of overreach this article is about.
Sources
- Jo Freeman: The Tyranny of Structurelessness. 1972.
- James R. Barker: Tightening the Iron Cage: Concertive Control in Self-Managing Teams. Administrative Science Quarterly, 1993.
- On the Zappos/Holacracy experience and the ~18 % attrition after the 2015 offer, and the later rollback.
- Buurtzorg: 2013 loss on roughly 9.5 million euros of unreimbursed care; difficulties replicating the model internationally.
- Frederic Laloux: Reinventing Organizations. 2014 — and the methodological criticism of its developmental stages.
- Gerhard Wohland, Matthias Wiemeyer: Denkwerkzeuge der Höchstleister. 2012.
- Dave Snowden, Alessandro Rancati: Managing complexity (and chaos) in times of crisis. A field guide. EU Joint Research Centre, 2021.
More from the Beta Library: betaos.org/library