Conflict without a boss
In a hierarchy, an unresolved dispute travels upward until someone decides it. Remove that escape route and you need a replacement — otherwise conflicts do not disappear, they go underground. This is the part of self-organization most often left unbuilt.
Why this cannot be skipped
Most accounts of self-organization spend their pages on structure and decision-making, and go quiet on what happens when two people cannot agree, when someone does not deliver what they promised, or when a colleague has to be let go. That silence is not an oversight — it is the hardest part.
The hierarchy at least had an answer, even a poor one: escalate. Somebody senior decides, everyone lives with it. Take that away without providing something else, and one of two things happens. Either the conflict is avoided until it becomes personal and someone quietly leaves, or an informal authority steps in — the founder, the longest-serving colleague, whoever has the most standing — and you have rebuilt the hierarchy without admitting it.
The four-stage path
The best-documented answer comes from Morning Star, a tomato processing company of several hundred people with no managers, running the same process for decades. Its logic is simple: the conflict stays with the people involved as long as possible, and each stage only opens if the previous one failed.
The two people involved
They talk. That is the whole stage, and it resolves the overwhelming majority of cases. The obligation is not to succeed — it is to try first, rather than complaining sideways or waiting for someone to notice.
The two, plus one colleague they jointly choose
If the conversation fails, they pick a colleague together to help — not appointed, not senior by rank, and with no authority to impose anything. The mediator's job is to help them reach their own agreement.
A small group of colleagues, again jointly chosen
If mediation fails, several colleagues hear the case together. They can recommend, and their recommendation carries weight through reputation rather than power. The number of cases that reach this stage is small.
The last resort, rarely used
Only when everything else has failed does the person carrying ultimate liability decide. This stage exists deliberately and is named openly — which is exactly what makes it rare. An honest last resort is better than a pretend flatness in which someone decides anyway, only without anyone admitting it.
Why the sequence works
It is graduated, not all-or-nothing
Ostrom found the same pattern in communities that governed shared resources successfully for centuries: sanctions and conflict handling must start mild. Where the only options are "let it go" or "full escalation", people choose to let it go — and the problem accumulates.
Nobody is assigned from above
The mediator and the panel are chosen by the parties themselves. That is what separates this from an HR process: the people in conflict retain agency over how it is handled, which makes them far more likely to accept the outcome.
The exit is named
Stage four is the honest admission that someone still carries final liability. Models that deny this end up with an unofficial version of it — which is worse, because it cannot be appealed and nobody agreed to it.
And dismissal?
This is the question that ends most enthusiastic conversations about self-management. Somebody has to be able to say that a working relationship is over — and in most jurisdictions, including Germany, that is a legal act requiring a legal entity behind it, not a team decision.
✅ The process before it
- Naming the problem early, through the same four stages.
- Colleagues, not a manager, giving the feedback.
- A documented attempt to resolve it before separation.
- The decision to recommend separation, made by peers.
❌ The legal act
- A termination is signed by whoever is legally entitled to.
- Works council rights, notice periods and protection against dismissal apply regardless of your operating model.
- Employment law does not recognize "the cell decided".
This is one of the clearest cases where the formal structure has a legitimate, irreducible job. It should be small — but for this, it exists.
Where it strains
It only works with psychological safety. Stage one assumes people can raise something uncomfortable face to face. Where that is not safe, the process is a form on a wall: everyone knows it exists and nobody uses it. Build that foundation first, or the rest is decoration.
Peers can be harsher than a boss. James Barker's finding applies here too — colleagues judging colleagues can produce more pressure, not less, because there is no longer a manager whose job includes absorbing some of it. A panel of peers is not automatically gentler than a supervisor.
Power imbalances do not vanish. A new joiner and a founder choosing a mediator "together" are not choosing on equal footing. The process makes influence visible, but it does not neutralize it.
It requires a written commitment to have been broken. The whole thing works better when there is something concrete to point at — an agreed deliverable, a seam agreement, a role someone took on. Conflicts about vague expectations are much harder to handle at stage one, which is an argument for writing agreements down in the first place.
How BetaOS relates to this
Two of Ostrom's eight principles in the governance view ask exactly about this: whether there are graduated responses when a commitment is broken, and whether there is a known, low-threshold path for conflict inside a cell and at its seams. Most organizations answer "no" or "unclear" to both — which is the finding, and the reason the question is asked.
The four-stage process itself is not modelled in the product. It is sketched as a possible future state machine, but building a conflict tracker before organizations have a conflict process would be the wrong order — and a piece of software is a poor place to hold something this personal. What the product can usefully do is make the agreements visible that conflicts tend to be about.
Sources
- Morning Star Self-Management Institute: the Colleague Letter of Understanding and the four-step conflict process (Direct Communication → Mediator → Panel → Rufer).
- Elinor Ostrom: Governing the Commons. Cambridge University Press, 1990 — principles 5 and 6 on graduated sanctions and conflict-resolution mechanisms.
- James R. Barker: Tightening the Iron Cage. Administrative Science Quarterly, 1993 — on peer control.
- Frederic Laloux: Reinventing Organizations. 2014 — conflict practices at Morning Star, Buurtzorg and FAVI.
- Amy C. Edmondson: The Fearless Organization. Wiley, 2019 — why stage one presupposes safety.
More from the Beta Library: betaos.org/library