The failure is not usually a surprise. That is the thing that survivors of co-founder splits most often say afterward: in retrospect, the signs were there early. The disagreement about the hire that never fully resolved. The equity conversation that ended ambiguously. The moment one person started describing the company's vision without "we." The failure of a co-founder friendship is rarely sudden. It is a long deterioration that accelerates at a terminal event, and the terminal event is usually preceded by years of small divergences that neither party named.
Co-founder friendships fail along patterns that are well-documented and predictably ignored. Noam Wasserman's research at Harvard, tracking hundreds of founding teams over years, identified a set of structural fault lines: equity splits decided too quickly under the social pressure to seem non-greedy with your friend; role ambiguity that functions as a latent conflict-avoidance mechanism until the company reaches a scale where the ambiguity becomes operationally destructive; vision divergence that is manageable in early days when the company's direction is abstract but becomes irreconcilable once a concrete choice has to be made. These are not personality failures. They are structural failures with predictable triggers, and the reason they keep happening is that the founding moment — the heady period of mutual excitement and shared commitment — is among the worst times to do the clear-eyed planning that would prevent them.
The friendship is both the asset and the liability. The trust that allows two people to quit their jobs together, to sleep on each other's couches during the early grind, to work eighteen-hour days without resentment — that trust is real and it matters. But it is also the mechanism that suppresses the conversations that would surface the divergence while it is still correctable. You do not tell your friend that you think they are the wrong person to be CEO when the company has six employees and you are still figuring out whether the product works. You manage around it. The management compounds. When the conversation finally happens — usually under investor pressure, or at a board meeting, or in a confrontation that has been building for two years — it has too much history on it to be resolved cleanly.
Law 5 — Revise — names the operating failure precisely: the co-founder relationship needs continuous revision, and the friendship makes continuous revision feel like continuous conflict. Founders who are friends tend to treat relational maintenance as relational equilibrium — as long as no one is openly fighting, things are fine. But the silence is not equilibrium; it is accumulation. The founding agreement signed in year one is not a living document; it is a snapshot of what two people thought they wanted before they knew what the company would become. The relationship it structures requires active revision as the company evolves, and the revision requires exactly the kind of direct, uncomfortable naming that friendship's social logic tends to defer.
What makes this a collective problem rather than a personal one is the scale of predictability. The same failures happen in founding teams across industries, funding stages, and cultural contexts. The field of study is mature enough that the failure modes are classified. The interventions that would prevent them — explicit co-founder agreements, early clarity on decision rights, deliberate role definition, founder therapy — are known, available, and underused. They are underused primarily because the founding moment is not a moment that feels like it requires defensive planning. It feels like the beginning of a joint adventure with someone you trust. The planning feels like distrust. The distrust would feel like betrayal. So it is deferred, and the failure arrives on schedule, and everyone is surprised.
The collective revision that is overdue is a change in how founding teams are prepared — not just in tactics or term sheet literacy, but in a candid reckoning with the specific ways that friendship's logic conflicts with the organizational logic of a scaling company, and with the specific conversations that need to happen early rather than late. The archive of co-founder failures is extensive. The use of that archive in founder preparation is not.