The startup world has spent fifteen years treating co-founder conflict as a leading indicator of failure.
The argument is that the founder relationship is a company's primary operating system. When that system is strained, everything downstream breaks. It's elegant, intuitive, and entirely untested against actual evidence.
The flaw is not in the observation that founders do experience tension, but in the causal leap that follows. The reasoning skips directly from "tension exists" to "tension predicts failure" without ever showing the mechanism. No numbers, no cohort analysis, no comparison between companies with documented founder friction and their control groups.
Look at what actually happened at Stripe. Patrick and John Collison have never hidden their fundamental differences in temperament and working style. Patrick operates as a visionary impatient with implementation detail, while John is a ruthless optimizer obsessed with execution mechanics. By the logic of the invisible-tax argument, this pairing should have created drag. Instead, the company reached a $95 billion valuation. The tension between their styles appears to have been a feature, not a bug.
A strained relationship that produces clarity through conflict is not the same as a strained relationship that produces paralysis.
The original claim is right about one thing. The founder dynamic matters enormously. But "mattering" and "being predictive of failure" are not the same thing. A strained relationship that produces clarity through conflict is not the same as a strained relationship that produces paralysis. What we've actually learned from tracking founders is that market timing, capital velocity. Early hiring choices explain far more variance in outcomes than whether the founders like each other. The corrected picture changes how you should think about any high-stakes partnership, not just startups. It's not whether you get along, but whether your friction produces decisions or only exhaustion.