Someone forwarded me a Threads post last week. The question it opened with: why do two bright, ambitious, independent people come alive around each other - building careers, moving fast - and then go flat the moment they merge their lives and become one?
The author’s answer wasn’t psychological. It was thermodynamic. Two principles: Carnot’s theorem and the perfectly inelastic collision.
I read it thinking about romantic partnerships. I kept coming back to co-founders.
The thermodynamic case against alignment
Carnot’s theorem defines the upper limit on any heat engine’s efficiency: η = 1 - T_cold/T_hot. Efficiency depends entirely on the temperature gap between the two reservoirs. When temperatures equalize - when T_cold approaches T_hot - efficiency approaches zero.
The engine needs a differential to run.
Most co-founder advice works against this. The standard investor question: do you and your co-founder share the same vision? The answer they want to hear is yes. But complete shared vision is temperature equalization. The more fully two founders converge - on risk appetite, time horizon, what the company should become - the less heat the engine produces.
This doesn’t mean founders should want opposing things. It means the structural relationship between them should preserve difference. One founder holding a six-month view and one holding a five-year view isn’t misalignment to fix. It’s pressure differential - the kind the engine runs on.
The alignment trap works like this: every time two founders come out of a meeting having resolved a disagreement into consensus, it looks like the partnership is working well. They agree more. They friction less. The temperature gap quietly closes. Ask most co-founder pairs to describe a productive week and they’ll describe a week with less friction - which is also a week with less differential. Consensus is how you run out of pressure without noticing.
The co-founder pairs that stay productive for years tend to have a visible asymmetry they protect deliberately. One holds the accelerator, one holds the brake. One carries the long view, one has the kill switch on current reality. They don’t agree on everything - they agree on what they’re each responsible for, and they stay different. The structure is what keeps the gap open.
What the collision actually costs
The second principle is the perfectly inelastic collision.
When two objects collide and merge into one, momentum is conserved. The combined mass continues moving. But kinetic energy is not conserved. It converts into heat, internal deformation, the structural cost of two things becoming one.
In co-founder terms, that deformation cost is the alignment tax. Decisions one person used to make alone now require consensus. Actions that were immediate now need coordination. Checking before changing something. Calls to synchronize before moving. Meetings to maintain the merge.
This is where the kinetic energy of the partnership goes - not into building, but into holding the fusion together.
None of this is optional once the merge has happened. The company keeps moving - momentum transferred. But the energy available for actual work is lower than the sum of what each person brought in, by exactly the amount that went into becoming one thing from two.
This is why acquisitions reliably lose velocity. Not because the acquired team got worse. Because a fraction of their energy is now structural overhead that didn’t exist before the collision. The momentum made it. The kinetic energy didn’t fully come with it.
What elastic looks like
The elastic collision is the alternative.
Both objects hit. Both change trajectory. The total kinetic energy of the system stays constant - the faster one slows slightly, the slower one accelerates, and when they separate, the system has lost nothing. Both exit changed, but the energy is all still there.
The structural conditions for elastic are specific. Each founder needs clear, bounded authority - domains where they’re final, where the collision stops before it becomes a merge. Time horizons should be genuinely different, not converged by alignment rounds. And the partnership needs enough tolerance for unresolved tension to let two people stay two people.
Diego Oppenheimer’s co-founder framework puts defined roles and explicit hierarchy at the center - not as bureaucracy, but as the thing that keeps collisions elastic. Clear domains mean a disagreement can resolve as two changed people with intact authority. Blurry authority means every disagreement risks becoming a merge event. Every merge event has a deformation cost.
This isn’t an argument for co-founder conflict as a value in itself. The data on partnership breakdowns is real, and advice to align your vision comes from watching actual companies come apart. But the standard response - more consensus, less friction, converge the priorities - treats the symptom and removes the mechanism. What Carnot actually says isn’t “keep the reservoirs at opposing extremes.” It says don’t let them equalize. Value alignment - shared mission, shared operating standards - isn’t a temperature to disrupt. Operational differential - risk appetite, time horizons, domain ownership - is. Most alignment advice collapses the two into one.
The test: when did you last come out of a conversation with your co-founder having changed your position - and they changed theirs?
A founder changed my view on two different questions in a single conversation last week. Neither of us left thinking what we’d thought when we arrived.