the binding affinity of cofounders

I was thinking about biophysical interactions, as I often do, and realized the same ideas apply to cofounders. In molecular biology, complementarity gets you docking: two shapes fit together. But docking scores are bad at predicting durable complexes. What actually predicts residence time is binding affinity, how hard it is to pull the complex apart once it has formed.

When people look for a cofounder, they usually screen for complementarity, one technical and one commercial. But complementary skills that are replaceable generate zero dependence. The moment outside capital or talent can replace your cofounder's function, the partnership is running on goodwill. Zuck and Eduardo had textbook complementarity, product and capital, and then Peter Thiel's check and Sean Parker showed up, Eduardo's function got commoditized, and it blew up. The Collison brothers look redundant on paper, two near-identical technical generalists, and Stripe is one of the most durable partnerships in tech.

So I think the more important variable is mutual leverage: each person's best possible path runs through the other, and both know it. This is also what kills resentment. Resentment needs a story where you'd be better off alone, and real mutual leverage means that story is never true.

Alignment is still the precursor -- same direction, same speed, same risk appetite -- because leverage is an amplifier, and amplified misalignment is probably worse.

The leverage has to do three things. It has to come from the value of staying rather than the cost of leaving (golden handcuffs are leverage, but they produce the most resentment). It has to stay roughly symmetric over time. And it has to compound as the company grows. Checks, intros, and the v1 codebase all decay, while context, trust, and owned domains compound. Most cofounder blowups have the geometry of front-loaded leverage decaying against permanent equity. Vesting is basically the institutional patch for getting this wrong.

Biotech has run this experiment before. Genentech was Herb Boyer and Bob Swanson, recombinant DNA and a 28-year-old VC who could raise money and build a company. Each put in $500, and neither had anything without the other. Regeneron has run on Len Schleifer and George Yancopoulos for almost four decades, one owning the company and the other owning the science, with the dependence compounding through every platform George built. The modern version is the wet lab / dry lab pair: one founder owns the experimental platform, the other owns the models, and every cycle the lab generates data only that platform can produce while the models turn it into predictions only that data can train. Recursion started exactly this way, Chris Gibson on phenomics and Blake Borgeson on compute, and Inceptive is Jakob Uszkoreit and Rhiju Das, a transformer co-author paired with a leading RNA experimentalist. This archetype is young, so we'll see how it plays out.

The implication I keep coming back to is that the best people have options by definition, so you can't get leverage on them by making their exit expensive. The only durable leverage on someone with options is the size of the thing you're building together. At the end of the day, the mission is the leverage generator.