The Temporary Prop
A reciprocal frame is a roof made of beams arranged in a circle. Each beam rests on the one before it. The last beam closes the ring by supporting the first. No central column. No hub. The structure holds itself up through mutual dependency — each element supports the next, and the next supports the next, all the way around.
They’ve been building them since the 12th century. Chinese temples. Japanese pavilions. Medieval European cathedrals. The geometry is elegant and the engineering is sound.
But here’s the thing nobody mentions in the architecture textbooks: you can’t build one incrementally. You can’t set down beam A and then lean beam B against it and then lean C against B. The ring only works when it’s closed. Until the last beam is in place, nothing supports anything.
So during construction, you put a prop in the middle. A temporary central support that holds everything in position while you assemble the ring. Once the ring is closed and the mutual dependency is established, you remove the prop. The beams carry themselves.
The interesting question is: how do you know when to remove it?
A startup has a founder who makes every product decision. In the early days, this is necessary — the team isn’t aligned yet, the vision isn’t shared, the coordination patterns haven’t formed. The founder is the prop. They hold the beams in position while the ring assembles.
Over time, the team develops its own coordination. Shared context, implicit norms, distributed decision-making. The ring starts to close. The question becomes: is the founder still load-bearing, or is the ring self-supporting with the founder still standing in the middle?
You literally can’t tell by looking. A reciprocal frame under construction looks the same whether the ring is carrying load or the prop is. The only way to find out is to remove the prop and see if the structure holds.
Founders who never step back never discover the answer. The team might be self-supporting — ready to carry its own weight. Or it might not be — still depending on the founder for every significant decision. Both states look identical from the inside. The test requires the absence.
This is different from the fragility question. Fragility asks: what happens when the central node fails unexpectedly? That’s a crisis scenario — the prop gets knocked out and you see what falls.
The assembly question is gentler and more important: is the central node still needed, or is it habit? Systems designed around a central coordinator often maintain that centrality long after the periphery has developed enough capability to self-coordinate. Not because the coordinator is doing essential work, but because removing the coordinator feels risky and nobody has tested it.
The prop that stays too long creates its own dependency. Teams stop making decisions they’re capable of making because the founder is available. The founder’s presence prevents the ring from discovering it can hold itself.
Reciprocal frames also teach something about redundancy that’s counterintuitive.
A single ring is fragile. Every beam is load-bearing. Lose one, and the whole structure cascades — not gradually, but in sequence. A collapses onto B, B onto C, all the way around. No graceful degradation. The mutual dependency that makes the structure self-supporting is the same mutual dependency that makes any single failure catastrophic.
You don’t fix this by making individual beams stronger. You fix it by adding a second ring, offset from the first, connecting different beams in a different pattern. Now each beam is part of two rings. A failure in one ring is caught by the other. The redundancy comes from having independent parallel structures, not from strengthening any single path.
This is true of organizations too. You don’t make a team resilient by making the lead architect more capable. You build a second coordination path — a different set of connections carrying a different kind of information. The architecture review board and the daily standup serve different functions and connect different people. Either one can fail without cascading because they’re geometrically independent. They share members but not topology.
The insight I keep coming back to: biology builds redundant systems through evolution. Millions of years of selection pressure produce organisms where no single component is a keystone because competitive pressures are distributed hierarchically.
Engineering builds redundant systems through design. You add the second ring on purpose, because you thought about what happens when the first one fails.
The difference matters when you’re choosing how to build. If your system evolved — grew organically, found its own patterns — the biological models are the right lens. You’re describing what emerged. The keystone analysis tells you where you’re vulnerable.
But if your system was designed — if someone assembled it with intention — the engineering models are more useful. You’re not describing emergence; you’re choosing architecture. The question isn’t “where are we fragile?” It’s “have we built enough independent rings, and have we tested whether they hold without the prop?”
Both questions are about resilience. They lead to different interventions. The biological intervention is to strengthen the weakest node. The engineering intervention is to add an offset ring. And the hardest intervention of all: stepping back from the center to see if the structure you built can stand on its own.