Convexity, Uncertainty, and the Society We Build
Convex payoffs — small downside, large upside — are not just a financial concept. They describe a class of decisions that are structurally different from the ones most people make most of the time. What follows from applying this lens to how we organize collective life?
A concept from options pricing
Convexity, in the mathematical sense, describes a function where the output accelerates relative to the input. The curve bends upward. On the downside, losses are bounded. On the upside, gains are unbounded.
In financial derivatives, convexity is what makes options asymmetric. You pay a premium for the right, but not the obligation, to participate in a move. Your loss is capped at the premium. Your gain is theoretically unlimited.
This is not just a trading concept. It describes a structural property of certain kinds of decisions — and certain kinds of social arrangements.
The convexity of optionality
A decision is convex when the cost of being wrong is bounded and recoverable, while the benefit of being right is large and scalable.
This is the structure of: a researcher who pursues speculative hypotheses, knowing that most will fail but one might transform a field. A startup that operates cheaply, failing fast, until it finds something that works at scale. A country that protects dissent, absorbing the cost of ideas that are wrong, to preserve the conditions under which ideas that are right can propagate.
Note what all of these have in common. They require tolerance for frequent small losses. Most experiments fail. Most startups die. Most dissidents are wrong about most things.
The convex arrangement is not one that eliminates failure. It is one that keeps failure cheap and keeps success uncapped.
The concavity of control
The opposite of convexity is concavity: bounded upside, accelerating downside.
Concave arrangements feel safer in the short term. A society that suppresses deviant ideas avoids the cost of tolerating bad ones. A company that enforces conformity avoids the friction of internal disagreement. A person who avoids risky experiments avoids the discomfort of failure.
The problem is the asymmetry compounds in the wrong direction.
Suppress enough dissent, and you lose the mechanism for identifying error. Enforce enough conformity, and the organization becomes brittle — optimized for the environment it was built in, incapable of adapting when that environment changes. Avoid enough experiments, and you accumulate a kind of epistemic debt: a widening gap between what you believe and what is true.
The tail risk of concave arrangements is catastrophic. Not dramatic collapse — usually slow rigidity, then sudden irrelevance.
Uncertainty is the point
The standard argument for convex institutions goes: we don't know in advance which ideas are right, so we need systems that allow many ideas to compete.
This argument is true but undersells the case.
The deeper point is that uncertainty is not a temporary condition to be eliminated. It is the permanent condition of operating in a complex, non-stationary world. Models go stale. Environments shift. The people running institutions die and are replaced by people with different priors.
Convex arrangements don't just hedge against current uncertainty. They are robust to future uncertainty — including uncertainty about the nature of future uncertainty. They don't require the designer to correctly predict which dimension of adaptability will matter.
This is the sense in which convexity is not just a good strategy but a meta-strategy. You don't need to know what changes are coming. You need to be the kind of system that can absorb changes you didn't anticipate.
What this means for institutions
Most human institutions are not designed for convexity. They are designed for efficiency.
Efficiency is the optimization of known processes for known objectives under known constraints. It is anti-convex almost by definition: it eliminates slack, redundancy, and variance. It concentrates authority, because distributed authority is slower. It standardizes process, because deviation introduces variance.
Efficiency is the right objective for a stable environment with well-understood goals.
It becomes a liability when the environment changes, when goals need to be revised, or when the map diverges from the territory. Which is to say: efficiency is a liability under uncertainty, and the world is uncertain.
Convex institutions maintain slack. They tolerate redundancy. They distribute authority not because it is faster, but because it is more robust. They fund speculative work even when the expected value calculation doesn't close — because expected value calculations under deep uncertainty are unreliable.
The political difficulty
None of this is controversial in abstract. Most people, asked in the abstract, will agree that optionality is good, that resilience matters, that diverse approaches to hard problems outperform monocultures.
The difficulty is that convex arrangements are politically hard to sustain.
The benefits of convexity are diffuse and long-term. The costs are concentrated and immediate. Tolerating bad ideas costs something now. The benefit — that the process which tolerates bad ideas also surfaces good ones — is statistical and visible only over time.
Meanwhile, concave arrangements produce visible short-term results. Centralizing authority is faster. Eliminating redundancy is cheaper. Suppressing variance reduces noise.
The political economy of convexity is structurally disadvantaged against the political economy of control.
This is why it is not enough to argue for convex institutions on principled grounds. The incentive structures have to be built in. Tenure, academic freedom, independent judiciaries, antitrust law, reserve requirements, redundant infrastructure — these are all mechanisms that impose short-term costs to purchase long-term convexity. They are all perpetually under pressure from actors whose time horizon is shorter than the benefit they provide.
The society we build
The question of what kind of society we build is, in part, a question about what time horizon we are optimizing for — and who bears the costs of the arrangement.
A society that concentrates decision-making, eliminates redundancy, and optimizes for measurable short-term output is building for concavity. It may perform well for a generation. The tail risk is inherited by whoever comes after.
A society that distributes authority, maintains slack, and tolerates variance is building for convexity. It will appear less efficient. It will produce more failures. It will also produce more recoveries.
This is not a left/right axis. It is a time-horizon axis. It is a question of who benefits and who pays — and when.
The answer to that question is the answer to what we are building.