The Strait of Hormuz shows that some problems cannot be optimized

The Strait of Hormuz measures, at its narrowest point, just over fifty-four kilometers. A thin line between Iran and Oman, a thread of water connecting the Persian Gulf to the open sea. And yet through that liquid corridor flows roughly twenty percent of the world’s oil every day. Tankers loaded with energy glide in line between two wary coasts, under military eyes that never sleep. Strategists call it a chokepoint — a bottleneck.

The metaphor is perfect: the flow concentrates, pressure increases, and very little is needed to block everything. Global markets tremble whenever someone, in Tehran or Washington, raises their voice. The world, in all its digital modernity, remains anchored to a fifty-four-kilometer funnel.

Hormuz is not merely a strait. It is a lesson in applied epistemology. It tells us that in reality—physical as well as psychological—there exist knots, bottlenecks, points of irreducible complexity that no alternative architecture has truly managed to bypass. Pipelines built to circumvent it exist, but they are insufficient. Alternative routes exist, but they are longer and more expensive. The problem remains. Bottlenecks are not anomalies to be corrected: they are the natural form complexity takes when it organizes itself. Ports, customs offices, border hospitals, railway junctions, bureaucracy, cold chains for vaccines: every system has its Hormuz. And every time someone promises to eliminate it through technology, history responds with a shrug. Complexity does not dissolve: it moves.

The debate around artificial intelligence is crossed by a current of radical optimism: the idea that AI systems might untie the irreducible knots of reality. Optimize supply chains until chokepoints become irrelevant. Predict crises before they erupt. Transform chaotic complexity into an orderly flow. This promise is not entirely unfounded: AI-based logistics models have already reduced enormous waste, forecasting systems have made warehouses more efficient, neural networks applied to maritime traffic have improved routing. Real progress. But there is a dangerous logical leap between “improves” and “eliminates.” And it is precisely this leap that Hormuz, in its stubborn physicality, invites us not to make.

This promise is not entirely without foundation. AI-based logistics optimization models have already reduced enormous waste in distribution chains. Demand forecasting systems have made warehouses more efficient. Neural networks applied to the management of maritime traffic have improved route planning. These are real advances, not negligible ones. Yet there is a dangerous logical leap between “improves” and “eliminates.” And it is exactly this leap that Hormuz, in its stubborn physicality, invites us not to take.

The Strait of Hormuz is not a computational problem. It is a physical, historical, human problem. And even an architectural one. Its critical nature derives from centuries of rivalry between civilizations, from borders drawn and redrawn, from national identities that have incorporated mistrust as a defining trait. The Iran that controls the northern coast of the strait is not a parameter to be optimized: it is a state with its millennia-long history, its narratives of humiliation and redemption, its internal contradictions between regime and civil society. No algorithm can “solve” Iran. No recommendation system can suggest the optimal geopolitical move that would transform Tehran into a cooperative actor.

There is an epistemic error that technological enthusiasm tends to make: confusing map with territory. AI models are maps - representations of a reality that is always richer and more contradictory than any depiction of it. Hormuz as a map is reducible to GPS coordinates and energy flow statistics. Hormuz as territory is where the economic survival of whole nations, nuclear ambitions, a people's dreams of autonomy are intertwined. Complexity is not a flaw in the world that technology can correct: it is an emergent property of systems in which many agents with different goals and historical memory interact. Each time one level is resolved, agents adapt and generate another. Financial markets are the most vivid example: each new quantitative instrument that promises to tame volatility creates new forms of systemic risk. The bottleneck shifts, but it remains.

We live in an age that confuses speed with depth, computational power with wisdom. Hormuz reminds us that the world is also slow, also physical, also opaque. That the heaviness of reality is not a bug to be fixed: it is the very condition within which human life unfolds. The bottleneck is a form of the world, not a disease. And intelligence -- artificial or human -- becomes true wisdom when it can recognize its own boundaries and learn to stay within the forms of the real, even the most uncomfortable ones.