Most systems that look irrational from the outside are not irrational. They are optimized, just not for the outcome you were told to expect. War persists in a world with near-trillionaires. Hunger persists in a world that produces enough food for everyone alive. Pension systems promise security to people who understand the math shows otherwise. The question that actually explains these is not “why hasn’t this been fixed.” It is “who is paid while it stays broken.”
Start with the mechanism, because it is simpler than it sounds and does not require a conspiracy. Any institution funded to solve a problem develops a structural interest in that problem never fully going away, because solving it ends the funding. A department that eliminates its own caseload eliminates its own budget next year. A charity that ends the crisis it was built around has to find a new crisis or shut down. Nobody has to plan this. Ordinary self-interest, keeping the job, protecting the budget, growing the department, does most of the work on its own. Robert Michels named a version of this the iron law of oligarchy in 1911, studying European socialist parties that were founded to abolish hierarchy and ended up building some of the most entrenched hierarchies in politics. Public choice economists James Buchanan and Gordon Tullock later formalized the same pattern across government agencies: officials maximize budget and survival like anyone else maximizes income, not because they are corrupt, but because the incentive structure rewards it.
War financing is the starkest case, because the money trail is unusually traceable. Wars are financed largely through debt, and debt is issued and often held by banks and bondholders who earn interest on it regardless of the war’s outcome or duration. The U.S. federal debt tied to post-9/11 wars in Iraq and Afghanistan is estimated by Brown University’s Costs of War project at over 2 trillion dollars in interest payments alone through 2050, money that flows to bondholders whether the war achieves its stated goal or not. This is not proof that banks start wars. It is proof that the parties financing a war have no structural incentive to want it to end quickly, which is a different and more durable problem than any single villain.
Hunger works the same way, through a different channel. The world produces more than enough calories to feed everyone alive, roughly 1.5 times global need according to the UN Food and Agriculture Organization, yet hundreds of millions remain undernourished. The gap is not production. It is distribution, price, and the fact that global food aid, subsidy, and trade systems each have institutional constituencies, agribusiness lobbies, aid bureaucracies, tariff-protected domestic producers, who benefit from the current arrangement staying roughly as it is. Amartya Sen’s work on famine showed decades ago that most modern famines occur without an actual shortfall in food supply; they occur because of who has the entitlement to buy it. A problem framed as scarcity is often a problem of who profits from the current allocation.
Pension systems are the version closest to home for anyone under forty. Pay-as-you-go pension systems work by using today’s workers’ contributions to pay today’s retirees, on the assumption that tomorrow’s workers will do the same. That assumption depended on a demographic pyramid, many workers per retiree, that has been inverting for decades across most developed economies; the OECD projects the ratio of workers to retirees in many member countries will fall by roughly half between 2000 and 2050. Governments that raised this alarm early would have had to raise taxes, cut promises, or extend working years, all politically costly in the short term for the people in office at the time. Deferring the fix is not incompetence. It is the rational choice for anyone whose horizon is the next election, at the direct expense of anyone whose horizon is their own retirement.
Here is the distinction that keeps this from collapsing into cynicism: structural self-interest and deliberate villainy are not competing explanations, they run side by side. A bank does not need to want war for war financing to be profitable to hold, and a pharmaceutical sales team does not need a systemic theory to keep pushing volume once the numbers reward it. Purdue Pharma’s own internal sales data showed executives tracking which doctors were prescribing in patterns consistent with abuse, and continuing to target them anyway. That is a deliberate choice sitting inside a system that would have rewarded the deliberate choice even if no single executive had made it. Flattening this into “no one is to blame, it’s just the system” is as wrong as pretending every bad outcome traces to one villain’s decision. Both mechanisms are real, and they compound each other.
The reason this framing matters practically, not just intellectually, is that it changes what you look for when something stays broken despite obvious effort to fix it. Stop asking whether the people involved are stupid or evil. Ask who gets paid, funded, reelected, or promoted precisely because the problem persists. That single question routes around most of the noise. It will not always find a villain. It will almost always find an incentive. And an incentive, unlike a villain, is something a system can be redesigned around, once enough people can see it clearly enough to demand the redesign.
I write about AI, systems, markets, and the incentives hiding underneath them. Get the next essay by email.
Unsubscribe any time. Privacy.