Inadequate Equilibria
Reads why visibly bad systems can persist even when many people inside them can see the problem and would prefer something better.
Full Practice · Knowledge · Reading What's Operating
Mechanism
Some systems are bad in ways that are obvious and still hard to fix.
The school rewards credentials more than learning. The hospital knows which process wastes staff time, but no one with authority benefits from changing it. The scientific field can see that replication is under-rewarded, yet the careers inside the field still run through novelty. The organization has a better policy available, but the person who implements it first absorbs the cost while the benefit spreads elsewhere.
Inadequate Equilibria reads the bad stable state that visible intelligence alone does not correct.The common naive question is: if this is so broken, why has nobody fixed it? The tool answers by refusing the fantasy that every visible improvement has someone positioned to capture the benefit of making it. A system can contain a large improvement and still leave no actor with enough incentive, authority, information, permission, or coordination capacity to realize it.
The distinction from Nash Equilibrium matters. Nash Equilibrium gives the formal strategic concept: no actor can improve their outcome by changing strategy alone, given what everyone else is doing. Inadequate Equilibria is the applied institutional diagnostic. Not every Nash equilibrium is inadequate, and not every inadequate system is best explained by the formal model. This tool asks why a bad system, visible as bad, still persists. Who could move? What would they gain? What would they lose? Who has authority? Who captures the benefit? Who pays the transition cost? What happens to the person who tries?
Eliezer Yudkowsky frames the tool through adequacy analysis: when should you expect civilization to have already picked up a visible improvement, and when should you expect visible improvements to remain unused because the system cannot exploit them? The market analogy is useful if handled carefully. Some markets are efficient because anyone who sees a predictable profit can act and, by acting, correct the price. Other domains are not built that way. If you can see the problem but cannot profit from fixing it, cannot gain authority by fixing it, cannot coordinate the actors required to fix it, or cannot even make the failed state legible to the people who matter, the bad outcome can remain.
Control misreads Inadequate Equilibria by turning the diagnosis into elite contempt: people inside the system are fools, therefore a superior designer should override them. Decay misreads it by treating inadequacy as proof that nothing can be done. The Range reading is more disciplined. It asks where the system is inadequate, why correction is blocked, whether local wedges remain possible, and whether your own confidence is strong enough to justify action.
Practice
The diagnostic question is: "Why has this visible improvement not already been taken?"
Use this when a system looks obviously broken and the easy explanations are too simple: everyone is stupid, everyone is corrupt, or everyone is waiting for someone brilliant to arrive. The better reading asks what correction paths the system actually allows.
Name the bad equilibrium. Be specific. "Education is broken" is too broad. "Schools reward credential accumulation more reliably than durable understanding" is readable. The state has to be stable enough that people can keep participating in it while complaining about it.
Find the would-be corrector. Who could fix this? A customer, employee, regulator, founder, voter, researcher, funder, patient, executive, judge, or outside entrepreneur? What power do they actually have? What benefit would they capture if they succeeded?
Distinguish inefficiency from exploitability. A visible waste is not automatically an available opportunity. Ask whether someone can act on the insight in a way that benefits them enough to cover risk, cost, delay, reputation loss, and coordination burden.
Locate the blocking structure. Is the problem a central-command bottleneck, a two-part failure, an information problem, a prestige incentive, a legal barrier, a career risk, a coordination problem, a trust failure, or a missing institution?
Look for the local wedge. A system may be too hard to repair globally and still adjustable locally. You may not fix medicine, but you may run a better experiment in your own life. You may not fix publishing, but you may build a better review rule inside one team.
The practice is dangerous when it becomes permission to trust every contrarian impulse. Most bad ideas also look like overlooked opportunities to the person having them. Inadequacy analysis has to run with Foundation intact: base rates, evidence quality, domain humility, feedback, and update conditions. The tool does not say you know better. It tells you when the fact that "someone would have fixed it already" is weak evidence.
In the Wild
A company had a painfully slow approval process. Everyone knew the process created delay. Every manager complained about it. The obvious fix was to remove two layers of signoff. But the signoffs existed because nobody wanted to be the person blamed for a bad decision. The cost of delay was spread across teams and customers; the cost of a visible mistake landed on the approver who removed the guardrail. The system was inadequate in the plain sense: it produced worse decisions more slowly, and no individual actor gained enough from fixing it to accept the risk.
A scientific field under-rewarded replication. Researchers valued replication in public. Journals and hiring committees rewarded novelty in practice. The young researcher who chose replication paid career cost for public benefit. The senior people who admired replication did not control enough of the incentive field to make the career tradeoff safe. The problem was not that no one understood the importance of replication. The problem was that the reward path for doing it was broken.
A parent noticed that a child's school was teaching to standardized tests in a way that hollowed out curiosity. The parent could complain, but the teacher was graded on test results, the principal was judged by district metrics, and other parents feared anything that might lower scores. The visible improvement, durable understanding, had no actor positioned to capture enough benefit inside the existing accountability system. A local wedge might still exist: after-school reading, a small parent group, a teacher willing to protect one pocket of inquiry. The whole system might stay inadequate while a smaller repair remains possible.
When you see a broken system, do not rush to either deference or superiority. Ask what correction paths exist. If the system gives no one a usable way to fix what everyone can see, the bad state may be stable for reasons that have nothing to do with intelligence.
Lineage
The Codex did not invent Inadequate Equilibria. It inherits the tool from Eliezer Yudkowsky's 2017 book Inadequate Equilibria: Where and How Civilizations Get Stuck and places it inside the Knowledge as an applied institutional diagnostic.
Yudkowsky's project begins with a question about when you should expect the status quo to be hard to beat. The book generalizes efficient-market reasoning beyond finance. In a domain where many capable actors can see opportunities, act on them, profit from them, and thereby correct them, you should be cautious about assuming that obvious improvements are lying around. In a domain where correction paths are blocked, bottlenecked, unrewarded, or split across actors who cannot coordinate, visible improvements can remain unused.
The book distinguishes efficiency, inexploitability, and adequacy. A system can be inefficient in the sense that it predictably produces bad outcomes, and still be hard for any individual to exploit or repair. A startup market may overprice a company, but outsiders may have no way to short it. A central bank may follow bad policy, but no outside actor can profit by correcting the central bank. A research field may neglect an intervention, but the people who could study it may lack career incentives and the people who would fund it may lack the right evaluation frame.
The formal background includes efficient-market theory, Nash equilibrium, microeconomics, public choice, and institutional analysis. The distinctive contribution of Yudkowsky's treatment is the practical question it leaves you with: where should I defer to civilization's competence, and where should I expect civilization to be stuck?
The Codex's translation is placement. Inadequate Equilibria belongs in Reading What's Operating because it reads stuckness before it prescribes intervention. Nash Equilibrium supplies the formal strategic language. Mechanism Design may later alter the game. Rules-in-Use may reveal the operative constraints. Leverage Points may show where a change has force. But this tool asks the applied question: why does the bad state persist when so many people can see it?
The tool has limits. It can make you overconfident, especially if you enjoy feeling like the one person who sees through the system. It can also become cynicism: everything is inadequate, everyone is trapped, nothing can be trusted. Both are failures. A good inadequacy reading is local, specific, and answerable to evidence. It names the domain, the blocked correction path, the actors, the incentives, and the conditions under which the reading would change.
Cross-references
Within the category. Prisoner's Dilemma reads the minimal cooperation trap. Inadequate Equilibria reads the wider institutional condition where improvement is blocked by incentive, authority, and coordination patterns. Moloch reads the competitive race that can generate inadequate systems. Rules-in-Use asks what formal or informal constraints actually govern the stuck state. Leverage Points helps distinguish a true intervention point from a complaint about symptoms.
Across the Workshop. Checking Your Map Against Reality matters because inadequacy claims are especially vulnerable to flattering error. If you think you see a civilizational failure, the map needs pressure. Foundation tools such as Base Rate Neglect, Dunning-Kruger Effect, and Bayesian Reasoning remain active in the background.
Limitation. Inadequate Equilibria does not license lazy contrarianism. It asks whether the status quo has a correction path. Sometimes it does, and the correct conclusion is that you probably have not seen what the people inside the system have seen.