The Ungovernability Theorem
For 3,500 years, Leto II ruled humanity with absolute power and perfect foresight. He was a prescient god-emperor fused with sandworms, immortal, invincible, and utterly committed to a single goal: making his own rule so oppressive that humanity would never tolerate such rule again. His tyranny was not a failure of governance but its logical terminus -- the demonstration that even perfect control must be destroyed to preserve the species.
Frank Herbert published God Emperor of Dune in 1981. Twenty-six years earlier, in a short essay for Fortune magazine, John von Neumann -- architect of game theory, co-designer of the atomic bomb, pioneer of computing -- had concluded something remarkably similar. Asked to assess whether humanity could survive its own technological acceleration, he offered no solutions. Instead, he wrote what amounts to a proof of impossibility: no stable institutional arrangement can remain adequate under conditions of accelerating technological change.
A mathematician and a novelist, working from opposite directions, arrived at the same destination. Their convergence deserves more attention than it has received.
The Mathematician’s Refusal
Von Neumann’s 1955 essay “Can We Survive Technology?” is notable less for what it says than for what it refuses to say. Here was a man who had helped design nuclear weapons, who had formalized strategic rationality, who understood control systems and computational limits as well as anyone alive. If anyone could have proposed a cybernetic world government, an automated stabilization regime, a formal framework for managing existential risk -- it was him.
He proposed nothing.
The essay’s argument is structural. Technology accelerates; institutions lag. Effects propagate globally; decision-making remains local. Reaction times shrink; coordination requirements grow. The problem is not any particular invention -- not nuclear weapons, not automation, not climate manipulation -- but the relationship between technological tempo and institutional tempo. That relationship is diverging, and nothing in principle can make it converge again.
Von Neumann understood that specific governance solutions are regime-dependent. A regulatory framework adequate to one technological environment becomes inadequate -- or actively harmful -- in the next. The search for permanent solutions is not merely difficult but structurally misconceived. What worked for the agricultural age failed for the industrial age. What works for the industrial age will fail for whatever comes next. The failure is not contingent but inherent.
This is why his conclusion -- “patience, flexibility, intelligence” -- sounds like hand-waving. Read correctly, it is not. He is saying: there exists no closed-form solution. Only adaptive judgment offers temporary stability, and even that is a holding action, not a resolution.
The Novelist’s Demonstration
Herbert approached the same question from the opposite side: not technology but power, not acceleration but prediction.
In the Dune universe, prescience is both the ultimate tool and the ultimate trap. Those who can see the future can control it -- but control narrows possibility space. Governance optimizes; optimization creates rigidity; rigidity becomes fragility. The Bene Gesserit, with their millennia-spanning breeding program, become prisoners of their own plan. The Spacing Guild, with their monopoly on interstellar travel, become incapable of adaptation. The Empire, with its carefully balanced institutions, becomes incapable of responding to genuine novelty.
Herbert’s famous dictum -- “people are ungovernable” -- is routinely misread as libertarian sentiment or cynical observation. It is neither. It is a claim about dynamics. Governance systems, when successful, suppress the variation that would allow them to adapt. The more stable they become, the more catastrophic their eventual failure.
Leto II’s Golden Path is Herbert’s most radical thought experiment. The God Emperor understands this dynamic perfectly -- prescience gives him complete knowledge of how civilizations calcify and die. His solution is to become so terrible a tyrant that humanity will never again accept such concentration of power. He deliberately engineers his own assassination, knowing that his death will scatter humanity across the universe in what he calls “the Dispersion” -- an enforced unpredictability that no future prescient ruler can ever fully gather back.
This is not a governance solution. It is the recognition that governance solutions are themselves the danger.
The Convergence
Strip away the mathematics and the mythology, and the underlying structure is identical:
Von Neumann: Technology accelerates faster than institutions can adapt. Therefore, no institutional arrangement remains valid indefinitely.
Herbert: Successful governance suppresses the variation necessary for adaptation. Therefore, no governance arrangement remains viable indefinitely.
These are not the same argument, but they close the same escape routes. If technology held still, Herbert’s logic would still apply -- power concentrates, systems rigidify, collapse follows. If humans were perfectly rational, von Neumann’s logic would still apply -- institutional tempo cannot match technological tempo. Each thinker blocks the exit the other leaves open.
Together, they establish something close to a theorem: permanent solutions to governance are not merely difficult to achieve but impossible in principle.
Why This Remains Unacceptable
If von Neumann and Herbert are correct, most political and institutional thinking is founded on a category error. Constitutions are not meant to be eternal. Regulatory frameworks are not meant to achieve final adequacy. International orders are not meant to produce permanent stability. The search for such finality is not noble but dangerous -- a misunderstanding of what governance actually is.
This conclusion offends nearly everyone.
It offends conservatives who believe in restoring proven orders. It offends progressives who believe in designing better ones. It offends technocrats who believe that sufficiently sophisticated systems can manage any problem. It offends radicals who believe that the right revolution will finally get it right. All of them share the same underlying assumption: that governance is a problem to be solved rather than a tension to be managed.
The discomfort is understandable. Accepting non-finalizability means abandoning the project that has animated political thought since Plato: the search for the ideal state. It means accepting that every institutional achievement is temporary, every constitution is provisional, every solved problem will eventually become an unsolved one again. It means designing systems not to last but to fail -- and more precisely, to fail in ways that enable their successors.
The Deepest Error
Perhaps the most insidious aspect of finalizing ideologies is that they tend to regard the absence of a final solution as a problem to be overcome rather than a condition to be respected.
Von Neumann and Herbert, from their different vantages, suggest something more unsettling: the search for final solutions is itself the primary failure mode of governance. Systems that believe they have solved the problem of order stop adapting. They optimize for the world they know rather than the world that’s coming. They treat variation as noise rather than as the raw material of resilience.
This applies with particular force to contemporary efforts at AI governance -- where the impulse to design permanent frameworks, locked-in values, and final solutions is overwhelming precisely because the stakes are perceived as existential. If von Neumann and Herbert are right, this impulse is exactly wrong. The correct response to existential stakes is not more rigid control but more adaptive capacity -- structures that expect to be revised, frameworks that build in their own obsolescence, authorities that are designed to be challenged.
Leto II understood this. He made himself into a lesson: this is what final solutions look like, and this is why they must die.
Ungovernability as Safety
Read this way, “people are ungovernable” transforms from a pessimistic observation into something like a design principle.
Systems that assume permanence collapse catastrophically when their assumptions break. Systems that assume impermanence can fail gracefully, handing off to successors before the damage becomes terminal. The difference is not between order and chaos but between brittle order and resilient turbulence.
Von Neumann, the mathematician, proved that technological acceleration makes permanence impossible.
Herbert, the novelist, demonstrated that human nature makes permanence lethal.
Neither offered comfort. What they offered -- to those willing to accept it -- was clarity.
The most dangerous belief a civilization can hold is that it has finally gotten governance right. Not because the belief is arrogant, though it often is. But because the belief is false -- and acting on false beliefs at civilizational scale has civilizational consequences.
There is no final order. There is only the endless work of building orders that know they are temporary, and the wisdom to let them go before they become tombs.
Only failure is eternal. Design for it.