Designing the ministry of truth
"Efficient truth production requires the rigorous separation of adversaries."
I have never liked big social-media companies. But if you were running one and you decided to insource truth, here are some things you might want to think about. But first, a quick primer in Aristotelian political science.
Political science in public and private
There are three forms of government: monarchy, oligarchy, and democracy. These are kinds of force. In every regime, all three exist. They are either dominant or compliant.
When we say that a regime “is” a monarchy, what we mean is that in that regime, the force of monarchy is dominant; the forces of oligarchy and democracy are compliant. We can humanize these forces by defining them as wills.
A regime that is not a monarchy is not a regime in which the will of monarchy does not exist. It is a regime in which there is some structural constraint which controls the will of monarchy—either preventing it from existing, or compelling it to remain ceremonial subordinate. This structure is a compliance device.
These forms of power also exist in the private sector. The monarchical will is the will of the CEO, the oligarchical will is the will of the employees, and the democratic will is the will of the customers. An effective business meshes all three wills harmoniously.
In public or in private, three forms of action characterize monarchical, oligarchical and democratic action. Democratic action is by consensus: everyone hears everyone agree. Oligarchical action is by process: legitimate actors follow authorized steps. Monarchical action is by command: subordinates are ordered to complete a mission.
In both the private and public sectors, the histories of both business and war indicate that organized human action is most effective when organized by command. Yet there are significant structural exceptions. There is a place for both process and consensus. (Indeed, hierarchical command is only a special case of process.)
Process is never absent from operations. In the public sector, no army bigger than a bandit gang is without some form of military law. In the private sector, process tends to develop around accounting, personnel, and quality assurance—and some essential process is offloaded to the public sector in the form of regulatory administration.
Consensus creates the ideal level of collective commitment, but is slow, hard to scale, and ill-adapted to faction and dissent. Many processes use consensus at small scale. Large-scale forms of consensus, like elections, are rare in the private sector.
Three ministries of truth
A great social-media company is a virtual regime. Like every virtual or physical regime, it has to know what is true. It therefore needs a source of truth. Either it decides to build its own truth engine, or “ministry of truth”—or it decides not to.
A regime with no ministry of truth has not chosen freedom—it has chosen subjection. It has not created a vacuum of power; it has abdicated its monopoly of power to some other power, internal or external. We all remember when we lived in a world where we thought there were other powers worth trusting.
Since truth is power, and there are three forms of power, there are three ways to build a ministry of truth: oligarchic, democratic, and monarchical.
The oligarchic ministry of truth
An oligarchic ministry of truth is just an interface to the powers that be. Facebook’s “Oversight Board” is an excellent example. The fault line in the Oversight Board is over whether “racism” against unprotected classes is actually racism. (It isn’t. It is totally normal and fine.) I quote from a recent holy decree of these solons of ethics:
The Board emphasizes the importance of context in assessing whether content is urging violence. In this case, the Board finds that the quotes from the poem “Kill him!” are an artistic and cultural reference employed as a rhetorical device. When read in the context of the whole post, the Board finds that the quotes are being used to describe, rather than encourage, a state of mind. They warn of cycles of violence and the potential for history to repeat itself in Ukraine.
Indeed. It is not far from “in Ukraine” to “in Minecraft.” Hopefully, dear reader, you did not get that joke.
Obviously, the Meta management has chosen safely. This is the normal, compliant way to do it. Oligarchic will seeps in from every external pore. It is not enough to ignore this force. If oligarchy is not controlled, it will dominate. Maybe it should!
Concretely: the staff is the oligarchic force within a corporation. Left to set policy on its own, this local oligarchy will merge with the general oligarchy. It will become an obedient arm of “civil society.”
The fuel of oligarchy is prestige. In a democracy, everyone is equal. In an oligarchy, no one is equal. Access to an oligarchic process depends on some formal credential or informal reputation.
Even the legitimacy of formal, credentialed prestige rests ultimately on informal, reputational prestige—like an MD from Harvard. Harvard does not have a piece of paper saying it is Harvard. Nor does it need one. Its 400th birthday is coming up.
If a new ministry of truth assigns an infallible respect to the existing prestige system, it is and will always be a satellite—a slave. Only if it can challenge this system—not with the petulant negation of a child, but as a mature, respectful competitor, aiming to play the game of the old regime yet play it better—can it succeed in any real way.
The democratic ministry of truth
The instant and automatic temptation is to outsource truth to pure digital democracy—to simply crowdsource truth. This will always fail.
Crowdsourcing truth should not be mistaken for insourcing truth. It is closer. Digital democracy is one essential part of the solution. It cannot be the whole solution.
Worse—it may seem to work at first, then mostly work, while failing in subtle and ugly ways that resolve into the same old oligarchy. Here is why.
Democratic force cannot be a disorganized mob—it demands structure. The marketplace of ideas needs a roof and stalls. These are supposed to be neutral structures, which should not affect the content of the ideas… but, they always do. At least, in any context in which democracy has power.
The paradox of democracy is that democracy corrupts itself. Once the marketplace of ideas is granted power, it is corrupted by power. The marketplace’s reward landscape changes—it now rewards ideas which give most people a rewarding feeling of power. (This is one functional definition of leftism.) The open society develops a leftist bias.
Once Popper’s open society is granted power, it is no longer open. Out of power, it is pure. It seems to deserve power. In power, it decays. It seems to not deserve power. This decay is a progressive condition.
The open society is corrupted not because it is punished, as per Popper’s villains, but because it is bribed—as per Popper’s acolytes. Power with a carrot is the same thing as power with a stick: power. It still corrupts. Democracy still decays into oligarchy.
Even the enemies of the oligarchy are corrupted by it. The classic Internet rightist is rewarded by outrageous ideas—dropping communists out of helicopters, and so on. Whether reveling in such visions helps defeat communism is not, it seems, the issue.
Thus an uncensored forum is not just rightist—but crudely, ineffectively rightist. The vulgar /pol/tard seems controlled by a cabal of his enemies. Indeed his frame, while gloriously democratic, is often simply an inversion of the oligarchic frame. There is no decentralized answer—not even in the complete absence of censorship.
We are trained to fear centralization. This training is a political formula: an idea that trains us to support our government. This formula is one of the many compliance devices that restricts the power of democracy to escape from oligarchy to monarchy.
While corrupt or insane centralized powers are dangerous, any decentralized insanity is far more dangerous—just as metastasized cancer is more dangerous than localized cancer. If you think that decentralized systems are inherently incorruptible, I would invite you to look again—you are looking at a security hole in your rational mind. Do you have a good reason to assume that nothing bad has come in through that hole?
All democratic force has a leftist bias which terminates in a stable leftist oligarchy. For example, the greatest achievement of democratic force on the Web is Wikipedia. Since its founding by an Objectivist, Wikipedia has become steadily more progressive.
There is a simple reason for this—Wikipedia has become steadily more important. It is impossible to create anything important which does not either control the force of democracy, or succumb to it. And since we exist within an oligarchic regime, the force of democracy is controlled by the will of the oligarchy.
Yet Wikipedia is still one of the treasures of the early digital era. There is hardly anything useful left on the Web besides Wikipedia and Substack/Medium. And in any field of ideas not corruptible by power—math, for example—Wikipedia is superb.
A good example of a young digital democracy today is Twitter’s Birdwatch system, which depends on bridge-based ranking. The ominous power of any mechanical ranking system is well expressed by the author’s introduction—my italics:
This report explores the potential of bridging and discusses some of the most common objections, addressing questions around legitimacy and practicality. It contrasts bridging with some of the most discussed approaches for reforming ranking: reverse-chronological feeds, ‘middleware’, and ‘choose your own ranking system’. (Unfortunately, without introducing bridging, all of these proposed reforms still reward those who seek to divide.)
A ranking must not reward… those who seek to divide. Which would be who, exactly? This sounds like an Orwellianism for some sinister adversary. It might well mean: democracy must not reward conservatives. Or: democracy must not reward Tamils. Deeper in the paper, we learn the nature of this fear:
Empowering each individual to tailor their algorithms might encourage a further splitting of the American polity, allowing groups to more easily find voices that echo their own views, sources that confirm their factual beliefs, and political leaders that amplify their own fears.
Power always says the same thing about speech. It says: you have to read our tweets. But not only do we not have to read yours—you can’t even read yours.
It is always fairly easy to disguise this essential structure as a concern for something wholesome—bringing us together, protecting the children, preventing violence, etc. We should come together, protect the children, and prevent violence. However, when we observe any rule whose real effect is “you have to read us, but you can’t read you,” we should not be faulted for indulging in a little Foucauldian deconstruction.
A system that acts only when both democratic factions agree, like Birdwatch today, does not have the effect of bringing us all together. It has a different laudable effect: by allowing each side to veto the other, it allows only uncontroversial speech.
This is a real engineering victory and a good thing. But it is still the lowest rank of achievement. Moreover, it does not feel utterly ungameable. All democratic systems work the best when they are the least important; as they grow important, they are ideologically self-hacked, like Wikipedia, by the change in the reward landscape.
As Schmitt explained, democracy is not a neutral force. It privileges leftism, just as monarchy privileges rightism. This was true of the early 20th-century parliaments Schmitt was critiquing; it is true of Wikipedia committees and edit wars. The long arrow of drift from constitutional monarchy to Cultural Marxism is perfectly clear. Democratic energy flows left—impeded only by the burden of its various illusions.
Structures with a leftist bias cannot police themselves even in their own self-interest, which is what makes them unstable and dangerous—like the scorpion riding the frog. As John Adams said, true democracy always commits suicide.
At least in the current technical moment, government still physically depends on the consent of the governed, and democratic force still remains the ultimate power. But democracy is the pure explosion of politics. It can only be effective as a transient force—designed to shift the regime from one stable state to another. An explosion is not a good thing in and of itself; it is only useful within some well-designed engine.
In any stable state, democracy is disabled by some kind of compliance device. Even periodic elections are a compliance device. Unlike in ancient Athens, which could execute Socrates any day of the week, the hot cocaine of power is fed to the masses only once every four years.
A truth engine designed to faithfully serve its master, be it a company or a nation, by the faithful production of reliable truth, needs reliable external guidance. The engine is not crowdsourced because of the inherent reliability of crowds. It is crowdsourced because of the inherent scalability of crowds.
But crowds cannot guide themselves. If they are not guided by a local monarchy, they will be guided by the global oligarchic ethos. George Soros’s thumb might as well be pulling them around the map with an app.
The monarchical ministry of truth
Eldon Gorst, one of the last imperial viceroys of British Egypt, used to say that he could rule Egypt indefinitely, in perfect peace and security, with one power alone: the power to hang one Egyptian, of his own absolute and unquestioned choice, every year.
Banning one user, even for life, is not the same as hanging one man. But the principle that the CEO has the personal right to inflict the ultimate penalty, for any reason or no reason, is essential to the logic of monarchical governance. Once this right is lost, all other rights will be lost in it—and power will find its usual sad way to oligarchy.
The goal of the ministry of truth is truth. Just as the goal of the king’s army is the king’s peace, the goal of the king’s ministry is the king’s truth. The purpose of the ministry is not to override the king but to serve him. The king needs to make sure he is always right about everything all the time. His ministers help him do this job.
The king always needs ministers. The ministers always need clerks. This pyramid of people, operated by command with maximum flexibility and minimum process, exists only to extend the will of the king. They are all soldiers of the king.
But the king’s soldiers are not and can never be enough. The king’s strength must be the strength of the nation—which is a free country, not a military despotism, Athens not Sparta. The king cannot command the people. The king must lead the people.
The right design for a ministry of truth is a mixture of monarchy and democracy. From monarchy, it takes its efficiency and precision. From democracy, it takes its scalability and redundancy. The mass of democracy, guided by the aim of monarchy, makes the strongest possible structure.
A ministry of truth does not even spurn the immense talents and labors of the existing oligarchical regime. The “reliable sources” are indeed mostly reliable. The muddy stream of official knowledge cannot be ignored; it can only be filtered. This stream is so massive that even filtering it is an unprecedented challenge. Yet again, to ignore it is to submit to it. The filter is the monarchy’s compliance device for oligarchy.
And the democratic part of a ministry of truth is a compliance device for democracy. So its its accuracy—when there is no systematic error for the muttering of crowds to expose, when every “conspiracy theory” is mere nonsense, peer-to-peer power is an engine without its best fuel. When any sincere corrections are fed into the king’s truth machine, they are not available to power any revolutions.
The integrated oracle
There are many possible structures for a truth engine which can unite the forces of monarchy, oligarchy and democracy. Here is one: an integrated oracle.
This structure is integrated because it integrates the three forces. It is an oracle because it works by producing a comprehensive answer to a simple question.
Birdwatch today is like a little baby version of something like an integrated oracle. Wisely, Birdwatch has given itself a much easier job, because it is designed to be wide rather than deep—to rapidly surface undisputed information at scale, not to deeply analyze disputed questions.
To go from Birdwatch, a humble and beautiful success for all the “community notes” I have seen, to “a Twitter that can think for itself”—and not just rival, but vastly exceed, Harvard and the NYT in the quality of that thought—is an enormous step. It is like the difference between going into space, and going into orbit. Or maybe to the moon.
An integrated oracle is an institution which can give a fully researched answer to any disputed question. The reliability and completeness of its answers will exceed the reliability and completeness of journalism, and rival or exceed the reliability and completeness of a judicial or administrative court. It will be ready to criticize or validate the hardest fields of science—scientists will use it to settle their disputes.
The oracle does not worry at all about its reputation. It is always worried about If the oracle is always right, its reputation will follow. Its critics will mutter and grow silent. Its voice will grow only clearer and more confident.
The oracle is a purely monarchical institution. Its judges are corporate employees. They serve at the CEO’s pleasure. The CEO can even overrule their decisions. Again, the oracle itself exists only to check and expand the judgment of the CEO.
But the monarchical judgment of the CEO is incomplete and impotent without its oligarchic and democratic elements. All three powers must be fully integrated.
Of course, “democracy” is always a euphemism. No open decision structure can weight all voices equally. Any kind of open democracy, if it works, only works by selecting the natural aristocracy that always lies hidden within the crowd. Digital democracy must select, recruit, and muster this rival oligarchy as an invisible elite.