Twitter should be in the truth business

"Discourse is not truth. Discourse is the raw material of truth."

Twitter is discourse. Discourse is not truth. Discourse is the raw material of truth. Discourse is to truth as coca leaf is to cocaine. Sure it can make a Peruvian peasant work all day in the sun at 10,000 feet. But no one ever did a line of minced coca leaf. You gotta refine that shit. Yet without the leaf—there is no powder.

Discourse at a certain scale is power. For instance, one form of power is nut power. When an idea gets enough reach, nuts will act on it. They will break into Nancy Pelosi’s house with a hammer, or snipe Republican congressmen at a softball game.

In most immediate ways, even when one voice can reach a million people in seconds, the discourse is much more powerless than anyone thinks—because of the enormous secular drop in the capacity for political violence. Modernites are limp as a month-old carrot and a million of them can do nothing at all. But every nut is one in a million.

Because discourse is power, uncontrolled discourse naturally attracts regime power. Even if discourse did not create such nutty externalities, it would threaten the strong. All government inherently rests on consent. Discourse inherently threatens consent.

Ergo, any organized structure of discourse—such as Twitter—needs a strategy for handling power, because discourse is belief and belief is power. While discourse is not truth, it is the raw material of truth. Twitter, as a company, manages what is now an essential artery in the whole discourse of man. Truth is power and power is truth.

The new Twitter management has three ways to work with truth: to ignore truth; to outsource truth; and to insource truth. The normal strategy is the second. The obvious alternative strategy is the first. The most visionary strategy is the third.

Outsourcing truth

The perfect working definition of outsourcing truth comes from Wikipedia, for which outsourcing truth—relying on “reliable” sources whose word is considered infallible—was clearly the optimal strategy at the inception of that institution. Wikipedia, though founded by an Objectivist, is now clearly an orthodox mainstream institution.

If Wikipedia lived in a Catholic world, it would let the Pope define a reliable source. It might look like the Catholic Encyclopedia. If Wikipedia lived in the Soviet Union, it would let the Communist Party define a reliable source. It might look like the Great Soviet Encyclopedia. But Wikipedia lives in… the global postwar liberal order. So…

If you read the Wikipedia definition of a reliable source, it tells you: a reliable source is a source that is reliable. Or rather, reputable. Any inquiry into this leads to the definition of fringe, which is defined as—not mainstream. Going deeper into the definition of mainstream leads to a disambiguation page topped by a 2020 movie.

There is literally no information on the Internet about what this “mainstream” is or how it came about. Suppose it just meant: controlled by the Mafia. We could all have been reading a Mafia-approved press, with Mafia-approved books, and never know it. And even Wikipedia would say nothing about the Mafia. (It might have something to say about the Italian-American Civil Rights League.)

But we do know one thing about this “mainstream.” We know it is passionate about social justice and creating change. What… what do those words mean, anyway? As Dr. King put it, the arc of history is long, but it bends toward social justice:

Isn’t that interesting? But what does it mean? As Buckaroo Banzai once put it: don’t pull on that. You don’t know what it might be connected to. (No, it’s not “the Jews.”) You could try to untangle what the heck is going on here. Trust me: that would take you a while. First, set your time machine to Mugwump time, 1895 or so…

Or you could say: whatever this mainstream is, I don’t want it. And whatever it is—maybe the rest of the world doesn’t want it. By definition—that makes it a bad product. It should be possible for a good product to compete with a bad product.

Moreover, once we distrust this mainstream, the concept of political balance has no relevance at all. “Centrism”—the idea that the truth can be computed by a democratic average, like the averaged Airbus joysticks on Air France 447—is completely logical, as long as we assume private truth is evenly distributed around a mainstream. Which may not be magically infallible, but is magically at the center of the distribution… sure.

Present historical reality does not have to be anywhere near close to either the official truth, or the average of all private truths. It does not have to be close to the average trending truth on Twitter. All we can say is it probably exists somewhere on Twitter.

But any algorithm for outsourcing the truth will connect itself to some mysterious historical force that seems to have started operating no later than 1895—the oldest case of the sinister jargon I could locate. Again, you could look into this—or not.

Not all moderation is about “misinformation.” But even the identification of concepts like “hate speech” or “offense” is fundamentally political. Since culture is downstream from politics, all cultural concepts are actually political concepts. What is the bias of the Great Soviet Encyclopedia—a political bias, or a cultural bias?

Ignoring truth

Everyone’s first strategy for handling power is to ignore it. This works, up to a point. Until then it is possible to be the “free speech wing of the free speech party.” After then it is not.

People always have copes to think they can go back to their childhood. Time is not a flat circle. Time is an arrow. Except it speeds up. Since nature abhors a vacuum, there is a limit to how big a vacuum of power can become. Whether you’re building a rocket or a car, you don’t want nature to start abhorring you. It’s the same in politics. Trotsky put it best: you may not be interested in politics, but politics is interested in you.

If Twitter tries to moderate without either outsourcing or insourcing truth, there are three possible results. First, power will take its head—creating a “$44 billion Gab.” Second, it will fall under the power of the mainstream anyway. Third, its moderation will be weird, arbitrary and capricious.

Only with a single source of truth can a moderation team define harmful or offensive content. The decision to ignore truth is an inefficient decision to insource truth. The official truth will seep in everywhere. Twitter might as well worship at the throne of the “mainstream,” whatever that is—it would just be getting there the long, hard way.

Partly outsourcing truth is simply not an option. The temptation to submit to power is omnipresent and universally seductive. It even feels like the responsible adult idea. Compromising, even a little, with an existentially determined power, is like scratching a mosquito bite. It feels like it will work and it does. The bite feels better right away.

Then in about a minute it itches even more, and you want to scratch again… you end up with a horrible, scarry sore the size of a half-dollar which is healing, bleeding, and itching at the same time. You start to consider a new career as a professional leper.

Power doesn’t take yes for an answer. Whatever you give it, it always wants more. Submitting it is feeding it. However you scratch power, power will keep itching.

Twitter could not be tame enough for power. Facebook could not be tame enough for power. The cathedral of the press will keep hitting these intractable structures, no matter how much they submit, until they do not just submit—but become the most enthusiastic standard-bearers of the cause we all share. Of change and social justice. Whatever that means.

The engineers who originally built most of these social platforms were typical 1990s lite-libertarians. They would be horrified at the union of power between centralized social networking and the ideological monoculture of the modern mainstream media. Their initial reflex was to resist power—an even worse idea.

Resisting power attracts power and makes it stronger. Generally, power doubles down when you resist. When the house can double down against you, the house always wins. Once it wins, it takes everything you have—a huge meal.

Insourcing truth

Insourcing truth means building the structural capacity for Twitter itself to make a call on what is right or wrong. This is an incredibly difficult problem on which the fate of the world may literally depend, so it should not be taken lightly.

Imagine a sort of Twitter court that had the intellectual firepower and confidence to decide any matter of right and wrong—much as a real-world court can. Was Covid a lab leak? Was Shakespeare Oxford? Did FDR have prior knowledge of Pearl Harbor? Could UFOs be real? Does amyloid cause Alzheimer’s? Is Hunter’s laptop authentic? And so on.

Forget for a moment how this truth engine works. Imagine only that it does work. Imagine that the only way Twitter uses it is to mark tweets as misinformation—with a link to some report from the truth engine. Twitter court, in this world, often can and freely does mark articles in Nature or the New York Times as untrue or misleading.

If and only if this device is in fact infallible, its prestige will naturally come to exceed that of the New York Times. Since the power of the Times is derived only from its prestige, the power of this device will come to exceed that of the Times. Stand back.

Structure of the truth court

The only way to prevent oligarchical biases, internal or external, from contaminating a judicial system, is to roll the judiciary back to the original constitutional design in which the judicial power is an extension of the monarchical executive. The supreme court is simply the king—who can overrule or overturn all judgments. Totally based.

In other words: the decision as to what is true and false, right or wrong, on Twitter, falls ultimately to the Chief Twit. Truth is an extension of his or her will. The reason the Chief does not make every decision personally is solely a question of scalability.

The only way to justify this power is for the Chief to be absolutely always right. If the truth of Twitter is the truth of the Chief, and the primary value of Twitter is not the minced leaf of entertaining raw discourse, but the refined powder of truth, any error in this truth is as fatal as a rocket crash. There is no room for winging it.

To be of significantly higher quality than whatever bathtub gravel the Times is selling, this new truth must be refined and developed by sophisticated engineering processes. The processes of this “court” will not have been invented in 13th-century Kent. They will be trusted because they work, not because they are venerable.

This truth engine will be powered by Twitter’s user base. It cannot be a poll. The goal of the truth engine is to refine the truest truth, utterly regardless of popularity—and certainly regardless of political bias. Since it can never afford to fail, it has to start with the highest quality—its reports have to be always incredibly right.

A truth engine should boot up slowly. At first, a Twitter court is doing well to try one case a month—maybe even one a quarter. Make it work first. Everything that works can be scaled; nothing that does not work should be. Eventually the truth engine will be turning out tractor-trailers full of kilo bags of uncut refined white powder truth.

Structure of an online trial

The fundamental structure of an online trial is the same as the structure of a regular trial. Lawyers for both sides work together to produce an argument. This argument is presented to a judge or jury, who renders a decision. Online, the details are different but the structure is the same.

Assume you are some kind of judge or juror. What is an online argument in its most helpful form? Is it a theater performance, like a classic courtroom? Cross-examining witnesses is certainly a good thing, but not needed for most Twitter-type problems.

The simplest possible argument is a pair of legal briefs. A pair of briefs and rebuttals is a slightly more advanced form. But these forms are built around the limitations of paper government and the structure of the adversarial process.

What would help the neutral judge the most is an argument tree, which makes it easy to look through every argument from either side, see the other side’s rebuttals to each of those arguments, the rebuttals to the rebuttals, and so on—until one side is tired of repeating itself, and refers back to its last point. This kind of collaborative document was unthinkable to the scribes of 13-century Kent, or whatever, but is trivial today.

Even a structured argument cannot be composed by a crowd. The tree has to accept public contributions, but those contributions should be curated, and perhaps even just composed, by a team of advocates. These advocates can only be qualified by the combination of current Twitter clout, demonstrated loyalty to their own team, and obvious intellectual talent.

As for the judges, they are… brilliant, open-minded generalists. They work for Twitter and report to the Chief Twit. Outsourcing arguments is possible and necessary; outsourcing judgment is impossible. Executive review of all judgments is automatic; every losing side wants to appeal. The Chief can change any decision for any reason.

After the process is well-developed, a prediction market that can predict the judges’ response can be developed. A market is like an AI; it has to be trained on good data. Such a market can use crowd power to amplify and scale the judicial process. Until there is some sense of a coherent truth that the judges are converging on, it makes no sense to emulate that coherent truth with a prediction market (in which the winners will be Internet randos who think the same way as the judges).

A new truth engine should boot up under optimal conditions. Optimal conditions mean: a deep pool of high-quality advocates on both sides, for a clear and relevant question close enough to the confidence and expertise of truly undecided judges. Twitter picks the (unpaid) advocates and, again, employs the judges.

Once the easy questions are handled easily, the truth engine can scale in all kinds of directions. It can expand away from the safe ground of true and false and toward the treacherous terrain of right and wrong. It can resolve scientific disputes. It can even settle private beefs. Imagine the original Gamergate, but with a truth engine…

Truth and power

Building a truth engine is not resisting power (which is bad). Building a truth engine is competing with power (which is good).

Every new regime must win by the standards of the old regime. A military despotism can only be overthrown by force. The current regime is not a military despotism; the power to decide, which is the essence of power, belongs to intellectual prestige. This is how every theocracy has worked since the priests of Amun were running Egypt.

Creating centers of prestige which are higher than the regime’s centers of prestige—which lack venerable brands, but more than make up for that with sheer excellence—is as far as possible from being resistance as anyone could imagine. Not only is it not violent or illegal, no one even considers it immoral.

Yet, in a regime whose nervous system is built on prestige, there is no way for such a machine to not make an impact. The brands are just brands. The New York Times is not even mentioned in the Constitution. And even the power of the Constitution is—just a brand.

What if the 21st century can generate new institutions with more prestige than even the oldest, most venerable, most patriotic 18th-century brands? Then we could really have some fun.