How Diego Decides What Is True

Diego’s worldview on knowledge, expertise, and changing his mind

There is a comfortable chair in this room. Sit down before reading the inscription above it:

If you need to know what's right, come to me.

A little farther down the same 2025 post, Diego clarifies:

So I'm more right than righteous if you catch my drift.[⁎]

The asterisk is mine. The confidence is entirely his.¹

Diego treats understanding reality as a craft to which someone can devote a life. Other people specialize in making money, becoming attractive, maintaining harmony, or acquiring influence. He regards being right as his own specialization, and the abrasiveness sometimes accompanying it as one of the costs he has accepted. In his account, he has paid with friendships, opportunities, money, and social ease. The resulting confidence is the confidence of someone who thinks he has been at the bench for decades.

“Not by sorcery or arrogance, but by tenacity and love of the craft.” That is how he explains it. This room concerns the craft: what he does with an argument, where he goes looking for information, and what can persuade him to move the furniture.¹

The world does not take requests

Start in 2009. Diego opens his account of the obstacles to rational knowledge by separating what happens from what one wishes would happen. A universe that pays attention to moral disapproval would be remarkably convenient. In his philosophy, we have not been issued that universe.

He makes the distinction memorable with a Portuguese heading:

Dever não é Ser, Bom não é Verdadeiro.

My translation: Ought is not is; good is not true.²

For him, explaining an unpleasant behavior is a different operation from endorsing it. Finding that something occurs in nature does not settle whether it is desirable. Wanting a pattern to be false does not alter its frequency. These separations are load-bearing throughout his work on human groups, sex, incentives, religion, and diet.

That is also why he objects so strongly to treating a statistical claim as a moral confession. He wants the claim about the world examined as a claim about the world. The next conversation—what people should do about that world—has its own premises. As he puts it:

A estatística não nos diz o que deveríamos fazer, ela nos diz o que está se passando.

My translation: Statistics does not tell us what we should do; it tells us what is going on.²

His 2009 discussion then asks for causes as well as descriptions. A visible difference may provoke contempt when its causes remain invisible; making the causal story intelligible can change what people want to do about it. This is his own reason for demanding more explanation. The audience’s first emotional reaction is not the end of the inquiry.

Take the sentence apart

The year before, he had published twenty guidelines for becoming a good thinker. They are strikingly practical: draw relationships, identify levels of explanation, distinguish a factual problem from a naming problem, reconsider settled conceptions, learn psychology, and inspect how a representation was produced.

One instruction could sit over much of his subsequent work:

Fazer a engenharia reversa de tudo que nos aparece na frente.

My translation: Reverse-engineer everything that appears in front of us.³

A claim arrives wearing a smart jacket: a graph, an academic title, a fashionable vocabulary, a confident speaker. Diego wants to see the factory that made the jacket. Which observations produced the graph? Which distinctions did the argument erase? What rewards did the speaker face? Which level of the system is doing the causal work?

His philosophical tools matter here. He regularly asks whether two people are disagreeing about reality or merely using the same word for different things. He asks whether a proposed explanation has skipped the intermediate machinery. His 2009 criticism of overaggressive reductionism targets the leap from particle physics to sociology without actually explaining the steps between them. Saying that everything is physical is only the beginning of a causal account.

He also insists that thought be made accessible to the animal doing the thinking. His final 2008 guideline calls for arguments in terms that are:

macaco-inteligíveis e mais do que isso macaco-emocionáveis.

My translation: intelligible to monkeys and, beyond that, emotionally graspable by monkeys.³

This explains a good deal of the furniture elsewhere in the Castle. Lobsters, ants, gods, games, horny apes, and unruly computer spirits are working examples. An argument should be something a reader can manipulate in imagination and remember afterward. Sometimes the philosophical equipment has a tail.

Several maps, one place to get lost

His attachment to reality also coexists with a longstanding interest in how minds represent it. In a November 2009 essay on Hilary Putnam, Diego explores a question more complicated than simply choosing between universal reason and anything-goes relativism: how can rationality vary with purposes, history, and conceptual tools while remaining constrained?

His image, explaining Putnam’s position, is an anchor with a long, flexible cable. Different movements are possible; unrestricted drift is not. The essay follows language, evolutionary history, cognitive habits, goals, and values into the construction of what humans call rational. It considers both one theory’s compatibility with several possible worlds and one world’s description by several theories.⁴

The distinction matters when reading him. He can demand an objective answer while spending pages arguing about the machinery through which an answer becomes accessible. He does not imagine the human brain as a pristine judge delivered outside history. It is an evolved object, equipped with particular capacities and failure modes. Knowing how that object works is part of using it well.

His early list therefore includes an unusually Diego-shaped instruction about intellectual honesty: do not invent arcane knowledge of unfamiliar subjects in order to sound authoritative, while also avoiding the demand that one suspend judgment on everything short of certainty.³ He wants a thinker able to commit, discriminate, and act.

Expertise has an internal structure

Diego’s distrust of authority is easiest to misunderstand when detached from his esteem for expertise. He spends enormous amounts of time learning from people he thinks know more about a subject. He also thinks an institution can certify competence at one task and then receive unearned deference on another.

His 2024 essay on doctors and nutrition is a particularly explicit statement of this method. He distinguishes treating common illnesses efficiently, memorizing a body of practice, evaluating statistical methods, understanding biochemical mechanisms, investigating incentives, and interpreting humans evolutionarily. These are separate abilities in his account. A qualification that bundles them together in the public imagination has not made them identical.⁵

He looks for a combination of mathematical competence, conceptual precision, domain knowledge, independence, and a grasp of the incentives producing a literature. He asks who selected the questions, who selected the data, what the institutions reward, and which assumptions are being smuggled into the conclusion. His objections to peer review and meta-analysis belong to that larger picture: he does not regard either label as an automatic transfer of trust.

The same essay assembles knowledge from clinicians, analytic thinkers, nutritional travelers, historical diets, individual variation, and anthropological reconstruction. A person who understands a mechanism, a person who observes communities closely, and a person who detects a statistical trick can contribute different parts of the answer. His ambition is to bring those parts into contact.

An earlier formulation names the broader rule:

Compatibilidade inter-conhecimentos.

My translation: Compatibility across bodies of knowledge.³

The universe is one place, so he expects knowledge about its different parts to be capable of fitting together. This is why a discussion that starts with dinner can end up requiring biology, philosophy, economics, history, psychology, and a diagram. You may have wanted a quick recommendation. Diego has opened six departments.

Go where the information is dense

He has a practical answer to the problem of how any single person can manage this: choose the moment of judgment carefully.

In 2017 he calls it the Maximal Information Density Decision. Imagine information unevenly distributed through space and time. Some people, conversations, and places contain much more relevant knowledge than others. Your own knowledge also changes: you may understand a subject particularly well after an intense period of study, and much less well after months away.

His instruction is:

Focus and concentrate on deciding in the moment with maximal information density.

And the point of moving around this landscape is to get:

close to a compressed good and trustworthy information source, or you’ve made yourself into one.⁶

This is an epistemology with travel expenses. Meet the person who built the thing. Put yourself among people who understand it. Learn enough to recognize the unusually informative moment. Then make the decision while the relevant material is assembled, rather than trying to remain equally current on every dimension of life forever.

The method connects his intellectual life to his social one. Friendship, conferences, direct conversations, and immersion in unfamiliar places can change what becomes thinkable. Information has a geography. Diego likes visiting it.

The chair can move

For an especially clear case of revision, turn to his February 2026 post about China. He had spent years arguing against its eventual dominance. Then an argument about bottlenecks in industrial and robotic production changed his view.

None of my arguments is stronger than the argument Elon made.

His argument is stronger than all my arguments combined.

He described the magnitude of the shift numerically:

I'll move the odds from like 1% to 40% but not to 60% yet.

Bayesian updates come in phases and that's as much as I'm willing to move.⁷

The announcement ended without much ceremonial cushioning:

My mind is changed.

I was wrong. For 15 years.

Taking the L

No excuses.⁷

Here, the objective he prizes—being right—requires relinquishing a position he held for a long time. The commitment attaches to getting the answer, even when that means publicly announcing a replacement.

There is a price beyond losing arguments. In 2013 he described looking at the books around the Future of Humanity Institute and realizing how much he would never manage to read. He listed activities and pleasures he had surrendered to keep up with the subjects that fascinated him. The confession underneath the grand intellectual ambition was wonderfully specific:

Come to think of it, the only two things that really, really interest me are minds and evolution.

Only two things. Unfortunately, those two things required most of the library.⁸

That earlier essay already distinguished the value of knowledge from the use of intelligence to display intelligence. He wrote:

The goal is using your intelligence to pursue your goals.⁸

The visitor leaves this room knowing why Diego can be simultaneously an enthusiastic student, an exhausting adversary, a supplier of sweeping judgments, and the author of a public admission that fifteen years of his thinking just lost to one argument. He takes the activity personally. Reality is where he has invested a large part of his life.

Further doors

• Guiões do Bom Pensador (2008): the twenty early thinking tools.
• Os Principais Obstáculos ao Conhecimento Racional and Uma Análise do Conceito de Racionalidade à Luz de Putnam (2009): the philosophical workshop.
• Drowning In An Information Ocean (2013): what wanting to know costs.
• Seize the Maximal Information Density Decision (2017): a method for choosing when and where to decide.
• On Choosing Right (2025) and Epistemic update: China (2026): confidence and revision in his own words.

Continue: The Animal That Built Civilization