What It Costs to Think

Socrates did not write books, found a school, or produce a doctrine. He walked around Athens asking people to define the words they were already using, and then showed them their definitions did not hold. That was the entire method. He called it midwifery — he delivered nothing of his own, he only helped other people give birth to what they turned out not to actually believe.

Athens executed him for it.

Every essay recommending his approach leaves that out, or mentions it as a tragedy that happened to a good man rather than as information about the method. It is information about the method. A society had a man who made people examine their assumptions in public, and it decided, by a jury of five hundred citizens, that it would rather not.

That is where an honest argument for teaching independent thought has to start, because everything that follows is easier to sell and less true.

What the Method Actually Is

Not knowledge. This is the part that gets lost every time.

We use "critical thinking" to mean something like being smart about things, and then we teach more content and call it done. But the Socratic operation was never about the acquisition of information. It ran the other direction: take a belief the person is confident about, ask what it commits them to, follow the commitments until one of them contradicts another. The output is not a new belief. The output is the discovery that the old one was never load-bearing.

That is a strange skill, and it is not the one we test for. Testing rewards producing the correct answer. This rewards discovering that you did not have one.

What We Teach Instead

We teach conclusions. A student learns which answers are right, and learns to hold them the way they were delivered — as settled, as authoritative, as something a person is either in possession of or not.

I have argued elsewhere that the system is misassigning students to a curriculum most of them will never use. This is the deeper version of the same failure. Even where the content is right, the manner of holding it is what gets transmitted, and the manner we transmit is certainty. Twelve years of practice at receiving conclusions from an authority is excellent preparation for receiving conclusions from an authority, which is precisely what a person then does for the rest of their life, from whichever authority reaches them first.

What would actually be taught instead is narrow enough to specify:

Questioning assumptions. Given a claim, name what has to be true for it to hold. Most claims collapse at this step and it takes ninety seconds.

Argument structure. Distinguish a claim, a reason, and evidence. Notice when the reason and the claim are the same sentence wearing different clothes.

Source evaluation. Not "is this a reliable outlet" but "what would this source say if the opposite were true?"

Calibration. Say I am about sixty percent on this and mean it. Then check, later, whether the things you were sixty percent on came out right about sixty percent of the time. This is a trainable skill, it is almost never trained, and it does more work than everything else on this list combined.

Steelmanning as an assessment. Be graded on the strongest version of the case against your own position, written well enough that someone who holds it would sign it.

None of that requires new content. It requires grading a different thing.

The First Cost: It Is Not Armor

Here is where I have to break with the version of this argument I would rather make.

The comfortable claim is that critical thinkers are protected from manipulation — that a population trained to analyze arguments becomes resistant to demagogues and propaganda. I believed a version of this. The evidence is not kind to it.

Reasoning skill does not reliably move people toward truth. It moves them toward wherever they were already going, faster and with better justifications. Dan Kahan's work on motivated numeracy found that when a data problem was politically neutral, the more numerate subjects got it right more often — as you would expect. When the identical numbers were relabeled as gun-control data, the most numerate subjects became more polarized, not less. They used their skill to get to their side's answer.

Mercier and Sperber's account explains why: reasoning may have evolved less as a truth-finding instrument than as a device for producing arguments and evaluating other people's. That is a tool for winning, and it works beautifully for whoever is holding it.

So a well-trained mind is a better advocate, and an advocate is not the same thing as an inquirer. The most elaborately reasoned positions I encounter are frequently the most detached from reality, because elaboration is what a capable person does with a conclusion they arrived at some other way.

This does not make the discipline worthless. It relocates where the value is. The part that helps is not raw analytic horsepower — it is the specific, unnatural habit of actively looking for what would prove you wrong, and of tracking your confidence separately from your commitment. That habit is teachable. It is also the one part of the package that no amount of intelligence supplies on its own.

The Second Cost: It Is Slow and You Cannot Afford It Everywhere

Genuinely examining a foundational belief takes hours and often ends without a replacement. Inherited belief is instantaneous and usually good enough.

That is not a joke at anyone's expense. It is a real efficiency, and it is why the practice is rare. Nobody has the budget to audit their position on medicine, monetary policy, climate physics, criminal justice, and nutrition. You will take almost everything you believe on trust from someone else, and so will I, and the alternative is not a more examined life — it is paralysis.

Which means the skill is not question everything. It is choosing correctly which small number of beliefs to actually audit: the ones you act on, the ones that would be expensive to have wrong, and the ones you notice you are unwilling to examine. That last category is the highest-yield and the least visited.

The Third Cost: What Athens Did

The social bill comes due before the intellectual benefit does.

Say I do not know yet in a room where everyone has a position and watch what happens. It does not read as rigor. It reads as evasion, or cowardice, or disloyalty — because in most rooms the question on the table is not what is true but whose side are you on, and the honest answer is unavailable in that currency.

The person who has genuinely examined their group's core belief is a problem for the group whether they end up agreeing with it or not, because the examination itself signals that the belief was the kind of thing that could have failed. That is what Athens objected to. Not his conclusions — he barely had any. The examining.

So the discipline costs standing. It costs you the easy membership of certainty. Anyone recommending it while promising it will make you admired is selling something.

Why It Survives Its Own Bill

Given all of that, the case still holds, and it holds for a narrower reason than the one I started with.

Not because trained thinkers are immune — they are not. Not because it is efficient — it is not. But because the alternative is not neutrality. A person without the discipline is not unpersuaded; they are persuaded by whoever reached them first and said it most confidently. There is no unoccupied position. The choice is not between examined belief and no belief; it is between beliefs you have inspected and beliefs installed by parties with an interest.

And because of one property the alternatives lack.

The method can be shown to be wrong. Every finding in this essay that cuts against it — motivated numeracy, the argumentative account of reason, the practical impossibility of examining everything — arrived through exactly the procedure it undermines. The discipline generated its own strongest objections and then required me to print them.

Nothing built on certainty can do that. A closed system defends itself; it cannot be corrected by evidence, because evidence is admitted only when it agrees. This one is corrigible by construction. Show me a study demonstrating that trained calibration does not transfer outside the lab and I have to revise the fifth item on my list — and the revised version will be better than what it replaced.

That is not a small consolation prize. It is the entire advantage, and it is the reason the claim is worth making.

So I will not call this indubitable, or unassailable, or beyond argument. I would be contradicting the essay in its final line, and the contradiction would be the most memorable thing in it.

Here is the honest ending instead. What I have described is expensive, socially costly, insufficient on its own, and provably not a defense against being wrong. I think it should be taught anyway, to everyone, starting early. And if I am wrong about that, the method I am recommending is the one that will show it.