Spoke · Vol. III · Epistemic Cognition
Deanna Kuhn’s three levels of epistemic development, why AI has made Level 3 non-negotiable, and how to teach it inside a home schedule — without a curriculum.
A companion essay to the manifesto. The manifesto argues for epistemic depth as the goal. This essay argues for evaluativist epistemology as the mechanism — and names the reason it has just become urgent.
By
Chaitanya Prabhu Hak
A decade in investigative journalism, applied for fourteen years to two children’s home education.
I
Kuhn's Ladder
Deanna Kuhn’s developmental epistemology names three levels a mind can occupy in relation to knowledge itself. They are not ages. They are stances. Every human being lives on one of them at any given moment — and most of us shift between them depending on the topic.
Level 01
The Absolutist
Knowledge is a copy of reality. Experts have it. Disagreement means one party is wrong. There is a fact of the matter and a settled answer, and to encounter uncertainty is to encounter someone’s failure. Schools teach absolutism whenever they treat the textbook as the source. This is where most curricula end.
Level 02
The Multiplist
Knowledge is opinion. Everyone’s view is equally valid. Disagreement is subjective. There is no fact of the matter, only preferences. This is the shrug of the adolescent who has noticed that adults disagree, but has not yet noticed that disagreement can be adjudicated. Culturally, we celebrate this stance and mistake it for open-mindedness. It is neither open nor minded.
Level 03
The Evaluativist
Knowledge is constructed. It can be assessed on the strength of its warrants. Some positions are better supported than others; disagreement is a live argument that responds to evidence; uncertainty is comfortable but not paralysing. Positions are held provisionally, revised deliberately, defended with reasons. This is the stance of the working scientist, the honest historian, and every serious adult intellectual life. It is also the one institutional schooling most reliably fails to produce.
Institutional schooling actively obstructs Level 3 by treating textbook claims as unassailable. Homeschools that rebuild school at the kitchen table inherit the same defect.
II
The AI Amplification Problem
Prior to 2022, misinformation was a supply-limited problem. Someone had to author a plausible falsehood. That someone had to spend time. Time was scarce. The costs of production capped the volume.
Post-2022, the cost of producing a plausible sentence dropped to approximately zero. Any child with a phone can generate ten thousand words of confidently-worded, well-structured, factually-adjacent prose on any subject in a minute. The information environment has been permanently altered. This is not a hypothesis. This is the world we are already in.
A child stuck at Level 1 or 2 in 2026 is functionally illiterate against AI-generated content. This is not a future problem. It is a today problem.
Consider the three stances in that environment. The Absolutist trusts the first plausible source they encounter, because plausibility is their proxy for truth — and AI outputs are astonishingly plausible. The Multiplist shrugs and grants every generated claim equal standing with every peer-reviewed study, because “everyone has an opinion.” Both stances collapse under the new information regime.
Only the Evaluativist survives, because only the Evaluativist has a vocabulary for warrant — and can therefore notice that a beautifully written paragraph with no sources, no methods, and no citations is not evidence. It is prose.
The most important thing a child can be taught in the AI era is not what to think, but how to weigh a thought.
III
What Evaluativism Requires
Evaluativism is a working stance, not a taught topic. It is composed of habits. The homeschool advantage is that these habits are cultivated in the small interactions of daily life, not in a scheduled block.
Every fact in a book was placed there by a person, informed by a source, corroborated by other sources or not, revised at intervals. A child who sees the fact as an object misses this. A child who sees the trail behind the fact does not.
Primary, secondary, corroborated, uncorroborated, hostile. Peer-reviewed, self-published. Replicated, single-study. These are not academic categories — they are working tools. Introduce them in ordinary conversations, not in a class.
For every claim, the habit of asking: how would things look different if the opposite were true? If a child cannot describe the world under the negation of a claim, they do not really understand the claim.
The absolutist needs a settled answer. The multiplist has none. The evaluativist holds positions provisionally — strongly enough to act, loosely enough to update. This is not a personality trait. It is a taught disposition.
The single hardest habit. When new evidence arrives, the mature epistemic agent updates and moves on. They do not double down. They do not retreat into cynicism. They say: I was wrong, this is what I now think, this is why. Model this in front of the child. It is the most important thing you can model.
IV
Concrete Teaching Moves
V
Assessment
Multiple-choice measures recall, at best. It cannot see reasoning, and it cannot see the shape of a disposition. The evaluativist stance is invisible on a bubble sheet.
Test it instead by giving the child two conflicting sources on the same question — a peer-reviewed study and a well-written blog post arriving at opposite conclusions, say — and asking: which do you believe more, and why? The answer they arrive at is not what you are grading. The reasoning they show is.
A child who can defend a provisional position with reasons — and revise it in front of you when a better reason arrives — has passed the only epistemic test that matters.
Closing
Evaluativism is a taught disposition. It cannot be bought in a box. It is transferred, slowly, in a thousand small conversations, by parents who model it more than they preach it. It is the deep purpose of the “Project-as-Proof” model. It is the reason a child raised well can survive a generation of infinite plausible content.
Continue reading
The Authority Engine
Companion essays that build the same argument from adjacent lenses.
The Manifesto
Epistemic Depth vs. Academic Mastery — the pillar.
The Instructional Architect at Home
Cognitive load, backward design, and why buying a curriculum is not the same as designing one.
Genes, Networks, and Reasoning
Systems thinking as the foundation of a STEM curriculum, not its capstone.