High performance · High learning
Productive success
The ideal. The student performs well, and the underlying schema construction has occurred. Outputs and understanding are aligned.
When the Output Looks Like Learning.
The essay is polished. The thesis is clear. The transitions are smooth. What if learning never happened at all?
Enter the companion
Jump to the framework
Read journal article
Open NLU colloquium deck
Return to hub
Scroll to begin
§ 01
/
The scenario
An essay arrives.
It performs at proficient or advanced.
An eighth-grader submits a five-paragraph essay on the causes of the American Civil War. By every rubric metric the teacher designed, the essay performs at proficient or advanced. The teacher suspects most of the text was generated.
Hover the highlighted passages — or tap a card below — to see what the rubric measures, and what it misses.
Submission #4128 · Grade 8 history
The Causes of the American Civil War
Grade 8 · U.S. History · Mr. Castellanos
The American Civil War, fought from 1861 to 1865, was the result of deep economic, political, and moral tensions that had been building between the North and the South for decades. While slavery is often cited as the central cause, the conflict was driven by a convergence of forces that made compromise increasingly impossible.
First, the divergent economies of the two regions created irreconcilable interests. The industrialized North relied on wage labor and tariffs to protect its growing manufacturing sector, while the agrarian South depended on enslaved labor to sustain cotton exports. These economic structures shaped political coalitions that increasingly viewed one another as existential threats.
Second, the question of slavery's expansion into new western territories transformed a regional dispute into a national crisis. The Kansas-Nebraska Act, the Dred Scott decision, and the raid on Harpers Ferry each accelerated polarization, eroding the political compromises that had previously held the union together.
Ultimately, the election of Abraham Lincoln in 1860 served as the immediate catalyst for secession, but the deeper causes had been gathering for generations. The war was not an accident of politics but the inevitable result of a republic that could no longer reconcile freedom with the institution of slavery.
Rubric · thesis clarity ✓
The opening names three causal categories. It scores well. But it could have come from any source — including one that has never thought about the war.
Rubric · topic sentence ✓
Structurally textbook-perfect. No struggle to find the framing — which means no productive cognitive work has been externalized into language.
Rubric · evidence ✓
Three correct historical references in a single sentence. Yet there's no encounter with which event mattered most, or why.
Rubric · conclusion ✓
Returns to the thesis with a gesture toward inevitability. Reads as wisdom. But "inevitable" can paper over any uncertainty the writer never had to face.
The essay exists.
The learning has not occurred.
Miner names this cognitive bypass: the symbolic output of an assignment is produced
without the cognitive work the assignment was designed to require.
§ 02
/
Three terms doing the work
Before the argument can be made,
the vocabulary has to be precise.
Miner asks the reader to hold three terms in technical senses that differ from everyday usage. Each does substantive work in what follows.
term iDefined
Symbolic production
The act of creating representations through systems of signs. A human being is doing the cognitive work of translating understanding into signs.
— writing a paragraph, sketching a diagram, composing an email, recording a spoken explanation
term iiDefined
Symbolic environment
The total set of signs and representations within which learners work. It is not a neutral container. It shapes what counts as a credible argument, a competent explanation, or a finished piece of writing.
— textbooks, discussions, teacher explanations, search results, and now AI-generated content the student consumes and submits
term iiiDefined
Cognitive bypass
The condition in which the symbolic output of an assignment is produced without the cognitive work the assignment was designed to require. The essay exists. The learning has not occurred.
— distinct from cheating, which is an ethical category. Cognitive bypass is a learning category about whether mental work has taken place.
§ 03
/
A media ecology lens
Walter Ong named three stages of how media reshape consciousness.
This framework adds two extensions.
Algorithmic secondary orality names the transitional stage in which humans still created symbolic content, but algorithms increasingly decided what reached which consciousness. Tertiary algorithmicity names the next threshold: systems that both curate and generate symbolic content on the human's behalf.
I.
Pre-history → antiquity
Primary orality
Producer: the human voice. Thought is aggregative, situational, communal — shaped by the affordances of spoken language as the medium for preserving knowledge.
II.
c. 3000 BCE → 20th c.
Literacy
Producer: the writing hand. Ideas can be fixed, compared, and systematically built upon. The knower separates from the known in ways that support sustained abstract analysis.
III.
Telephone → broadcast → web
Secondary orality
Producer: the speaking human, broadcast. Electronic media reintroduces oral immediacy into cultures that have already internalized literate patterns of thought.
IV.
2006 → early 2020s
Algorithmic secondary orality
Producer: the human, sorted by the feed. People still create the posts, videos, comments, and images, but recommendation systems increasingly determine what reaches whom, when, and in what sequence.
V.
Late 2022 →
Tertiary algorithmicity
Producer: the model, often with the human as prompter. The technology no longer only mediates symbolic activity — it generates it. Students can delegate symbolic production before the learning struggle has occurred.
The fifth stage — what's different
For the first time, the medium becomes the maker.
In primary orality, humans produced speech. In literacy, humans produced text. In secondary orality, humans produced electronically mediated speech and text. In algorithmic secondary orality, humans still produced the content, but algorithmic systems increasingly curated its circulation. In tertiary algorithmicity, algorithmic systems share — and sometimes entirely assume — the role of generating symbolic content. This is a shift in who or what is making meaning, not merely in how meaning travels.
§ 04
/
Four characteristics that make this stage distinctive
Each carries a direct implication for school.
i.
N
Noetic displacement
The externalization of cognitive operations — analysis, synthesis, evaluation, argumentation — that previously required internal mental processing. Where literacy externalized memory, generative AI externalizes thinking itself.
A sixth-grader is asked to explain the water cycle in their own words. Before AI, their struggle to articulate the difference between evaporation and condensation was the learning. With AI, the productive gap never opens.
ii.
R
Rhetorical saturation
The condition in which AI-generated content becomes so prevalent and so polished that the symbolic environment is flooded with competent but undifferentiated prose. The scarcity value of competent writing — which historically functioned as evidence of learning — collapses.
An English teacher reviewing essays on The Great Gatsby finds prose that is structurally identical: competent thesis, orderly paragraphs, tidy conclusion — stripped of the idiosyncratic voice that marks genuine engagement.
iii.
E
Existential abstraction
The increasing distance between the learner and the experience of confronting genuine intellectual difficulty. Learning depends on the encounter with resistance — the moment when existing mental models prove inadequate. AI offers a way to bypass that encounter entirely.
Following a GPS voice versus reading a topographic map. You reach the destination either way. Only one of them leaves you with a mental model of the terrain.
iv.
I
Infrastructural consolidation
The growing dependence of schools on a small number of commercial AI platforms whose design choices, pricing structures, and content policies shape the symbolic environment of learning from outside the institution itself.
What counts as a good essay, a clear explanation, a reasonable argument — these choices migrate from teachers, departments, and districts to platform designers and model trainers most teachers will never meet.
§ 05
/
The trap, in two dimensions
Performance is not learning.
Manu Kapur's matrix makes the gap visible.
Productive failure research arranges four relationships between performance and learning. Three of the quadrants have long been familiar in schools. Generative AI dramatically expands the fourth.
Performance →
← Learning
i.
ii.
High performance · Low learning
Unproductive success
A correct or competent outcome without the cognitive processing the performance was meant to develop. The polished essay; the right answer; no schema. Generative AI is, in this framing, an unproductive-success engine.
iii.
Low performance · High learning
Productive failure
The counter-intuitive but well-documented finding: initial failure, when properly structured, can produce deeper learning than immediate success. The struggle is the mechanism.
iv.
Low performance · Low learning
Unproductive failure
What most educators already work to avoid: the student fails to perform and fails to learn. The familiar enemy of schooling.
Generative AI produces polished symbolic outputs from minimal cognitive inputs.— On the unproductive-success engine
§ 06
/
A metaphor for what gets lost
Two ways of getting somewhere.
Only one leaves you with a map of the terrain.
● Topographic map · with friction
Reading the map.
You analyze the landscape, feel the incline of the hills, make active choices about the path.
The cognitive resistance is the point. You arrive — and you arrive with a mental model. If the map disappears tomorrow, you can still navigate.
● GPS voice · frictionless
Following the turns.
You arrive at the correct destination. You maintain no presence in the journey.
If the GPS fails, you are lost — because you never developed a mental model of the terrain. The result is yours. The wisdom of the struggle is not.
Existential abstraction is the GPS condition, scaled across every domain that runs on language.
§ 07
/
Watch two essays write themselves
Same prompt. Same grade. Same word count.
Different cognitive events.
The output, on the page, is comparable. The mechanism that produced it is not.
Stream A · 12s elapsedvia LLM
The causes of the French Revolution.
Stream B · 42 min elapsedby student
The causes of the French Revolution.
Stream A is symbolic production by an algorithm; the student's contribution is the prompt. Stream B is symbolic production by a student, with all of its hesitation, deletion, and reach. The polished column will be graded the same. The cognitive event has only happened in one of them.
§ 08
/
The Pedagogical Friction Framework
If the problem is structural,
the response has to be designed.
Pedagogical friction is the intentional design of learning environments that preserve the productive cognitive resistance necessary for genuine understanding — in contexts where tools make bypassing that resistance both possible and attractive. The framework names four dimensions along which it operates.
Dimension 01 · cognitive
Noetic friction
Schema construction · accommodation
The cognitive resistance encountered when a learner's existing knowledge structures prove insufficient and must be revised. Difficulty is not an obstacle to learning. It is a constitutive feature of it.
Threatened by AI when — the signal of not knowing is muted by a polished output before accommodation can occur.
Design moveRequire initial attempts before any AI interaction. Assess the visible trajectory of thinking, not only the finished product.
Dimension 02 · expressive
Rhetorical friction
Writing as a mode of thinking
The resistance encountered in the act of communicating to others — the struggle to find the right words, anticipate objections, organize complex ideas. Writers discover what they think through the labor of trying to say it.
Threatened by AI when — production of text is decoupled from the cognitive processes writing was meant to engage.
Design moveAssignments anchored in personal experience, situated knowledge, live dialogue, and visible revision. Make voice irreducibly human.
Dimension 03 · dispositional
Existential friction
Sitting with not-knowing
The encounter with genuine uncertainty, intellectual vulnerability, the limits of one's own understanding. Curiosity, persistence, and intellectual humility are forged in repeated experiences of not knowing — then coming to know.
Threatened by AI when — answers arrive before students have fully experienced the questions. The habit of inquiry atrophies.
Design moveSequences that require students to sit with ambiguity before resolution. Teachers model intellectual vulnerability as a stance, not a weakness.
Dimension 04 · institutional
Infrastructural friction
Policy · pacing · assessment
The structures — grades, schedules, pacing guides, technology policies — that either support or undermine the preservation of productive friction. A teacher's good assignment can be hollowed out by a system that rewards frictionless output.
Threatened by AI when — policies address only integrity and data privacy, while pedagogy is left to chance.
Design moveAI policy as a three-pillar instrument: integrity, privacy, and an explicit pedagogical stance on the work to be preserved.
Friction without support does not produce productive struggle. It produces aversion to challenge.— A necessary caveat
§ 09
/
Not all friction is the same
Productive friction drives learning.
Exclusionary friction blocks access to it.
Any honest framework has to distinguish the two — or it becomes a rationale for withholding tools that serve equity goals. The design challenge is to reduce exclusionary barriers while preserving productive cognitive demands.
An English learner reading a complex text.
● Reduces exclusionary friction
AI translation during the reading phase
The language barrier is unrelated to the construct being assessed — historical reasoning, literary interpretation, scientific argument. The tool removes a barrier that was never the point of the assignment.
● Would erode productive friction
AI generation during the writing phase
Language production is the construct here. Allowing AI to compose the student's response bypasses the cognitive demand the assignment was designed to build. The tool, in this position, replaces the lesson.
A district's AI policy cannot be one rule applied uniformly. It is a set of principles teachers and leaders interpret in light of which students, which goals, which moment in the learning sequence.
§ 10
/
Closing
The question was never whether to ban it
or embrace it.
That framing — which has dominated public discourse since late 2022 — is a false binary that obscures the more fundamental questions:
What cognitive work must be protected, and how can institutions protect it?
Schools that treat AI as just the latest technology to integrate into existing practice will find that the practice itself has been hollowed out. Schools that develop the conceptual and institutional infrastructure to distinguish between productive and unproductive uses of AI — between friction that serves learning and friction that impedes it — will be better positioned to fulfill their core mission in a period of substantial technological change.
The Pedagogical Friction Framework is offered as one contribution to that infrastructure, with the recognition that much work remains to be done.
Original article
Miner, M. J. When the Output Looks Like Learning: Tertiary Algorithmicity, Unproductive Success, and the Case for Pedagogical Friction in K–12 Schools. i.e.: Inquiry in Education, Vol. 18, Iss. 1, Article 4. Published by the Center for Inquiry in Education, National-Louis University, Chicago, IL.
↩ Continue
NLU Doctoral Colloquium Deck
Use the companion presentation for a session-oriented version of the same argument about AI, unproductive success, tertiary algorithmicity, and pedagogical friction.
Open the deck Open the hubend of companion · designed as a reading aid · not a substitute for the article itself