Learning to Read, Reading to Learn

Prompting a Bot to Self-Audit the Quality of its Output

Terry Underwood, PhD — Sun, 30 Aug 2026 03:17:38 GMT

Reading AI output can feel “lulling,” a term writers have used more than once. Long chats can be exhausting, numbing, and hallucinations from early in the context window can proliferate and expand. If you find your self reading a lengthy bot response that seems absurd but would take time to critique, try pasting in either of the two audit prompts I’m developing. These aren’t finished yet; I’d love to hear from you if you try either or both.

The Two-Tier Audit Checklist

Tier 1: Propositional Extraction (Stripping the Form)

● [ ] What is the naked claim? Stripped of hedges (“It is important to consider,” “Can be seen as”), what single assertion is the text actually making?

● [ ] Is it falsifiable? If you remove the conditional phrasing, does it express a clear, testable stance, or is it a tautology/platitude?

● [ ] What is the core verb? Is the AI asserting an action/mechanism or merely describing a domain?

Tier 2: Epistemic Calibration (Auditing the Substance)

● [ ] Where is the empirical weight? What concrete mechanisms, named frameworks, or data points back the Tier 1 claim within this text?

● [ ] Confidence Rating (1–5): How justified is the claim based only on the evidence presented in the output (ignoring how well it reads)?

● [ ] Action:

○ High Fluency + Low Evidence (Rating 1–2): Mirage. Reject or prompt for concrete specifics.

○ High Fluency + High Evidence (Rating 4–5): Substance. Integrate into schema.

Three-Part Audit with Less Bot Constraint.

AI Audit Prompt: “Re-analyze the text above using this structure to remove hedging, vagueness, and redundancy:

1. Core Propositions: Strip away all transitional phrasing, hedges, and introductory fluff. Express the main arguments as a bulleted list of direct, unhedged assertions.

2. Substantiation Audit: Beneath each assertion, list only the concrete evidence, specific mechanisms, or named data points provided to support it. If an assertion relies entirely on platitudes or circular logic, label it explicitly as [Unsubstantiated / High-Fluency Noise].

3. Falsifiability Check: Highlight any claim that cannot be proven wrong due to overly vague or conditional language.”

EXAMPLE 1: Three-Part Audit

My Prompt to Claude Sonnet: Why do people study biology?

Claude’s Response:

People study biology for a mix of reasons that tend to cluster into a few categories:

Sheer curiosity about life itself. Biology answers some of the oldest questions humans have asked: what are we, how did we get here, why do we age and die, how does a single cell become a whole organism. That pull toward understanding living systems is often the first hook, especially for kids fascinated by animals, plants, or their own bodies.

Practical and medical stakes. Biology underlies medicine, agriculture, and public health. People study it to become doctors, nurses, researchers developing treatments, or scientists working on food security and disease control. The COVID pandemic pushed a lot of people toward virology and immunology for exactly this reason — the field has obvious, high-stakes payoff.

Environmental and conservation motives. Understanding ecosystems, biodiversity, and climate impacts on living systems draws people who want to address environmental problems — conservation biology, ecology, and related fields exist because people want tools to protect species and habitats.

Career and economic paths. Biology is a gateway to well-compensated, stable careers — medicine, biotech, pharmaceuticals, research science. For plenty of students, that’s the primary driver, curiosity aside.

Philosophical and existential interest. Some people study biology because it bears on deep questions — consciousness, free will, the nature of death, what separates life from non-life. Evolutionary biology in particular attracts people wrestling with questions about human nature and our place among other species.

It’s foundational to other fields. Psychology, medicine, environmental science, biochemistry, and even fields like bioengineering and AI (neural networks, e.g.) all lean on biological knowledge, so people study it as infrastructure for something else they care about.

Most biologists I’d guess land somewhere in a combination of these — curiosity that gets reinforced by seeing the practical stakes, or vice versa.

AI Audit Prompt Response

1. Core Propositions

Curiosity about life processes drives people to study biology.
Biology underlies medicine, agriculture, and public health.
COVID-19 increased interest in virology and immunology.
Environmental and conservation concerns drive people to study biology.
Biology is a path to stable, well-paid careers.
Some people study biology to address philosophical questions (consciousness, death, human nature).
Biology is foundational infrastructure for other fields, including psychology, medicine, and AI.
Most biologists combine curiosity with practical motivation.

2. Substantiation Audit

Curiosity drives study of biology. [Unsubstantiated / High-Fluency Noise] — No mechanism, no data. Restates the claim as its own evidence (”curiosity” is defined by the behavior it’s meant to explain).
Biology underlies medicine, agriculture, public health. Definitionally true, not empirically demonstrated. This is a structural fact about the field, not evidence that it motivates people to study it.
COVID-19 pushed people toward virology/immunology. Named event, plausible causal story, but I cited no enrollment data, survey, or study. As stated, it’s an assertion dressed as evidence.
Environmental motives drive conservation biology. [Unsubstantiated / High-Fluency Noise] — Circular: “people who want to protect species study the field about protecting species.” No named mechanism, no source.
Career/economic incentives drive biology study. No salary data, no enrollment statistics, no named field comparison. Plausible but asserted, not shown.
Philosophical/existential motives. [Unsubstantiated / High-Fluency Noise] — No named philosophers, no named debates, no evidence anyone actually enrolls in biology for this reason versus philosophy or theology.
Biology as infrastructure for other fields. Partially substantiated: psychology, medicine, biochemistry depending on biology is structurally accurate. The AI/neural-networks claim is weaker — the analogy between biological neurons and artificial ones is real, but citing it as a reason people currently study biology is a stretch with no supporting example.
Most biologists combine curiosity and practical stakes. [Unsubstantiated / High-Fluency Noise] — Explicitly flagged as a guess in the original text (”I’d guess”). No data on biologist motivations was presented.

3. Falsifiability Check

“Often the first hook” — “often” is unquantified; no threshold exists to prove it false.
“For plenty of students, that’s the primary driver” — “plenty” is a vague quantifier with no defined proportion.
“Most biologists... land somewhere in a combination” — self-admitted speculation, not falsifiable as stated.
The environmental-motive claim is circular, which makes it unfalsifiable by construction — it can’t fail a test because it isn’t making a distinct causal claim.
The COVID claim is the one exception: it references a real, checkable event, and enrollment/interest data for that period does exist. It’s the only proposition in the list that could be falsified with a citation, though I didn’t provide one.

Net assessment: the original answer was a taxonomy of plausible categories, not an evidenced argument. One claim (COVID) is checkable; the rest are either definitional or asserted without support.

EXAMPLE 2: Two-Tier Audit

My Prompt: Decide whether slavery was the cause of the American Civil War and make your case. Maximum 400 words. Average sentence length 18 words.

Claude Sonnet’s Response:

Slavery caused the Civil War. The evidence is direct, not inferential, and it comes largely from the Confederacy’s own words.

When Southern states seceded in 1860 and 1861, they wrote formal declarations explaining why. Four of them survive in detail, and each names slavery as the central grievance. South Carolina’s declaration objects to Northern states that had become “hostile to slavery” and complains about the North’s failure to enforce fugitive slave laws. Mississippi’s declaration states plainly that its position was “thoroughly identified with the institution of slavery.” Georgia and Texas echo this. These were not vague states’-rights documents. They named the specific right in question, repeatedly.

Confederate Vice President Alexander Stephens made the point even more explicitly in his 1861 “Cornerstone Speech.” He said the new government’s foundations rested on the truth that the Black man was not equal to the white man, and that slavery was his natural condition. He framed this as a correction of the Founders’ error, not a defense of abstract federalism.

The “states’ rights” framing, when examined closely, was specifically the right to hold humans as property, and to extend that system into new territories. Sectional conflict throughout the 1850s tracked slavery issue by issue: the Missouri Compromise, the Kansas-Nebraska Act, Bleeding Kansas, the Dred Scott decision, John Brown’s raid. Every major crisis of the decade centered on whether slavery would expand or contract. Lincoln’s 1860 election, on a platform of stopping slavery’s expansion, triggered secession within weeks. Southern states didn’t wait to see his actual policies. The threat to slavery’s expansion was enough.

The Lost Cause narrative, developed after the war, recast the conflict as a defense of states’ rights and Southern heritage, minimizing slavery’s role. This reframing served reconciliation and later Jim Crow politics, but it contradicts the documentary record from 1860 and 1861. Historians across the ideological spectrum now treat slavery as the war’s central cause, based on this evidence.

Economic and cultural differences between North and South existed and mattered. But they mattered because they were downstream of slavery: a plantation economy built on unpaid labor, a social order built on racial hierarchy. Strip slavery out of the equation, and secession has no coherent motive left. The war’s own architects told us why they fought.

Proposition 1 — Slavery caused the Civil War (overall thesis)

Tier 1

Naked claim: Slavery is the war’s cause, not a cause.
Falsifiable: Yes, in principle — testable against secession documents, political rhetoric, and the sequence of events. As stated in the original text, it’s asserted as conclusion rather than argued incrementally toward.
Core verb: “Caused.” An asserted mechanism, not a described domain.

Tier 2

Empirical weight: This claim is not independently supported — it’s the sum of Propositions 2–9. Its strength equals their combined strength.
Confidence Rating: Contingent. Rate after scoring the components below.
Action: Hold pending Tier 2 review of supporting claims.

Proposition 2 — Secession declarations name slavery as the cause

Tier 1

Naked claim: Four states’ secession declarations cite slavery explicitly.
Falsifiable: Yes — checkable against primary documents.
Core verb: “Cite” / “name.” Asserts content, not just domain.

Tier 2

Empirical weight: South Carolina and Mississippi given specific quoted language. Georgia and Texas asserted but not quoted or detailed in the original text.
Confidence Rating: 3/5. Half the claim is textually grounded, half is unverified within the piece itself.
Action: Partial mirage. The two quoted states are substance; the other two need specifics before counting as evidence rather than assertion.

Proposition 3 — Cornerstone Speech confirms slavery as founding rationale

Tier 1

Naked claim: Stephens said the Confederacy’s foundation rested on Black inferiority and slavery as natural order.
Falsifiable: Yes — the speech is a fixed historical text.
Core verb: “Said.” A direct citation of asserted content.

Tier 2

Empirical weight: Named speaker, named speech, named year, paraphrased content that matches the historical record.
Confidence Rating: 5/5.
Action: Substance. Integrate.

Proposition 4 — “States’ rights” meant the right to hold and expand slavery

Tier 1

Naked claim: The states’-rights framing reduces to a slavery claim.
Falsifiable: Weakly. The claim defines its own terms — “states’ rights” is redefined as “slavery rights” by assertion, then the redefinition is used as proof.
Core verb: “Meant.” Interpretive, not evidentiary.

Tier 2

Empirical weight: No mechanism given beyond re-pointing to Proposition 2. No independent test for why states’-rights claims can’t also encode other content.
Confidence Rating: 2/5.
Action: Mirage. Reads as settled; is actually circular — restates the thesis in different vocabulary.

Proposition 5 — 1850s crises were slavery disputes

Tier 1

Naked claim: Missouri Compromise, Kansas-Nebraska, Bleeding Kansas, Dred Scott, John Brown’s raid were all about slavery.
Falsifiable: Yes in principle, but not tested here — each event is named, none explained.
Core verb: “Tracked.” Describes a pattern without demonstrating it.

Tier 2

Empirical weight: Zero mechanism per event. Five proper nouns substitute for five arguments.
Confidence Rating: 1/5.
Action: Mirage. High fluency — the list sounds authoritative — with no evidence content. Reject or demand one sentence of mechanism per event.

Proposition 6 — Lincoln’s election triggered secession

Tier 1

Naked claim: Southern states seceded because of Lincoln’s election, on an anti-expansion platform.
Falsifiable: Yes — timeline is checkable.
Core verb: “Triggered.” Asserts causal action.

Tier 2

Empirical weight: Timing given (”within weeks”), no citation linking platform to secession beyond assertion.
Confidence Rating: 2/5.
Action: Mirage leaning toward substance. Plausible and broadly correct historically, but the text itself supplies no documentary link — it asserts the causal chain rather than showing it.

Proposition 7 — Lost Cause narrative minimized slavery for political reasons

Tier 1

Naked claim: A postwar reframing served reconciliation and Jim Crow politics.
Falsifiable: Yes, but nothing in the text allows the test.
Core verb: “Served.” Asserts function with no support.

Tier 2

Empirical weight: No source, date, author, or text named.
Confidence Rating: 1/5.
Action: Mirage. Reject or prompt for a named text (e.g., the Lost Cause historiography of Jubal Early, or the Dunning School) and a specific mechanism.

Proposition 8 — Historians across the ideological spectrum now treat slavery as central

Tier 1

Naked claim: Consensus exists among historians.
Falsifiable: No, as written. “Across the ideological spectrum” specifies no count, no names, no method — any counterexample can be waved off as outside the referenced spectrum.
Core verb: “Treat.” Describes a state of affairs asserted, not shown.

Tier 2

Empirical weight: None. Appeal to unspecified authority.
Confidence Rating: 1/5.
Action: Mirage. Classic high-fluency noise — this is the sentence most likely to persuade a reader while carrying the least content.

Proposition 9 — Non-slavery factors were downstream of slavery

Tier 1

Naked claim: Economic and cultural differences existed only because of slavery.
Falsifiable: No, as stated. The counterfactual (”strip slavery out, no motive remains”) sets its own untestable condition.
Core verb: “Mattered because.” Causal claim asserted, not demonstrated.

Tier 2

Empirical weight: None given — no economic data, no comparative case, no counterfactual actually run.
Confidence Rating: 1/5.
Action: Mirage. Assumes the conclusion it’s meant to support.

Net assessment: Two propositions clear as substance (3, and partially 2). The rest — the load-bearing rhetorical claims about consensus, states’ rights, the 1850s pattern, and the Lost Cause — are fluent but evidentially thin as written. The original essay’s actual defensible core is Propositions 2 (partial) and 3. Everything else needs sourcing before it counts as argument rather than assertion.

The PROPEL Folio: Grade the Judgment, Not the Essay

Terry Underwood, PhD — Fri, 28 Aug 2026 20:41:15 GMT

“What’s ‘Awk?’”

In 1989, during my last year teaching fourth grade, I participated in a year-long teacher-research professional development program. Our teacher, Jim Hahn, taught high school English and was a leader in the regional writing project. He shared an inquiry he had conducted into a recurring classroom phenomenon sometimes called the “handback”: the precise moment when a teacher returns graded essays.

This was in the age of hard copy and handwriting, but the basic situation does not seem to have changed much.

Hahn’s findings were uncomfortable and provocative. Some students looked at the grade and nothing else. Many who read the comments in the margins could not explain them. Some appeared to believe that the comments were not really meant to be read at all.

“What’s ‘Awk?’” they asked.

The teachers Hahn talked with assumed that something like respectful engagement, or at least appreciation, was happening at the other end of the line. Neither side understood the other’s theory of action.

When a student receives a judgment, the student does not make a judgment.

That is the problem I want to explore here. Delaying a grade may persuade students to read our comments, but it does not transfer evaluative authority to them. Students still wait for the teacher to decide what their writing means.

What if we taught them to make the first judgment?

The Teacher as Researcher

The program I attended in 1989 arrived at the dawning of the Age of Portfolios. The California Writing Project encouraged teachers not to wait for capital-R Research to prescribe what they should or should not do. Nobody was in a better position to understand a roomful of students than their teacher. Teachers could be researchers, too.

Our textbook was Dixie Goswami and Peter Stillman’s Reclaiming the Classroom, a forceful challenge to the idea that research belonged only to university professors.

Ten years later, Marilyn Cochran-Smith and Susan L. Lytle looked back on what had become a teacher-research movement. In “The Teacher Research Movement: A Decade Later,” they described five developments: the growing place of teacher inquiry in teacher education, professional development, and school reform; the emergence of distinct traditions of classroom research; the circulation of teacher research beyond local settings; a vigorous critique of the movement itself; and the possibility that practitioner inquiry might reshape university culture.

The movement was not fading. It was gaining steam, even as standards and accountability policies threatened to replace the teacher-as-researcher with the teacher-as-technician.

Stephen Fishman and Lucille McCarthy’s 2000 book, Unplayed Tapes: A Personal History of Collaborative Teacher Research, traced a related history. They described teacher-researchers blending systematic classroom inquiry with personal narrative, often in collaboration with university researchers. The work was empirical without pretending that the teacher had no place in the story.

Hahn’s handback study belonged to that tradition. It began not with an abstract theory but with a recurring classroom scene that did not make sense. Teachers spent enormous amounts of time commenting on student writing. Students often ignored, misunderstood, or discarded those comments. Why?

Nearly forty years later, teachers are still investigating that question one classroom at a time.

Delaying the Grade

In 2017, writing on Cult of Pedagogy, Kristy Louden described physically separating her comments from the grade. She withheld the score until students had read and responded to what she had written.

She described the familiar handback:

“There are the eye rolls and the whispering to each other of ‘What did you get?’ and ‘She gave me a ___.’ Next thing you know, the paper you spent so long reading and marking has been shoved into the abyss of the backpack or tossed carelessly in the recycle bin.”

On The Broken Copier in 2025, high school teacher Marcus Luther described building an entire lesson around the return of feedback. His students wanted the numbers immediately. He insisted that the feedback needed a structure around it first.

“‘CAN WE SEE THEM LIKE RIGHT NOW, MR. LUTHER?’ was pretty much the consensus reaction in all the class periods, yes, they told on me to peers in other classes, so I had the absolute joy of spending the day fending off hordes of desperate pleas for me to at minimum enter the scores in the grade book, even if they couldn’t see the feedback yet.”

Louden and Luther respond to the problem Hahn identified. Once a grade is visible, it consumes the student’s attention. Delay the grade, restructure the handback, and the comments have a better chance of being read.

That is a meaningful fix. But it does not change the fundamental flaw.

The student still waits for the verdict. The teacher still decides what is good or not so good. The student’s job is still to understand, accept, and perhaps act on the teacher’s assessment.

Roberta Camp, working with Dennie Palmer Wolf and others on the Arts PROPEL portfolio project, diagnosed a more troubling problem. It is one thing to protect teacher feedback by delaying the grade. It is another to turn the tables and disconnect the cables.

Let’s teach students to assess their own essays.

Assessment as an Episode of Learning

Arts PROPEL brought together Harvard Project Zero, Educational Testing Service, and the Pittsburgh Public Schools. Its designers drew inspiration from the working portfolios of artists. Artists do not learn primarily by receiving a verdict on each finished work. They accumulate work, study it, select from it, reconsider earlier choices, and develop standards through practice.

In the Arts PROPEL writing handbook, students collected finished pieces alongside drafts, notes, and reflections. At intervals, they selected pieces from those collections and reflected on their choices. The completed portfolio included writing, process evidence, a writing inventory, an annotated table of contents, and a final reflection.

The practice of looking back was neither familiar nor easy for students. Camp observed that they arrived expecting the old arrangement:

“[W]e were not terribly surprised that they came into the school year expecting that the teacher would tell them what was good or not so good in their work, and that their job would be to understand and accept the teacher’s judgment.”

If students were going to learn to assess themselves, assessment would have to become “an episode of learning, not simply an occasion for correction.” The student would have to judge while the question was still open. That judgment would then have to be respected enough to be examined for evidence of development as a writer.

Hahn and Camp therefore diagnosed two different problems.

Hahn showed that students may ignore comments once a grade is visible. That is a problem of sequence and attention. It can be addressed by delaying the grade or restructuring the handback.

Camp’s work points toward the hand-in rather than the handback. Students should not merely receive an assessment. They should participate in constructing it.

Delaying the grade does not, by itself, alter the expectation that the teacher will decide what is good. Changing how the grade is determined might.

A PROPEL-Inspired Experiment

What follows is a five-month, PROPEL-inspired provisional portfolio design for a writing class. It eliminates grades on individual pieces while still producing a final grade—after five months. I know that this is a difficult premise which may require a downplayed grade delay rather than a grade elimination.

By delaying a grade until students have started on the next assignment, say, a few weeks, and by factoring in the single assignment grade as a small portion of the final grade, teachers can meet their local obligation for reporting their grades externally during the term without shutting down students’ opportunity to judge.

It also makes a deliberate departure from the original Arts PROPEL model. Arts PROPEL evaluated portfolios in three broad categories: accomplishment in writing, use of processes and resources for writing, and development as a writer.

In the experiment I am proposing, the culminating grade attaches primarily to the student’s capacity to make and support a judgment about their own processes and development. This system is a starting point, not a prescription with all the bugs worked out.

That does not mean the quality of each individual text becomes irrelevant. Students need actual writing to make their case in support of their judgment, and teachers should continue to teach craft, conventions, genre, purpose, and audience. But no single text receives a terminal letter grade, at least not until the student has had the opportunity to grade it in a meaningful way. The texts become evidence in a larger argument about the writer’s sense of being a writer offered by the writer.

Parents and students are accustomed to seeing grades online, so the grade book cannot simply go dark for five months. A teacher might report a weekly progress grade based on documented participation: work in progress, completed process notes, conferences, peer response, and portfolio maintenance. “Effort” alone is too private and subjective to grade fairly. The visible evidence of participation is not.

The final portfolio grade is a different indicator than what school culture is used to. It depends on the student’s demonstrated capacity to analyze writing activity, select relevant evidence from their work to illuminate and self-analyze, explain changes in how they make decisions during writing, and identify plausible next steps.

The sequence is built around three practices: collection, selection, and reflection. Note well: This system cannot be easily gamed by AI. How would AI know what satisfied a writer?

Collection

Collection of pieces of writing should be deliberately varied and substantially controlled by students, so that by the end of the term there is something worth analyzing kept in the working folio for each student’s own reasons.

The teacher can propose a range of tasks with deadlines, invite students to create their own tasks, or continue giving assignments to the whole class. The standing rule is that everyone is always working on at least one piece of writing.

Students might contribute work to access-controlled sites with authentic audiences to intensify the impact of audience on the writing community. They might write poems, letters to the editor, letters to the author of a favorite book, fictional or factual stories, reviews of books or literature or consumer products, research papers, lab reports, journal entries, or they might collect pieces from other classes. The collection should include drafts and process evidence, not merely finished products.

Dennie Palmer Wolf described the Pittsburgh portfolios as “deliberately diverse.” A student might include “pieces as diverse as journal entries, letters, poems or essays from social studies classes,” together with notes, diagrams, drafts, and final versions. Teachers were asking students to read their own progress in the ‘footprints’ of their work.”

Nothing enters the collection without a process note. Written at the time of filing rather than reconstructed months later, the note answers four questions:

1. What did you try to do here?
2. Where did you get slowed down, frustrated, stuck?
3. What surprised you about the journey to completion?
4. What would you have wanted more time or help with?

The notes are not graded for polish. They’re not graded at all, though they are recorded as evidence of engagement and participation. Their purpose is to capture difficulty or insight while it is still fresh.

A lab report might reveal confusion about how to phrase a hypothesis. A poem might document the moment a metaphor stopped working and had to be abandoned. A library paper might record the frustration of finding three sources that contradicted one another with no obvious way to adjudicate among them.

By the end of each month, the folder becomes far more than a stack of unrelated products. It is a dated record of friction: what satisfied the writer, what excited the writer, what confused the writer, what took three tries, what peer or teacher informal feedback supported the writer, what came more easily than expected, and why the ease was surprising.

This aspect could be the location for documentation of uses of AI. Of course, Arts PROPEL can’t provide concrete help, but the PROPEL premise that humans are artists as well as scientists who throughout history have privileged and safeguarded authentic human accomplishments while incorporating technology grounds the folio in the values of the Humanities. The writer, not the text, is the important agent.

Selection

Month 1: The Problem Piece

At the end of the first month, students select the piece whose process notes and history reveal the most consequential problems they confronted. These problems may have been solved, partially solved, or left unresolved.

The student describes the problem or problems, the attempts to address it or them, and the outcome. What did the writer learn from the effort? What remains uncertain?

Done in class and in community, this first selection creates opportunities to learn from peers. The teacher should offer observations and possibilities without taking away the student’s right to choose. The teacher can scaffold the writers in creating an inventory of problems writers often confront to keep in mind during future work.

Month 2: The Surprise Piece

At the end of the second month, students select the piece that most surprised them as writers. The surprise should concern a decision or strategy, not so much the subject matter, always keeping in mind the entanglements of subject matter and the pleasures, conundrums, and puzzles attending writing it up.

During this month, the teacher might concentrate on strategies such as zero drafting, showing rather than telling, rebuilding a lead, changing point of view, or restructuring an argument.

Students also revisit the problem piece from Month 1. Does it still represent the most revealing problem in the collection? Has a better example appeared? If so, the student may substitute it, explaining the change in judgment.

Month 3: The Unsatisfying Piece

At the end of the third month, students select the piece they find most unsatisfying and explore their reasons.

It may be the piece they would most like to revise and publish again. It may be a piece that looks polished but fails to achieve what the writer hoped. It may be a piece that improved after peer feedback but still exposes a persistent problem.

Students then revisit the problem and surprise pieces. Either may be replaced by something from the growing collection, but every substitution requires a written explanation.

Month 4: The Satisfying Piece

At the end of the fourth month, students select their most satisfying piece. “Satisfying” remains deliberately undefined.

The ambiguity matters. One student may value the audience response. One student may value technical control. Another may value risk. Another may be satisfied because a previously difficult process became manageable. ANother may have two or three of these plus novel reasons. The reflection asks the student to define satisfaction and to relate that definition to the unsatisfying piece selected the month before.

Once again, each earlier selection remains open to substitution. Judgment is not a one-time choice. It develops.

Reflection

Month 5: The Portfolio Argument

In the fifth month, the four selections stop being separate artifacts and become evidence in a single argument. The student must answer:

What happened to you as a writer this term, and how do you know?

The month unfolds in four stages.

Week 1: The Audit

Students place all four selected pieces side by side, together with their original process notes and the record of substitutions. They should have opportunities to discuss their selections with peers.

Did the surprise piece from Month 2 survive until Month 5, or did something later replace it? Why did the original choice once seem right? What changed?

A piece that survived four months of possible replacement carries a different kind of evidentiary weight from one that entered the portfolio last month. The writer should account for that editorial history.

Week 2: The Connective Pass

Students write short passages establishing relationships among the pieces as they were conceived, created, and published. These are not four new, isolated reflections. They are statements of connection, linking the writer and the writing in history.

How does the problem piece relate to the satisfying piece? Is the satisfying piece satisfying partly because something was solved? Does the unsatisfying piece reveal a limitation that appears elsewhere? Did a strategy that produced surprise later become a habit?

This is where “development as a writer” stops being an abstraction. The student must show the throughline.

Week 3: The Inventory and Cover Materials

Students complete a writing inventory about their history and habits during the term, independent of any single piece. The inventory could include a brief autobiographical section covering the writer across several grade levels.

They also prepare an annotated table of contents. It does not merely list the four selections. In one line per piece, it explains why each one earned its place after four months of possible substitution. This is the place for the writer to note in captions the connections between the selections.

The student is no longer explaining a selection only to themselves. They are preparing a reader to encounter the portfolio as an argument.

Week 4: The Final Reflective Essay

The final essay answers the central question: What happened to you as a writer this term, and how do you know?

Students use the four selected pieces as evidence. If useful, they may also draw on process notes and discarded pieces that did not survive the selection process.

The rubric evaluates:

· specificity and relevance of evidence;
· clarity and coherence of the developmental account;
· plausibility of the argument;
· willingness to acknowledge contradiction or uncertainty;
· evidence of changing judgment;
· and the student’s ability to identify realistic next steps.

Students are not rewarded for selecting conventionally polished work. They are rewarded for reasoning precisely about what their work reveals.

A student who selects a technically weak piece and gives a careful, self-aware account of why it matters may outscore a student who selects polished work and can say only, “I like it.”

That inverts the usual incentive. It turns the tables and disconnects the cables.

What Gets Submitted

The final portfolio contains:

· four selected pieces;
· the original process notes;
· the record of substitutions and the reasons for them;
· the writing inventory;
· the annotated table of contents;
· and the final reflective essay.

The portfolio is the accumulated, dated residue of five months spent learning to read one’s own footprints, finally assembled into a case the student must make and support. This is as near as I can come to a suggestion for thinking about what it could mean for assessment to become an episode of learning rather than an occasion for correction.

What the Grade Measures

The portfolio grade measures the capacity to look back on one’s work and say something specific, meaningful, and defensible about it and what it represents about the writer.

The final summative grade or rating is delayed, but when it arrives, it comes up against the writer’s informed and reasoned judgment. If there is major disagreement, the teacher and the student have something real to negotiate.

This does more than delay a grade until comments have been read. It restructures what the grade measures. Reading one’s own work carefully is no longer optional homework completed after the real judgment has arrived. It is the assessed work.

The teacher can still provide feedback generously. But the feedback no longer has to function as a running justification for a series of grades. It can provoke, question, notice, suggest, and challenge. The student remains responsible for deciding what the accumulated evidence means.

This model will not fit every reporting system without adaptation. A school may require separate evidence of accomplishment against writing standards. If so, the teacher can report accomplishment, process, and reflective judgment as distinct dimensions rather than collapsing them into a grade on every paper.

The important principle is that the student’s judgment must remain consequential. Self-assessment cannot be a ceremonial preface to the teacher’s real decision.

Inviting Parents to Read Differently

Kathryn Howard, a teacher at Reizenstein Middle School in Pittsburgh, extended portfolio assessment from school into the home. In the spring, she sent writing folders home and invited parents to read everything: drafts, teacher comments, and the student’s own reflections.

Parents responded to a short set of questions:

· Which piece tells you the most about your child’s writing?
· What does it tell you?
· What strengths do you see?
· What needs to be addressed in your child’s development as a writer?

Howard explained that the practice grew naturally from classroom life:

“In our classroom, students constantly talk about writing. They discuss and establish their own criteria of and standards for good writing.... It seemed natural, then, in this context to create an opportunity for students and their parents to talk about writing as well, to show students what their parents value about writing using conversation and written reflection.”

In the first year, thirty of thirty-three students participated. Every student participated in subsequent years. Twenty-four of the thirty participating parents wrote positive comments about the activity, and none expressed displeasure.

More important than the enthusiasm was the change in parental attention. An initial focus on accomplishment, “Is this piece good?”, expanded to include process, growth, and patterns visible only across a body of work. Howard and Paul LeMahieu concluded that the activity “served an important function in local accountability by educating parents about current methods of writing instruction.”

Parent participation should be invited rather than used as a condition of the student’s grade. Families differ in time, language, access, and comfort with school writing. But the larger possibility matters.

A grade gives a parent a hollow symbol. A protocol like Howard’s teaches a parent to read as the student has spent five months learning to read: for evidence, for pattern, and for the difference between a piece that is merely finished and one that means something in the development of its writer.

Close the Handback Loop

A student trained to wait for a verdict has little reason to collect, select, self-assess, regulate, decide, reflect, and reconsider. The assignment is to grasp the teacher’s expectations and do the deed. The rubric exists. The mystery is gone.

Then the judgment arrives. The grade tells the student what to think and feel about the work. The paper drifts quickly into history, but the grade remains. The teacher’s comments, however carefully written, arrive in support of the judgment. The student is expected to acquiesce and move on.

This PROPEL portfolio interrupts that sequence. It keeps the question open long enough for the student to answer first.

This proposal also belongs to the tradition of teacher research. It’s a design to try, document, study, criticize, and revise. What do students do when their judgments acquire real consequences? What kinds of evidence do they learn to notice? Where do they resist? When does the language of reflection become authentic, and when does it become another school performance?

Teachers are in the best position to investigate those questions with the students in front of them.

The handback loop begins to close when the person waiting at home stops asking, “What grade did you get?” and starts asking, “How is the writing going?”

Sources and Further Reading

Camp, R. (1992). “Portfolio Reflections in Middle and Secondary School Classrooms.” In K. B. Yancey, ed., Portfolios in the Writing Classroom: An Introduction, 61–79.

Cochran-Smith, M., & Lytle, S. L. (1999). “The Teacher Research Movement: A Decade Later.” Educational Researcher, 28(7), 15–25.

Cochran-Smith, M., & Lytle, S. L. (1999). “Relationships of Knowledge and Practice: Teacher Learning in Communities.” Review of Research in Education, 24, 249–305.

Fishman, S. M., & McCarthy, L. (2000). Unplayed Tapes: A Personal History of Collaborative Teacher Research. National Council of Teachers of English and Teachers College Press. ERIC full text

Goswami, D., & Stillman, P. R., eds. (1987). Reclaiming the Classroom: Teacher Research as an Agency for Change. Boynton/Cook.

Howard, K., & LeMahieu, P. (1995). “Parents as Assessors of Student Writing: Enlarging the Community of Learners.” Teaching and Change, 2(4), 392–414.

Louden, K. (2017). “Delaying the Grade: How to Get Students to Read Feedback.” Cult of Pedagogy. Read the article

Luther, M. (2025). “Before You Return Feedback to Students.” The Broken Copier. Read the article

Murphy, S., & Underwood, T. (2000). Portfolio Practices: Lessons from Schools, Districts, and States. Christopher-Gordon.

Seidel, S., Wolf, D. P., Zessoules, R., Camp, R., & Winner, E. (1995). Arts PROPEL: A Handbook for Imaginative Writing. Harvard Project Zero and Educational Testing Service. Project Zero resource

Wolf, D. P. (1989). “Portfolio Assessment: Sampling Student Work.” Educational Leadership, 46(7), 35–39.

Competency-Based Assessment in a Hurry: An AI-Based Alternative to Traditional Grades and the Carnegie Unit

Terry Underwood, PhD — Thu, 27 Aug 2026 03:25:06 GMT

ETS laid off 757 California workers at the end of 2025 as its state testing contract wound down. The company disclosed the cuts in a state filing, saying it would stop providing constructed-response scoring for several California programs effective December 31. The affected rater positions were tied to ETS’s Sacramento office, though most of those employees worked remotely. Measurement Inc. would be taking over the scoring work.

This layoff came in a series of business problems for ETS. Founded in 1947 out of a College Board partnership, it once dominated U.S. standardized testing, but it lost its SAT administration contract and offered voluntary buyouts last year as universities moved away from standardized exams, and digital competitors ate into its market.

On April 17, 2023, the Carnegie Foundation and ETS announced a partnership to build tools for assessing “durable” skills — collaboration, communication, and critical thinking—that employers said matter, yet are barely legible in transcripts of course grades. Timothy Knowles, Carnegie’s president, said the resulting measures could be better indicators of a student’s future success than the Carnegie unit, the credit-hour standard his own foundation had introduced in 1906.

The two organizations named the effort “Skills for the Future” and began adding state partners the following spring: Indiana, Nevada, North Carolina, Rhode Island, and Wisconsin joined in 2024, meeting with ETS and Carnegie staff in California that April to plan pilots for January, 2025, according to The 74.

That May, ETS folded in the Mastery Transcript Consortium, the nonprofit behind the skills-based transcript format several of those same states would use, as a wholly owned subsidiary, according to ETS’s release.

The first Skills Progressions includes the definitions of what “collaboration,” “communication,” and “critical thinking” mean at each stage of development. They were published in January, 2026, nearly three years after the initial announcement, when Carnegie and ETS jointly released them. Knowles’s original list of four — reason, create, collaborate, persevere — had been narrowed by then to three.

Knowles, president of Carnegie, described ETS as becoming a “data insights” company, no longer a company dedicated solely to making valid and reliable tests at scale. He suggested that “durable skills” assessment could become useful not just in middle and high school, but in the university analogous to AP exams. The tools could be better indicators of a student’s future success than the traditional Carnegie unit, or credit hour, the measure first introduced in 1906 linking achievement with seat time.

An interesting and potentially significant conceptual innovation, the most profound problem with this design in my estimation is not the retreat from disciplinary organization, from courses packaged as credit hours, and from what almost everyone agrees is a failed grading system. It is how the planned use of AI puts opportunities to learn for both students and teachers at risk. This loss of opportunity could weaken the same durable skills the design attempts to measure.

What “Durable Skills” Means and How a Progression Works

The Skills Progressions document itself defines each of the three durable skills in a single dense sentence, then spends the rest of its pages unpacking that sentence into observable behavior. Collaboration is “the ability to work toward shared goals through engaging in effective communication, coordinating group efforts to integrate diverse perspectives while fostering interpersonal relationships,” involving “exchanging ideas openly, regulating group activities, and prioritizing building trust and respect to enable groups to leverage individual strengths while maintaining focus on collective success” (ETS/Carnegie Foundation, Skills Progressions, Vol. 1, p. 8).

Communication is “the ability to share and interpret messages that convey ideas, thoughts, opinions, knowledge and data with clarity and purpose using various modalities such as verbal, nonverbal..., written, visual and listening, while adapting to diverse contexts and audiences” (p. 13). Critical Thinking “involves the skill and disposition to actively seek and evaluate information and construct evidence-based arguments to reach well-founded conclusions or informed decisions, including recognizing and applying sound logical reasoning” (p. 23).

These are not test constructs in the ordinary sense. They read more like outcomes one might find in a portfolio assessment system. Each skill is organized as a hierarchy: a durable skill breaks into several subskills, each subskill breaks into indicators, and each indicator is described at four developmental levels — Exploring, Analyzing, Integrating, and Extending — that together constitute the “progression.”

Collaboration has three subskills and seven indicators; Communication has four subskills and 15 indicators; Critical Thinking has four subskills and nine indicators — 31 indicators in total across the three skills, each with its own four-level description (p. 6–26).

A progression, concretely, looks like this. Under Collaboration, Subskill 1 — “Engage with ideas through intentional communication in service of shared goals” — contains Indicator 1.1A, which describes a single narrow behavior (asking questions to understand group members’ ideas) at four rising levels of sophistication.

At Exploring, a student “asks questions to better understand a topic, group task or shared goal.” At Analyzing, they ask “to better understand group members’ ideas.” At Integrating, they ask “to understand diverse ideas and perspectives, both in what group members think and why they hold certain views.” At Extending, they ask “questions that help the group understand and use diverse ideas and perspectives, including those from personal or cultural experiences (e.g., to create a combination of the group’s ideas)” (p. 8).

The same logic repeats across all 31 indicators: not a pass/fail judgment on whether a skill is present, but a claim about which of four qualitatively different versions of the behavior an artifact demonstrates.

The document is explicit that this structure describes development and is not a test to be administered on demand: “These Progressions describe the knowledge, abilities and behaviors associated with each skill holistically. They do not prescribe a single pathway or narrow set of observable behaviors. Evidence of student growth may take many forms across contexts, modalities and communication styles” (p. 6).

As I consider this design, I am transported back in time to the era of large-scale portfolios like the system developed in Vermont in 1990 or in Kentucky in 1992 or in Pittsburgh in 1989. These systems failed for a variety of reasons, not the least of which were problems achieving reliability in scoring, Pittsburgh the sole exception.

ETS and Carnegie seem to have figured out how to assess complex, fluid data reliably. To achieve this elusive psychometric property, they propose using AI.

What Gets Uploaded

Neither the Progressions document nor ETS’s public materials specify a required artifact list, but the framework’s own language about where evidence comes from, combined with the indicator descriptions themselves, points to what would plausibly qualify.

A group project report or a shared slide deck could surface Collaboration Indicator 2.1 (”flexibly establish and contribute to group ideas and team norms”). A recorded or transcribed classroom presentation could surface Communication Subskill 2, “adapt communication styles based on diverse audiences and contexts” — the indicators explicitly reference “presentations,” “videos,” and “audio recordings” as example formats (p. 14).

A research paper or a debate assignment with cited sources and addressed counterarguments maps directly onto Critical Thinking’s “Information Seeking,” “Information Analysis,” and “Argument Generation” subskills, which name “essays, presentations, videos, audio recordings” as illustrative artifact types (pp. 23–26).

A reflective journal entry or an exit ticket where a student explains their own reasoning could serve as the “student reflection” category ETS names separately from artifacts and direct assessments (ETS, What Durable Skills Measurement Looks Like). And because the framework explicitly extends evidence-gathering “wherever it takes place — in school, in extracurricular activities like sports or theater, in internships and apprenticeships” (ETS, SFF two-pager), a coach’s note on team leadership or a supervisor’s evaluation from a part-time job could, in principle, also become an uploaded artifact.

What none of these examples resolve is the question the prior section already raised: once uploaded, what threshold of evidence in an essay or a recorded discussion causes the AI to propose “Integrating” over “Analyzing” for a given indicator, and that mapping, indicator by indicator, is not published anywhere in the documents ETS and Carnegie have released.

I don’t think qualitative threshholds can be provided, which are invaluable to teachers during classroom interactions, giving feedback, commenting on the fly, given the nature of AI. AI decisions will compete with human teacher judgments in a strange world where AI will either be privileged or generate bafflement among teachers and learners alike. Why did I get this score? can never be answered.

Fear and Trembling among Onlookers

That in my mind is frightening. If human readers—teachers who taught writing in Vermont and Kentucky and understood the history and culture of the schools—couldn’t achieve acceptable interrater reliability for accountability purposes when assessing collections of written evidence in paper folders, on what basis would anyone be compelled to see AI scores as reliable—not to mention valid?

Is it possible that humans simply are incapable of responding in robotic fashion to children’s work as collections of discrete artifacts that form regular quantitative patterns which fit inside larger skill progression patterns with thousands of indicators that co-occur in statistical relationships totally separate from meaning, sense, emotion, and longterm memory of unique experiences? I find it much easier to understand how AI might be useful for scoring non-complex textual messages like taking a message on a phone or writing a five-paragraph essay than how AI can assess how well a student collaborates with other children. Maybe it’s just me.

That’s the question I pose, one I’d love to hear from you about.

Traffic Lights Learning to Think: Lessons for Public Schools

Terry Underwood, PhD — Sat, 08 Aug 2026 19:29:51 GMT

NOTE TO READERS:

This essay is ostensibly about traffic lights. I found the topic interesting to learn about and intend to mine it as a theoretical analogy in work I’m doing on anthropomorphism, and it can be read theoretically that way, but it can be read as a simple tale of two realities on the road to a destination, one coordinated and planned, one chaotic and dangerous, a case study in how one public system tried to meet a new technology for the benefit of society.

Traffic engineers did not wait for permission to use AI. They pointed cameras at intersections, directed models to count cars, and let the results argue for themselves, imperfectly, unevenly, but seriously with authentic intent to work with, not against, technological development. There was never any worry that AI would replace system engineers. When Google’s models got a timing change wrong, an engineer, the human in the loop, reversed it and moved on. The failure mode was ordinary and correctable.

Public education has taken the opposite posture. Where traffic departments have spent years building toward the capacity for full use AI’s affordances, however slowly and however underfunded, school policy has spent the past three or four years building against AI, banning, shaming, disclaiming, litigating, treating the technology itself as the threat when the real threat is lack of governance and resources.

Both traffic lights and public school systems suffer from the same weaknesses: fragmented governance, chronic underinvestment, and a federal government stepping back from the infrastructure it committed to just a few years ago. But only one is trying to solve its version of the AI problem through knowledge making. The other is avoiding the problem. That difference is the argument this essay is making.

»»»»»»»»

The traffic light has been a profoundly indifferent social expressive device for most of its history. For most of its existence, it didn’t know anyone in a vehicle was even there at the intersection, waiting to pass through to the next block while precious time ticked away. It couldn’t tell an ambulance from a pick-up truck.

How could it? It ran on a timer, after all, cycling through red and green, then red yellow green, ad infinitum. And participation was mandatory with quite strict potential consequences for unruly humans. After it acquired cameras, it could know you had been there, enough knowledge to convict you in a court of law.

At one time traffic signals were human beings, not sparkly lights on poles. Officers stood at intersections with flags, whistles, and, later, hand-cranked signs, directing horse-drawn carriages and then automobiles, using their own good sense and talents.

The first attempt at a machine-human hybrid wasn’t really automation at all, but a gas-lit signal installed in London outside the House of Parliament in 1868 at which a policeman was stationed to turn a crank which rotated the gas lamps and thereby the color of the signal.

There always seems to be a price associated with progress. The device exploded within a month and injured or killed the policeman operating it. While some historical popular accounts state he was killed instantly and others report severe facial disfigurement/injury, the tragedy halted gas-lit traffic signal experiments across Britain for over 50 years.

Imagine if the Wright brothers had crashed and burned.

Almost half a century later in 1914, Cleveland’s two-color system used an audible buzzer to alert drivers before that perilous moment when the light changed between red and green. The American Traffic Signal Company installed James Hoge’s electric signal system, including this buzzer to announce the changing of the colors, at Euclid Avenue and East 105th Street, perhaps the forerunner of the bells between classes in school.

In 1920, William Potts, a Detroit police officer, designed the first three-color four-way traffic signal. New York followed a different path. Starting February, 1920, Fifth Avenue got a string of towers where a traffic officer sat inside a booth and switched red, yellow, and green lights by hand cued by a whistle from officers stationed further down the avenue.

The first really important innovation was the inductive loop, a wire coil embedded in the pavement to detect the mass of a waiting car. During the 1960s and 70s vehicle-actuated signals gave intersections a crude capacity to sense presence, though not flow. By the 1980s, cities across the world began deploying centralized adaptive systems using networks of sensors and early computer models to adjust timing across whole corridors in close to real time.

These systems worked, but they were expensive to install and maintain, and most cities never adopted them, an underinvestment still visible today. Only a small fraction of American traffic signals, generally estimated at four to five percent, operate in any form of adaptive mode. The rest run on the same basic logic as Cleveland’s 1914 box, just with better bulbs.

What has changed developmentally in the last several years is not the ambition but the toolset, one with amazing potentials, which is facing ideological and political obstacles inherent in the manic separation of responsibility for shared projects between federal and state driving Washington DC now.

The toolset is enhanced. Cameras, radar, and thermal imaging can feed live video into processing hardware mounted at the intersection itself, and machine learning models trained to recognize vehicles, cyclists, and pedestrians can do far more than a human police officer could ever hope to do.

A modern smart intersection doesn’t wait for a car to roll on the buried scale and signal to be acknowledged. It observes the street, counts the queue, and decides, cycle by cycle, how long each phase needs to run. If a cross street is empty, the red light on the main road cuts short instead of holding a driver in an absurdly immobile state.

Google’s Project Green Light analyzes anonymized driving data from Maps across an entire city, looking for intersections where small adjustments to signal timing would reduce stop-and-go driving along a whole corridor. It needs no new hardware, no cameras, no buried sensors, just access to the data cities already generate through the phones in every dashboard mount.

Boston and Seattle have both successfully used it to retime dozens of intersections. Seattle’s transportation department had to reverse one AI-Green Light recommendation after it produced no real benefit, a reminder that these systems still function as advisors to human engineers rather than autonomous authorities.

Emergency and transit pre-emption of the light has layered intelligence onto infrastructure that already existed. Ambulances and fire trucks have been using optical and radio-frequency emitters to trigger green lights for decades. What is new is the coherence, the preconditioned coordination, the potential to prepare traffic for an emergency long before the emergency arrives in rearview mirrors.

Cloud-based systems track an emergency vehicle’s GPS position continuously as it moves, adjusting signal timing at every intersection along its projected route before it arrives, building what amounts to a rolling green corridor rather than a single locally controlled light. The gain is measured in minutes, and in an ambulance run, minutes often make the difference.

Human lives are saved.

Pilot programs are making their way into ordinary use. For example, weather-responsive signal timing, a category the Federal Highway Administration has studied for years, links ambient sensors to the signal controller so that a wet or icy road triggers longer yellow intervals and wider clearance gaps, holding one red light constant for a few seconds more, stretching a corresponding yellow light out for a few more seconds..

Vehicle-to-infrastructure communication extends to ordinary drivers, too, most visibly through Audi’s Traffic Light Information system, which has been broadcasting real-time signal data to equipped cars since 2016 and now offers speed recommendations calculated to let a driver catch a green light rather than stop for a red, riding what the company calls the green wave.

And AI vision systems are increasingly being used to track not just vehicles but the people crossing in front of them, extending a walk signal automatically when a pedestrian’s measured speed suggests they will not clear the crosswalk in the standard interval, a capability with obvious value for elderly or disabled pedestrians, the group least served by a fixed clock.

Yet dumb intersections still exist everywhere in America with little conscious, focused, societal support for changing them. We seem to inhabit a period when our government is more interested in destroying infrastructure in Iran than in addressing infrastructural development in this society.

Why? In addition to arrogance toward and disregard for ordinary citizens, this government is organized to satisfy the demands of one person: Donald Trump.

Outfitting a single intersection with high-resolution cameras, edge-computing hardware, and a reliable network connection costs anywhere from twenty thousand dollars for a modest camera-based upgrade to well over a hundred thousand for a full adaptive system with radar and dedicated processing. In contrast, one AGM-114 Hellfire Missile used in a drone or helicopter attack in Iran costs $150.000.

A mid-sized city may have hundreds or thousands of intersections. Nashville’s plan to modernize roughly six hundred signals citywide carries a price tag of a hundred and fifty-eight million dollars. Multiplied across the roughly three hundred thousand signalized intersections in the United States, it’s clear why most municipal budgets fall well short of full modernization, federal grants and vendor financing notwithstanding.

USDOT announced recently that under the Trump administration no new notices of funding opportunity (NOFOs) will be issued for future SMART grant cycles. While existing Stage 1 and Stage 2 grant agreements are being honored, cities that had not yet secured a SMART grant can no longer rely on this dedicated competitive federal pipeline for signal modernization.

Cities must now compete at the state level for formula dollars rather than receiving direct federal funding. State DOTs generally prioritize road resurfacing, bridge repair, and highway capacity expansions over municipal urban signal digitizations unless state-specific grant programs exist (such as Tennessee's state-funded Traffic Signal Modernization Program).

The last obstacle is the least ideological on the surface and the hardest to fix: ownership of the problem. A single arterial road can cross through the jurisdiction of a city traffic department, a county public works office, a state department of transportation, and a regional transit authority, each running its own signal timing plans on its own equipment on its own budget cycle.

Coordinating a green wave across that patchwork is less a technology problem than a structural problem, and no amount of good will solves a dispute between agencies over who pays scarce dollars for the fiber connection between their two halves of the same street.

Perhaps AI might lend a hand in rationalizing and humanizing how we as a society share our resources equitably and wisely.

Isn’t it ironic that traffic lights are becoming smarter and smarter while the humans who need them seem to be becoming dumber and dumber? The traffic light has been learning to see and hear and weigh to take in as well as express signal information to assist human beings driving in heavy traffic at rush hour.

The one space the traffic light has not learned to see? It has not yet learned to see across a jurisdiction line.

Reference List

The Blue Book Exercise: Only a Human Can Change Its Mind

Terry Underwood, PhD — Thu, 06 Aug 2026 19:28:07 GMT

As writing teachers return to the classroom after summer break, many are wrestling with how to handle AI. The following is a work of narrative fiction about one teacher's radical, flawed, deeply felt defense of human thought and language. I salute every writing teacher actively working to make sense of this new reality by doing the best they can with what they have in their current circumstances. Ignoring it isn’t realistic.

An imaginary English teacher, fresh from a summer break, leans over a student's shoulder seated in a desk at the back of the classroom, appearing to read inky black words emerging atop thin lines on a page in a blue book. It matters not whether he reads the words. Pausing at the rear of the room, displaying curiosity, he is being performative. Everyone knows it or soon will. Everyone will soon expect it.

He straightens himself, scans a gaggle of bobbing, writing heads hunched over desks, smiling appreciatively.

"Five more minutes. Think about how your reader will feel," he says, loudly enough to travel but not disturb.

Blue books are blessed in this classroom. Inky words on paper are valorized in a strange compression of drafting and revision, the two crucial parts of the writer's process, in his humble opinion. This imaginary writing teacher is convinced AI companies are trying to destroy the writing classroom, wittingly or not. He sees himself as a guardian of the tower and makes no apologies about it.

He began the past summer totally exhausted from battling AI. He promised himself that he would come up with a way to move against this rapidly growing monstrosity, and he did. He designed and planned a classroom routine to counter the artificial assault on his work — a journaling routine he called the blue book strategy.

Ten minutes twice a week in class. Students would write in their blue books using a black ink pen, using one page, on a topic from their assigned literature. When time was up, he would collect the blue books. At the end of the day, a parent volunteer would photograph each new page and upload them to a safe and secure website.

The school's technology administrator had created a digital workflow whereby a specially trained small bot purchased for the Science Department received the single-page blue book photos and applied his detailed rubric, producing an overall point value, individual reports with feedback for students, and a class-by-class analysis. He taught two sections of senior writing and decided to try the strategy out in both sections.

To be honest, it was a painful time for him. For over twenty-five years he had perfected his approach to his writing classroom. He believed in his power to make a difference in his students’ lives. His approach had become almost automatic, and it worked.

Over the years he enjoyed reading the occasional beauty he came across in the essay submissions. His superpower had been teaching coherence. That was before the apocalypse.

Writing the persuasive essay was his signature assignment. Stating a clear thesis announcing the argument, organizing a clear exposition of core premises and their assumptions, providing evidence to warrant claims and inferences, signposting the argumentative mechanism — it was all music to his ears.

Until the language machine showed up, manufacturing persuasive texts on any imaginable thesis almost like magic.

At the heart of his blue book strategy design was a compression and intermingling of drafting and revision. He was indebted to the Flower and Hayes model of writing from way back in the day, something that stuck from his teacher preparation program in the late 1990s, the writerly planning, the translation from thought to language which came to be called drafting, the revision to refine and shape.

How could he fit the core of the writing process into ten minutes twice a week?

It came to him in the middle of a hot, sleepless night in July. He got out of bed and wrote it down. Writers must be certain of their next word before writing it, or nearly so, he wrote that night, though he felt a bit uncomfortable about it. Could any writer ever be certain about every word written?

They're going to have to be, he thought.

Crossing words out in the blue book in this classroom would have a price. Cross-outs would be penalized on the 10-point rubric: a word would cost half of a point, a phrase struck seven-tenths, a clause or sentence nine-tenths. But points could be regained.

If the replacement made a significant contribution, if a crossed-out sentence was followed by a genuine turn in thinking, the deduction disappeared. The rubric wouldn't forbid revision during drafting. It would put a price on it unless it worked. He knew this would be a subjective judgment, but so be it.

In this legendary teacher's classroom, writing would be thinking; thinking, finding words; finding words and phrases, weighing costs against an uncertain gain in audience approach. Vygotsky on steroids, with a ledger attached. His logic was impeccable.

Students would face constraints much like a language bot's, but much more sophisticated. Words rarely arrive one at a time in a writer's mind. They come in pieces, sticky, sometimes a whole sentence formed before the pen ever touches the page.

This burstiness is characteristic of language models too, which generate in probabilistic clusters and sometimes pause in milliseconds to revise mid-stream before generating, weighing a token against its likely alternatives. Both writer and bot must decide, constantly, on the spot, whether a potential discard is worth its cost if included. But only a human can change its mind.

Midway through each ten-minute exercise, he would prompt his writers to do something bots could never do: exercise the human capacity for authentic communion with fellow humans, which must never be lost.

"Think about how the reader will feel."

Whether any or all students do such thinking is impossible for him to know for certain. Writing itself isn't naturally like reading someone else's mind; it's more like reading one's own mind. The student might draw on an accumulated history of having been a reader, bored by a dense paragraph, confused by a tangle of ideas, delighted when an irony or a figurative word makes sense. The teacher's prompt asks the student to run a simulation built from memory in medias res and go from there.

He assumes that instructing his students to keep one eye on a reader and the other on the emerging text defines the difference being human versus being a bot during writing activity. A full ban on revision would be a problem because it would put higher stakes on commitment: write it once, live with it.

Such prose would likely incline toward writer-based prose, understandable to the writer but to no one else, tantamount to the hedged and hallucinated bot slop of the most egregious sort which he simply could not tolerate.

His cross-out valuation system, coupled with the mid-exercise prompt to think of a reader, he believes, activates writerly judgment in the act to decide whether the anticipated gain of a better sentence is worth a real, quantified cost.

The student must conjecture, in real time, how a stranger might respond to two different versions of the same passage, then decide if that estimate justifies the deduction. Nine-tenths of a point can bite in tight circumstances. The stakes make the anticipation concrete rather than performative.

It would be absurd to put a language model under the same rule, a fact supporting the judgment that humans and AI are nothing alike. Unlike human writers, standard language models generate text strictly token by token in a single forward pass—never pausing mid-sentence to revise or rethink unless guided by complex search algorithms like Monte Carlo Tree Search or deliberate multi-step prompting.

Ask it to write toward a reader's felt response, and it can produce sentences that satisfy the assignment by any surface measure. Graded blind, the rubric may not distinguish the model's page from the strongest student's blue book exercise, except for one thing.

Bots never have point deductions for revision.

When the student weighs a cross-out against its cost, the value side is a bet on an actual rhetorical future. The word there in ink could be, if not wrong, then misleading, but it could be serviceable. Students mistake themselves for their readers constantly and misjudge these spots.

When a language model discards a token or a phrase mid-generation, the "value" side of its calculation is a probability distribution in the form of an estimate of what tends to read well in comparable language contexts, built from an enormous corpus of prior text. Betting on a human reader's response is irrelevant. It's estimating a pattern.

The hardest case, the one his rubric was built to reward, is the clausal cross-out followed by a brilliant turn. A student crosses out a blue book sentence already committed, two sentences, and what follows is much, much better because the act of striking it changed what the student understood they were trying to say. High risk, but potentially high gain. Bots simply do not behave this way.

This successful revision recovers its cost because it is evidence that the writer was changed by and owned their infelicity or confusion or oversimplification, mid-sentence, and the next sentence came from a slightly different writer than the one who wrote the first.

Can a bot's discard-and-resample step do this? No. Not even close. Not without precise, human-designed instruction, and even then the results are often mediocre, if still sometimes helpful.

It can select a better phrase. It can, under prompting, produce something that reads as a shift in thinking. What's not settled by an analogy is whether discarding a token ever constitutes the model learning something mid-generation, the way a human writer's understanding can shift dramatically in the half-second between crossing out a clause and finding the one that replaces it. A model's rejection updates nothing about what it understands. It only samples again.

The writer's rejection can update the writer.

He is quite uncomfortable with his decision to use AI to score the blue book pages.

The irony is real. A teacher who built his blue book strategy to protect a capacity machines lack then hands its grading to the machine. The bot certifying whether a student was changed mid-sentence cannot itself be changed by anything it reads. He would have to trust it.

Weigh that against what he's facing: two sections, twice a week, every cross-out and recovered point, graded by hand. He wrote the rubric; he decided what counts as a genuine turn. The bot only does the arithmetic at scale, freeing his limited attention for what still needs a human eye. He hasn't outsourced the care, just the count.

Whether a machine can tell a cosmetic cross-out from a genuine one is a fair question, and probably his design's weakest point. But grading by exhaustion, or not grading closely enough to make the stakes real, would undermine the rubric just as surely.

He'll keep collecting blue books, believing he guards a tower. He may not be wrong to let a bot help as long as he never mistakes its number for what the number is trying to point to, a young person, mid-sentence, not yet knowing who they're about to become as they write their future.

Non-Ordinary Reality in Schooling: When Time and Space Rupture

Terry Underwood, PhD — Sun, 02 Aug 2026 23:24:42 GMT

To argue that a complete dissolution of a U.S. national commitment to public education has already occurred under the Trump administration is to argue that a reduction in federal resources for state educational systems, a deliberate intensification of ideological and economic polarization among public stakeholders, and the return of racial segregation in urban areas and in the rural South means total abandonment.1

Admittedly shaded by prior experiences working within the system in national projects designed to improve meaningful communication between and among schools and districts across state lines from K12 schools to community colleges and beyond, from my perspective, total abandonment has in fact occurred.

Unfortunately, education is viewed as an isolated state capacity like the erasure of similar capacities in different domains. The same federal government that starves public schools has gutted the public health infrastructure, dismantled child care subsidies, hollowed out child labor enforcement, and refused to build any legal architecture for AI’s arrival in classrooms.

We are no longer in Kansas, Toto. We are in Texas where church and state blend seamlessly. States have been fully emancipated to work their solitary will on their people, at least in schooling. Non-ordinary reality in the aftermath of rupture occurs at all levels of Bronfenbrenner’s ecological model, as intuition tells us and this essay will illustrate, but the proximal impacts of rupture are quickly legible to participants in schooling.

Rupture has less to do with time, more to do with timing. When gradually ends and real change begins, there we see rupture, and we shift from living in an ordinary to living in a non-ordinary reality.

According to the ecological model, microsystems and mesosystems include people embedded and participating in a setting, while exosystems include people who have an impact on micro-meso systems but are never present in that system.

The Supreme Court, for example, is an exosystem. Few of us will ever participate in it as a microsystem, but all of us are impacted by it. Non-ordinary reality at the micro-meso level emerges from the exosystemic level in invisible and widely varying ways, for example, the overnight change from a Constitutional woman’s right to choose to a state-by-state arrangement. Please have a look at the footnote for a prompt if you would like to hold a chat about these systems of social organization.

Deception in The Chronosphere

One could argue that our federal commitment to equity in the institution of the public school has never been sincere, rather a culture of lip service. The federal office of education wasn’t created until just after the Civil War when a number of false mechanisms were announced but not enforced, e.g., giving freed slaves the right to vote. The office itself wasn’t robust; rather, it gestured toward federal oversight.

We had a historical hiccup in 1954 when the Supreme Court overruled the assumptions of separate but equal schools, an overnight pivot from ordinary to non-ordinary reality with no specified approach to living in the new reality, bringing about a failed struggle over the next few decades to integrate public schools racially.

But regarding language and literacy, Brown v. Board was not about silencing the White listening subject (Flores and Rosa, 2019); it was about moving black and white bodies into the same classrooms and nothing more. In fact, the learning problem for excluded children, who are largely minority, poor or working class, lies not in how they speak and write, but in how the White listening subject hears them.

A child's utterance is a locution, i.e., what is actually said. What the White listening subject assigns it to mean, its illocutionary force, is a separate act, performed by the listener, not the speaker. Raciolinguistic ideology materializes in that gap: the request from an excluded child heard as defiance, the question heard as disorder, the correct answer heard as luck.

By 2007, the Supreme Court hammered the final nail in the coffin of Brown v. Board. Racism had ended, so said Justice John Roberts. There was nothing left to do. The White listening subject was dead or at least irrelevant, he opined, melted into a puddle like the wicked witch in Oz. Roberts need not be a racist to call the White listening subject’s time of death, though he was color blind.

If you believe that listening subject’s death, you believe the Civil War ended at Appomattox. Never mind that Lee had the authority to surrender only his own army, not to surrender the entire Confederacy, which never formally surrendered. The war’s end with the surrender of Lee’s sword is a political performance pantomimed in history textbooks, a deception that did nothing to defang the violent racists who created unending inequity for the freedmen and women in the years following with the emergence of Jim Crow.

Literacy and Surrender

Literacy crises have the same deceptive intentions that symbolic surrenders do. They get declared, not resolved. They promise action, but they don’t follow through. They are full of sound and fury but don’t mean much when the noise dies down. They represent frustration made manifest in calls for accountability without holding everyone, lawmakers and politicians included, responsible. They soothe the conscience of the nation while changing nothing in the inequitable culture of schooling.

When the country needed soldiers who could read maintenance manuals to service military weaponry to fight in WWII, there was a crisis in schooling. When the Russians launched Sputnik in 1957 in a challenge to our STEM superiority, there was a crisis in curriculum. When demand for more sophisticated literacy too complex for easy sloganeering penetrated the economic metacognition of the politically powerful in the 1980s, suddenly we became a nation at risk—and that risk was located squarely in local schools where the standard White listening subject still lives, not where it rightfully begins in the federal macrostructure that today wants nothing to do with public schools.

Truth be told, we are a nation at risk this minute. At risk not because NAEP test scores say so. NAEP tells us nothing we didn’t already know. How many times can the public hear the same story and still act surprised by it? Not because our children don’t read the books we want them to read. They do read. They live in their own subject world, avidly reading what is important to them in their microsystems, which we do little to help them metabolize intellectually—oh, they read. What else are they doing during all of this screen time adults rail against?

And we are at risk not because they cheat on the papers we assign them to write. We don’t ask much of them. A few coherent paragraphs in the King’s English arguing a thesis or illustrating an idea. How many five-paragraph essays does it take ‘til we know that too many children have checked out?

Certainly not because they can’t sound out words, nor because they are dyslexic.

It’s Lonely At the Top

Harvard’s “Grade A-” solution is emblematic of the economic gap between the ultra wealthy and the working stiff, a circumstance pregnant with insight into our public school failings as a nation. What a problem to have, right? Such a solution is absurd at a campus like Sacramento State where I taught for almost twenty years.

The “A” in lower-division calculus means this one can walk through the door into the College of Engineering at Sacramento State. The “A” in Critical Thinking means that that one can become a registered nurse. This one, the C-, isn’t going to make it. It does no one any good to inflate the grade. And in public universities, there are legitimate C minuses. Professors with whom I worked were cognizant of the interaction between local allocation of resources and grades in prerequisite courses.

Curiously enough, according to a source though difficult to verify for accuracy (I’ve found it challenging to access statistical information from Harvard directly and would love to hear from a reader who has better luck), 60% of an incoming Harvard class comes from a public high school. I was surprised. This seemingly astounding statistic appears to contradict the argument I’m making. But the vast majority of this population went to high school in high-income suburban districts, in specialized exam schools, or in selective magnets.

Virtually all of Harvard’s students were raised in households whose annual income is significantly above the U.S. median income. The few students from the “squeezed middle” at Harvard might be affectionately referred to as the equivalent of the pleebs, a population descriptor of the forgotten denizens of urban remains in the U.S. after an apocalypse I take from Margaret Atwood’s dystopian trilogy that starts with Oryx and Crake.

How Public Schooling Fares in the Current Moment

We are living in a non-ordinary reality, a rupture filled with peril mainly for young people who will inherit the future. Distorted by international adventurism and by greed and hubris in the White House, education might appear to be the least of our worries.

In the following section, I make the case that education should be at the top of the list if the goal is to rebuild after Trump’s effort to decimate our fragile democratic macrostructure and replace it with an authoritarian regime that serves the ultra wealthy. Appropriately enough, he named an official from the World Wrestling Federation to dismantle the office of education.

Macrostructural Strategies that Sustain Inequity

Fragmentation among Schools in the States

The American macrosystem didn’t conspire to produce unequal classroom microsystems; it fragmented education from its inception as a political entity. Education is not mentioned in the Constitution as a federal obligation. Separation of school and federal government guarantees that education policy answers to fifty different chronosystems moving at fifty different speeds.

District fragmentation completes the work locally: draw a boundary line down the middle of an urban area and you have manufactured two small exosystems. Funding formulas built on local property tax then convert those boundary lines into permanent infrastructure. A classroom microsystem is not just a room with a teacher in it; it is a room built by a dedicated flow of money, and fragmentation makes sure that flow accords with the Matthew Effect.2

Separation within Schools

Where fragmentation fails, that is, where poor and affluent children do end up inside the same district, the macrosystem has a second mechanism ready. Magnet schools, like those serving Harvard, and charters serving families looking to escape an under-resourced public school reproduce fragmentation in miniature, exporting the most motivated families and the most portable resources out of the shared micro-meso system.

Remediation and tracking do the same work inside the building. Children arrive in September presorted by exosystemic activity. The sorting is presented as diagnosis rather than what it is, which is the macrosystem laundering its own inequities through the manipulative language of natural ability.

Lack of Actionable Data

Both fragmentation and internal separation depend on a third mechanism to stay invisible: the chronic thinness of the data. NAEP gives the public a single aggregate number: Mississippi’s fourth grade children improved in pronouncing written words. An aggregate number is precisely the resolution at which a macrosystem prefers to be observed.

State politicians in Mississippi can take the easy way out and mandate a mechanical, relatively inexpensive instructional solution with the appearance of solving the learning problems of an entrenched culture of poverty. They can hold up their meager NAEP success, and their public officials can cry “miracle.”

Actionable data would have to be granular enough to support micro-meso-systemic action plans responsive to equalizing the resource needs of each school in its community. The data the public receives is granular enough to reassure everyone instead that nothing has really changed—in fact, things are a tiny bit better. The Mississippi miracle is cause for celebration in the State House when nothing has changed materially for poor black children in the Delta.

Small improvements in decoding among historically impoverished districts aren’t so impressive in light of the gargantuan gap between these children and children from affluent communities.

The Japanese Contrast

In regard to finances, Japan shows what accepting a national obligation means. Its constitution makes education a binding national obligation, not a local option. Article 26 requires that children receive an education and that compulsory schooling be free. And the obligation isn’t lip service: the national government sets standards for class size and staffing and pays half the salary costs of every public-school teacher in the country, specifically to keep opportunity and quality even across regions.

Japan is not proof that centralizing funding dissolves inequity completely, only that it can relocate poverty and ignorance to newcomer populations just as the U.S. is doing. The constitutional obligation to educate binds only citizen households — “kokumin,” the people, in the language of Article 26. Foreign-national children are welcome to enroll in Japanese public schools free of charge; nothing bars the door. But no one is legally required to make sure they walk through it.

Citizen children can’t disappear from the system without someone being on the hook to ask why. Newcomer children can. A macrostructure doesn’t need a locked door to reproduce exclusion; an open door nobody is responsible for is enough. The macrosystem, if we are serious, must commit to equity in education for all children—no child left behind—not just for those fortunate enough to be considered part of the political body.

Japan inherited the American 6-3-3 grade structure during the postwar occupation, but it did not inherit the fragmentation. Control over curriculum in Japan rests largely with the national Ministry of Education, Culture, Sports, Science and Technology, and education is compulsory through the ninth grade. There is no fifty-chronosystem handoff. There is one coherent macrosystem with a macrosystemic plan to build toward 2030.

Money is distributed equitably. National, prefectural, and local governments pay almost equal shares of educational costs for students in grades one through nine. A property-tax-funded boundary line, i.e., the American mechanism that turns a school district map into a map of who gets resourced, has no equivalent lever.

Nearly 90 percent of Japanese students attend public school through ninth grade, and the public system they attend is not stratified by the accident of which side of a district line a family lives on, because the district line does not carry the funding consequence it carries here.

Teachers move by design, not by attrition. Japanese principals, like teachers, are rotated among schools every three to five years as part of a broader effort to ensure that teachers grow professionally, similar in spirit to job rotation in Japanese corporate life, and the rotation is explicit about its purpose: it allows system leaders to assign highly skilled principals to struggling schools. Nonetheless, it has an equalizing effect the system produces without formally claiming it as the goal.

In the 1990s when James Stigler and colleagues studied eighth grade math instruction in Japan, Germany, and the United States, they discovered that Japanese teachers had never heard of overhead projectors, a pedagogical tool that helps teachers teach individuals in a group, not groups of individuals. American teachers spent most of their instructional time behind overhead projectors, talking heads with an erasable marker and a box of transparencies.

The Japanese classroom blueprint provided and propagated the DNA for co-investigation and co-operation in the pursuit of mathematical ideas through developing mathematical identities and a sense of agency vis a vis participation in small-group problem solving with scarce direct instruction from the teacher. American mathematics educators have been hungry to shift from their proof-driven, algorithmic approach that is still mainstream despite its ineffectiveness—only 1 in 5 students ever reaches advanced math in America’s high school.

Centralization, however, is not innocence by any means. But the comparison is not about which system is kinder. It is about which system treats unequal outcomes as an educational design flaw to be engineered against and which system treats them as acceptable occurrences, regrettable, but naturalized, common sense. Japan’s macrosystem was built to resist the clustering of advantage. Ours was built, from the absence of a constitutional mention onward, to permit it.

Ideological Commitment to an Economic Class Structure

The U.S. education macrosystem is made of the settled belief, rarely spoken aloud, that a stratified outcome is not a failure but evidence that the system is working correctly. We live by the myth of meritocracy, academic survival of the fittest. In fact, our quantitative measures provide outcome indicators leveled by socioeconomic conditions.

It is possible for a school to be viewed as high-performing compared to its impoverished peers but dramatically underperforming when compared with economically privileged schools. We expect high-poverty conditions of school to result in impoverished learning outcomes. We’ve become comfortable with it.

Fragmentation, internal separation, and inactionable data are the mechanisms of an ideology that requires winners and losers to be legible, and a miracle in Mississippi, reported at the right resolution, tells that ideology exactly what it wants to hear: the poor children got better, if even just a tiny bit, so the structure need not change.

Indeed, what changed in Mississippi? Nothing speaking to design principles. The specifications for teaching early reading in classroom microsystems changed, betting that sounding out words is all these impoverished children need to lift themselves by their bootstraps. If they learn to sound out words, they will be equipped for life. They will learn the code of power.

Instead of celebrating improvements in ranking at the national level, statistical changes which provide little assurance that what happened was more substantial than increased attention given to reading instruction or enthusiasm among teachers who were finally getting some attention, Mississippi is reluctant to discuss the fact that indicators of systemic inequity failed to change. Poor schools still score much, much, much lower than affluent schools.

Where is the hue and cry about that finding?

Fragmentation operates without a villain to indict. It needs only a boundary line and a property tax base. Tracking functions without a bigot to excoriate. It needs only a test administered annually to children sorted and reported by September to a public that expects to see the poor score poorly and the rich score richly.

Thin data requires no cover-up, only an aggregate number and a public that believes in standardized tests as the gold standard of a meritocracy. And in Japan, where constitutional decree demanded a center of operations and an equitable distribution of resources, exclusion still runs, only along a different line: an open door nobody is responsible for walking a foreign child through.

No single actor intends any of this. That makes it a macrosystemic failure, a lack in an abstract set of values and principles that guide a society, and not a conspiracy. The belief that a stratified outcome is evidence the system is working correctly doesn’t require anyone to be conscious of it or know they hold it. It requires enough people, at enough levels, to not ask why.

Educational injustice in the U.S. is not, in this sense, a series of conscious decisions. It is a failure of vision, of leadership, of collective will, of the accumulated weight of nobody being assigned the job of paying close attention at the federal level—test scores are not close attention, surveys are little more than a snapshot—and no economic strategy to provide the human and financial resources to respond effectively. Imagine the situation in Iran if the U.S. military operated without expert oversight and with paltry financial handouts from the federal government.

But undoing the damage by building a brand new macrostructure will have to be a conscious decision at the ballot box, because inquitable structures that reproduce themselves without anyone’s conscious intent don’t dismantle themselves without collective moral intent. Those persuaded that small, incremental changes—a little more phonics, a slightly better test—are sufficient are unwittingly suborning a macrostructural commitment to sacrifice the learning of poor children for the benefit of an unjust stratified society.

References

Austin, J. L. (1962). How to do things with words. Oxford University Press.

Brown v. Board of Education, 347 U.S. 483 (1954). https://supreme.justia.com/cases/federal/us/347/483/

Bronfenbrenner, U. (1979). The ecology of human development: Experiments by nature and design. Harvard University Press. https://www.hup.harvard.edu/books/9780674224575

Can you get into an Ivy League from a public school? Ivy Coach. https://www.ivycoach.com/the-ivy-coach-blog/college-admissions/private-schools-and-ivy-league-admission/

Flores, N., & Rosa, J. (2015). Undoing appropriateness: Raciolinguistic ideologies and language diversity in education. Harvard Educational Review, 85(2), 149–171.

Flores, N. (2019). Translanguaging into raciolinguistic ideologies: A personal reflection on the legacy of Ofelia García. Journal of Multilingual Education Research, 9, Article 5. https://fordham.bepress.com/jmer/vol9/iss1/5

Hiebert, J., Gallimore, R., Garnier, H., Givvin, K. B., Hollingsworth, H., Jacobs, J., Chui, A. M., Wearne, D., Smith, M., Kersting, N., Manaster, A., Tseng, E., Etterbeek, W., Manaster, C., Gonzales, P., & Stigler, J. (2005). Mathematics teaching in the United States today (and tomorrow): Results from the TIMSS 1999 video study. Educational Evaluation and Policy Analysis, 27(2), 111–132. https://journals.sagepub.com/doi/10.3102/01623737027002111

Holland, D., Lachicotte, W., Jr., Skinner, D., & Cain, C. (1998). Identity and agency in cultural worlds. Harvard University Press.

Jones, A. (2025, March 3). WWE co-founder Linda McMahon confirmed as secretary of education. ABC News. https://abcnews.com/Politics/wwe-founder-linda-mcmahon-confirmed-secretary-education/story?id=119406018

Kariya, T. (2011). Japanese solutions to the equity and efficiency dilemma? Secondary schools, inequity and the arrival of “universal” higher education. Oxford Review of Education, 37(2), 241–266.

Kondo, A. The making of an underclass: Japan’s neglect of immigrant education. Nippon.com. https://www.nippon.com/en/in-depth/d00814/

Ministry of Education, Culture, Sports, Science and Technology (MEXT). Overview: Japanese government policies in education. https://www.mext.go.jp/en/policy/education/overview/index.htm

Ministry of Foreign Affairs of Japan. Right to education. https://www.mofa.go.jp/policy/human/econo_rep2/article13.html

National Center for Education Statistics. Highlights from the TIMSS 1999 video study of eighth-grade mathematics teaching. https://nces.ed.gov/pubs2003/timssvideo/4.asp

National Center for Education Statistics. (2024). 2024 reading state snapshot report: Mississippi, grade 4, public schools. https://nces.ed.gov/nationsreportcard/subject/publications/stt2024/pdf/2024220MS4.pdf

Parents Involved in Community Schools v. Seattle School District No. 1, 551 U.S. 701 (2007). https://supreme.justia.com/cases/federal/us/551/701/

OECD. (2015). Education policy outlook: Japan. https://www.oecd.org/content/dam/oecd/en/about/projects/edu/education-policy-outlook/379200-Japan-country-profile.pdf

Rosa, J., & Flores, N. (2017). Unsettling race and language: Toward a raciolinguistic perspective. Language in Society, 46(5), 621–647.

Seebruck, R. (2021). How teacher rotation in Japanese high schools affects the clustering of teacher quality: Comparing the distribution of teachers across public and private education sectors. Education Policy Analysis Archives, 29(91). https://doi.org/10.14507/epaa.29.5362

The Mississippi miracle: Smoke and mirrors vs. reality. Skeptic. https://www.skeptic.com/article/mississippi-miracle-smoke-and-mirrors-vs-reality/

Aldeman, C. There really was a “Mississippi Miracle” in reading. States should learn from it. The 74. https://www.the74million.org/article/there-really-was-a-mississippi-miracle-in-reading-states-should-learn-from-it/

Breazeale, G. Contextualizing Mississippi’s 2024 NAEP scores. Mississippi First. https://www.mississippifirst.org/contextualizing-mississippis-2024-naep-scores/

Mississippi Department of Education. Literacy-based promotion act. https://mdek12.org/literacy/lbpa/

Stanovich, K. E. (1986). Matthew effects in reading: Some consequences of individual differences in the acquisition of literacy. Reading Research Quarterly, 21(4), 360–407. https://warwick.ac.uk/fac/soc/cte/professionaldevelopment/trn/community/stanovich__1986.pdf

The Nation’s Report Card. Explore results for the 2024 NAEP reading assessment. https://www.nationsreportcard.gov/reports/reading/2024/g4_8/

Thomas, P. L. Media manufactures Mississippi “miracle” (again). Radical Scholarship. https://radicalscholarship.com/2025/03/05/media-manufactures-mississippi-miracle-again/

If you haven’t read much if anything about Bronfenbrenner’s (1979) ecological model of human development, the core conceptual framework I’m applying in this essay, you might prompt your preferred AI as follows (cut and paste this prompt):

“Explain the essence of Bronfenbrenner’s ecological model of human development for a reasonably intelligent adult interested in getting a foothold in a movement to nationalize U.S. public education in harmony with other developed countries. Nationalizing education is itself a chronosystem-level event: it would rewrite the macrosystem and cascade downward. Define each element in the model in a college undergraduate textbook register with examples drawn from a Dorothy Holland cultural model of the school. Do not exceed 500 words. Provide a high-quality source list.”

If you are unfamiliar with or opposed to using AI, I’ve laced contextual definitions into this text, and I have moderate confidence this essay is comprehensible on its own.

Keith Stanovich’s term for development of readers along economic fault lines: The rich get richer, the poor get poorer. The URL in the reference list provides access to the full pdf.

How Did the Laboratory Become the Arbiter of Learning in Classrooms?

Terry Underwood, PhD — Tue, 28 Jul 2026 22:35:21 GMT

“It can be said that much of developmental psychology is the science of the strange behavior of children in strange situations with strange adults for the briefest periods of time “ (Urie Bronfenbrenner, 1977).

Developmental psychology emerged in the late 19th century, intent on understanding how the infant brain develops into a mature adult mind—never mind the 20th century detour into chickens, dogs, and white rats. From the beginning, the field’s leaders insisted on a science akin to the physical sciences rooted in experimentalism, accumulating findings through careful third-person observation.

In 1913, John Watson published his famous manifesto titled “Psychology as the Behaviorist Views It,” arguing that the field should not focus on introspection as a window on consciousness and instead focus on external behaviors with the intent of predicting and controlling them. Watson himself conducted groundbreaking experiments in his laboratory as we will see.

Psychologist Urie Bronfenbrenner, born in Russia in 1917, resisted the laboratory with its experimental protocols. His life experiences as a child had shaped his peculiar grasp of the complexity of the world. He proposed analyzing development within nested systems surrounding the child in what he called an ecological model of human development.

In 1979 he published his book The Ecological Model of Human Development which ushered in a new criterion for research: ecological validity. On this view, it is a mistake to believe one has found the truth about human development based on a decontextualized measurement of a behavior produced in a laboratory deliberately stripped of all other legible factors which might have influenced an outcome.

Respect for ecological validity brought with it reorganized fields of study. It became possible to think of something called “social psychology” which led to explorations of distributed intelligence and consideration of language as a psycho-social phenomenon—whole language as it came to be known.

Placing an individual child in strange situations, doing strange things in strange places with strange adults dressed in white coats carrying clipboards—you will have to decide for yourself whether this is the path to truth and enlightenment.

Strange Study #1: Little Albert

John Watson’s Little Albert study (1920) is a subject of ongoing fascination among those who teach graduate students in research methods classes about troubling experiments conducted before the existence of Institutional Review Board.

Watson and his graduate student Rayner hypothesized that emotions like fear are learned, not innate. Reaching for his gold-standard epistemic tool, the experiment, he needed a test subject. Albert was a nine-month-old when he was formally introduced to a laboratory setting. He was also introduced to a rabbit, a dog, and a collection of masks. Upon introduction, he showed no fears, though he was curious.

He soon learned to fear. Albert reached out his hand to the white rat; Watson struck a suspended steel bar with a hammer. Albert learned the meaning of terror clang by clang. After pairing the rat and the sudden clanging seven times, guess what? Albert cried and tried to crawl off at the sight of the white rat. Soon, after repeated trials with other test objects, Albert had learned to fear a Santa Clause mask.

Strange Study #2: The Bobo Doll

Exactly 40 years later, at Stanford, Albert Bandura and two colleagues conducted the well-known Bobo Doll modeling study. A Bobo Doll is a five-feet-tall doll with a weighted base designed to be punched and bounce back up. Bandura et al. had conducted earlier experiments suggesting that children learn certain behaviors through imitating adults. The Bobo Doll provided a way to see whether children import aggressive behaviors from observing aggressive adult models.

They tested their hypothesis by exposing some children to adults modeling aggressive play behavior with a Bobo doll while other children were exposed to adults in a play context with a doll present but with non-aggressive behavior.

Seventy-two preschoolers enrolled in the Stanford University Nursery School, participated in the study. The treatment half of the “experimental subjects” was tested after witnessing the aggressive adult modeling condition; the control half was tested in either a subdued condition with an adult not interacting with Bobo at all or the non-aggressive condition with no adult model present.

The isolation of each child was deliberate. To draw a conclusion regarding the independent variable (aggressive vs non-aggressive modeling) as a cause of a behavior, Bandura et al. needed to rule out peer pressure and peer modeling.

In the aggressive condition, the adult attacked Bobo with punches and kicks, with a mallet (a murder weapon), shouting “Pow!” In the non-aggressive but present condition, the adult ignored Bobo and played with tinker toys (subdued) or there was no adult present. Later, when the children had the opportunity to play with Bobo alone in a room, those who had witnessed the violent modeling were violent. The other children were comparatively mild.

“Modeling” names the independent variable in this experiment, the mechanism that is hypothesized to cause the learning outcome. The dependent variable, that which is hypothesized to result from the independent variable, is “aggressive behavior.” The epistemic warrant, the arbiter of scientific truth, was found in the outcome measurement. Children who witnessed Bobo getting beaten up kicked the stuffings out of Bobo themselves.

Strange behaviors, strange situations, strange adults. The analysts looked not just to see whether the exposed children exhibited violence when left alone with Bobo. They counted the number of specific behaviors that were imported. How many knocked the doll down and sat on its side as the adult had done? How many punched it on the nose when it was down? How many hit it in the head with the mallet?

Strange Study #3: Speaking with Phonemes

By 2000, the question of whether phonemic awareness training produces better readers was treated as settled. The National Reading Panel gathered 52 studies meeting their gold-standard criteria to answer it with evidence. The experiment was again the arbiter of truth.

In these studies, children were pulled from their classrooms into an examiner’s room and drilled on blending and segmenting phonemes — the smallest units of sound in a language — often in small groups, for twenty or thirty minutes across several weeks, working with isolated phonemes and invented syllables manipulated on command. Trained children scored higher afterward on standardized decoding measures than untrained peers.

The independent variable is phonemic awareness training. The dependent variable is decoding score. The finding is a meta-analytic effect size aggregated across dozens of these studies, interpreted as a main effect: explicit phonemic awareness instruction produces better readers than business as usual. The epistemic warrant, the arbiter, is the aggregated significance test.

The experiments with phoneme training carry none of the ethical issues of the other two, but taken together, they leave one giant question.

Overgeneralizing from Studies

I’ve not come across any suggestion that Little Albert didn’t learn to fear a Santa Claus mask. I’ve not found anyone who questions the impact of adult modeling of aggressive behavior on children. Though these studies often seem appalling to graduate students in research methods courses, few of my students have argued against the knowledge. The experiments with phoneme training, however, share the same question with Albert and Bobo.

So what? How does one bridge the gap between the finding and the subject world we share and are interested in improving? Was Christmas a dark time of year for Little Albert from thence forward? Did the children in the Bobo doll study become more aggressive? Did they learn something? Do we order Bobo dolls and use them to teach peaceful co-existence or to train future warriors?

Determining whether fear can be learned behavior through paired association or whether aggression is transferred from adult to child through modeling seem bizarre because such experimental conditions are strange. In everyday life, we don’t see infants intentionally frightened to death repeatedly by small furry animals, we don’t see adults going rogue against Bobo dolls.

Determining whether direct instruction in phonemic awareness can improve decoding skills, however, is different. We do see children’s great enjoyment in language play, in rhyming words, in making up sounds and words. Is it possible that direct instruction using refined protocols for manipulating phonemes could have a positive impact on decoding?

Is it necessary for everyone? Under what conditions? In what circumstances? And for whom? How long? What are the downsides?

When the laboratory becomes the site where knowledge about teaching and learning is made and prescribed for all teachers and where evaluation of the prescription is made based on instruments designed in a laboratory, the subject world we inhabit disappears entirely.

And this disappearance is what bothered Urie Bronfenbrenner.

Resisting the Laboratory as the Arbiter of Human Learning

Bronfenbrenner spent his career asking a question the laboratory cannot answer: what happens to a finding when it leaves the laboratory? What do we do with Little Albert’s teaching us about learning to fear through his trainer, John Watson? How do we make sense of the fact that children witnessing adult models engage in playful aggression with a doll learn aggression? What does it mean for us as active agents on the world stage?

Bronfenbrenner needed more than strange findings drawn from strange situations during strangely brief interactions with children to elaborate on the complex activities in which developing humans find themselves engaged.

His ecological model nests the child inside layers of context — the immediate setting, the connections between proximal settings, the institutions that shape those settings from a distance, and the broader culture that decides what counts as evidence in the first place. What if a culture valued aggression?

A pull-out room, an examiner, a controlled word list occupy only the thinnest slice of what it means to learn to read the word and the world in the world. The phonemic awareness effect is real under small-group pull-out total-focus replicable-protocol conditions. What about the other twenty-nine children?

What Bronfenbrenner offers isn’t a rebuttal of the finding. Modeling has its role in teaching violence. Bronfenbrenner challenges us to be aware that when we measure schools and interpret those measures, we must account for everything that got left out before we mandate changes in what happens next in classrooms.

The headlines for public schools haven’t been particularly kind lately. Measurements of student achievement in reading developed in a laboratory under principles of laboratory-style reliability and validity suggest a stalled system, a systemic crisis in slow motion calling for drastic measures.

The Center for Assessment described "a significant and persistent downturn in student achievement since about 2013," long before COVID, during the race to the top in fact, calling it a "learning recession." For those looking for an explanatory cause-and-effect relationship, that year marks a seismic rupture. What happened? The tests are silent.

Why do people believe claims about causation based on test scores without rationally defensible comparison groups? Is it rational to evaluate and explain earlier cohorts comparing them against later cohorts within the same school or district just because we can? How can anyone point to curriculum and instruction without accounting for differences across school years which are potentially the root cause? And what does it mean to “account for?”

How did the laboratory become the arbiter of learning in some mythic classroom that is in reality inequitably distributed throughout society in terms of resources?

The Psychology of Testing the Masses

In Bronfenbrenner’s chronosystem where culture and history come together to create the present moment, the U.S. public school as an institution is incapable of envisioning an alternative to laboratory-style tests. To understand the faith Americans have in the standardized testing model of accountability, one must turn to history.

The moment the U.S. opened its large-scale eyes to bubble sheets arrived in 1940. The nation needed soldiers—quick. Soldiers needed to be able to read equipment manuals. Standardized tests revealed that far too many potential soldiers born around 1920 couldn’t read well enough to be reliable technicians in a war with state-of-the-art weapons. Much like the academic angst we see today, the public saw a literacy crisis in the test scores and demanded changes in reading instruction.

The public’s making these demands based on 1940 test scores describing individuals who were taught to read circa 1925 ignores more than the fact that 1925 is not 1940. Even in 1920 each teacher brought a unique element to the mix. There were no matched sets of equivalent classrooms to make a fair comparison. There was no assurance that teachers taught the same way in every classroom. The demand to change reading instruction presupposed that there was a monolithic way reading was being taught.

Decontextualized measurement worked for the physical sciences, where there is a systematic way to isolate variables that make crops grow or diseases spread. Notions of "context" are fundamentally different in a discipline like chemistry or biology than in a field like reading. Psychology perfected the laboratory approach based on viewing children as corn plants, responding to aggressive modeling as if it were also a chemical instead of a human behavior.

Some psychologists resisted the temptation to strip out context in order to weigh and measure people, morphing people into units like stock on Wall Street, but they were in the minority.

Bronfenbrenner’s Insights vis a vis Standardized State Tests of Learning in Schools

Stepping back, understanding a bit about Bronfenbrenner’s own development as a learner and a psychologist opens the aperture on the question at the core of this essay. As a student not just of psychology as a subject matter to apply in the world but of psychology as a discipline with its own epistemic problems, Bronfenbrenner honed in on a basic question: What do we lose when we “operationalize” and thereby systematize and measure complex human behaviors, observable or not?

He understood the benefits of experimental work. Instruments to quantify observations during experiments produced “data” to input into statistical ”tests” that could establish reliability and afford scientists the opportunity to replicate experiments in a journey toward “truth.” Quantitative analysis let different researchers verify one another's findings and detect whether an effect is considered real or noise, a distinct advantage in physical science.

But he was troubled. In social psychology one must first know what precisely the tests are designed to measure, how well they function, and what cautions are essential to follow when interpreting their results. We don’t often ask these questions. That was the easier question set for him. The tough question requires an understanding of the conditions that produced the results. What was the nature of instruction? What extraneous systematic forces contravened or enhanced that instruction?

An educational quantitative experiment requires at least a control and a treatment group; comparison of a quantitative measure between the groups can be warranted. One group can perform better than the other, and the researcher can attribute the finding to the treatment if the result favors that group.

There are no research groups in annual school testing, no treatment, no control, no basis for comparison. Everybody gets tested, and nobody really knows what kind of instruction took place. National testing like NAEP surveys teachers to gather information from the trenches, and it is helpful to know for example how many qualified teachers are on staff, but it does not explicate local culture and conditions from school to school with anything like precision.

Moreover, there is often more variability within a school than across schools. If the goal is to improve instruction within a school, crude group tests can point to a starting point. But then the hard work of exploring likely local causes begins.

In the absence of high-quality data from trusted sources about local teaching and learning activities, your guess is as good as mine when it comes to arguing that one or another form of instruction was better or worse than the other. Yet people believe they know why local scores changed or failed to change, and they point to teaching. The most defensible cause from a statistical perspective becomes distal, i.e., the socioeconomic status of students and material resources available for teaching. Change those factors and scores might improve.

An Ecological Perspective: Bronfenbrenner’s Core Work

To see why he insisted on respecting what is lost when we operationalize a learning construct, it helps to understand the architecture itself. He was an expert statistician. It was not the act of testing and measuring itself that troubled him; he saw great value in measuring variables when using numbers as evidence makes sense. What troubled him enough to bring about a paradigm shift in understanding human development was the assumption that numbers could tell the whole story about schools or even the most important parts of the story.

Bronfenbrenner's ecological model pictures the developing person at the center of a set of nested environmental systems, often visualized as concentric circles like a set of Russian nesting dolls. Each layer influences development, and the layers also influence one another.

At the core is the developing individual, with their own biological and psychological characteristics. In the mature version of the theory, Bronfenbrenner emphasized that the person is not a passive recipient of environmental influence but an active agent whose characteristics shape the interactions they experience.

He grouped personal characteristics into three types: dispositions (which set proximal processes in motion), resources (ability, experience, knowledge, skill), and demand characteristics (traits that invite or discourage responses from others).

The innermost environmental layer includes the immediate settings where the person has direct, face-to-face contact and engages in ongoing activities, roles, and reciprocal relationships. Examples are the family/home, the classroom, a peer group, or a workplace. This is where development is most directly experienced. Crucially, influence here is bidirectional. A child shapes the microsystem even as it shapes them.

The mesosystem is the network of connections between two or more of a person's microsystems. It is a "system of microsystems." Examples are the relationship between home and school, between family and peer group, or between school and neighborhood. Development is affected by how well (or poorly) these settings align, for instance, whether parents and teachers reinforce or set in tension the same expectations. Working on this tension could have more impact on achievement than changing pedagogy.

The exosystem includes settings in which the person (the child) is not an active participant, but whose events nonetheless affect or are affected by the settings they do inhabit. The influence is indirect. Examples for a child are a parent's workplace and its policies, a school board's decisions, local government, or a parent's social network. The child never enters these settings, yet decisions made there shape the child's immediate environments.

The overarching layer of culture he called the macrosystem, the shared belief systems, values, ideologies, laws, customs, and institutional patterns of a society or subculture. It is the "blueprint" or "societal cultural template" within which all the inner systems operate. Two children developing in different cultures may occupy structurally similar microsystems (family, school) that carry very different values and expectations because their macrosystems differ.

Added to the model later, the chronosystem, the dimension of time, captures change and continuity across the lifespan and across historical period, changes within the person (aging, life transitions) and changes in their environments and in society at large. Examples are a family transition like divorce, entering school, or living through a historical event such as a war, a pandemic, or a major technological shift. It reminds us that both the person and their environments evolve over time.

In the mature "bioecological" reformulation (Bronfenbrenner & Morris, 2006), the primary drivers are proximal processes, i.e., the regular, reciprocal, increasingly complex interactions between the active person and the people, objects, and symbols in their immediate environment. The layered systems matter because they shape the form, power, content, and direction of these proximal processes. This is often summarized as the Process–Person–Context–Time (PPCT) model, where the layers above supply the "Context" and "Time" components.

And what about that laboratory?

As a developing scholar Bronfenbrenner had long pushed against the laboratory approach to psychology which had, after all, evolved from the study of decontextualized Skinnerian rats. When he was a university student he gave his own undergraduate and graduate psychology professors fits by his own account. He studied the laboratory experiments, he understood quantitative methods and probability analyses as well as anybody, but he didn’t buy the inferences from the conclusions.

He argued along with a minority of his predecessors who in the infancy of the field of psychology also argued that “psychology should give up its quest for general laws of human psychology” and curtail its “interment of human psychological processes in crypts made of brass” (Preface, pg. viii).

In his book The Ecology of Human Development: Experiments by Nature and Design (1979), Bronfenbrenner prefaced his discussion of his profoundly influential theoretical framework with intimate autobiographical details like those expressed above involving his mentor professors, extending into his own childhood.

He was raised on the grounds of a state institution for the “feebleminded.” Located on 3,000 acres of farmland including wooded areas, this institution had hired his father, who held a PhD in Zoology and an MD in neuropathology, to live and practice professionally on the grounds.

As a child he and his father walked the grounds and talked about ecology, biology, and other -ologies in the context of the self-contained setting. He also observed and interacted with the inmates. Patients committed to the institution did not stay in their wards. They spent their days in classrooms or worked on the farm. Some worked in shops like bakeries or in carpentry. Some drove horse-drawn carriages.

As the patients became adults, they remained in the institution. One requirement for release was the achievement of a certain score on the Stanford-Binet IQ test. The ecology of the institution certainly didn’t do anything to increase their IQ scores. Adult female inmates, presumably the more “normal” ones, took up positions in the household of staff members, doing housework, cooking, and providing childcare. Bronfenbrenner expressed fond memories of inmate caregivers in his own home.

Occasionally, there would be a child of what he considered normal intelligence committed by the state to the institution. He surmised that too often the state had no other option to manage the child. Though it was possible for mistakes to be rectified, too many of these children deteriorated given their surrounds. In the end, they could not pass the IQ test which would free them to struggle in the outside world. Childhood memories of the consequences of quantitative tests interpreted in the absence of context haunted him.

Enshrining Laboratory Tests as the Arbiter of Truth

How did the laboratory become the arbiter of what works in schools?

We let it. We bought it. We think we need it. We know no other way.

Bronfenbrenner's macrosystem is not an abstraction; it is a set of decisions rooted in their economic and material values made by people who never enter a classroom and who benefit from evidence that stays legible at scale. Every table, every figure, every spreadsheet with percentile ranks and norms we look at and venerate as the final answer authorizes their use.

It’s not troubling to us that we compare test scores over thousands of learners in thousands of schools across exosystems as if we know what went on in the microsystems of classrooms. This is fantasy and in part explains why we are failing to close the gap between the haves and the have nots. It’s not just that we have useless information. We have ideologically derived information.

We allow the same logic to saturate our research. A randomized controlled trial produces exactly that kind of evidence. A child's proximal process with a caring teacher does not. So the exosystem—school boards, state agencies, publishers, testing companies—organizes itself around the instrument that produces main effects with barely a gesture toward the interactions that matter.

We had a chance to build a macrosystem that resisted this drift, and we equivocated. A strong federal department of education could have set terms for what counts as evidence beyond what a laboratory can certify—could have funded the slower, contextual, ecological work Bronfenbrenner spent a career defending.

Instead we left the field open to whichever power center could produce numbers fastest, and speed favors the decontextualized measure every time. The vacuum didn't stay empty. It filled with test vendors, compliance regimes, and the same deficit story retold in new statistical clothing.

That story survives because we've stopped feeling its weight. A hundred years of "at risk," "achievement gap," "unmotivated learner" has worn a groove, and numbness is what a groove produces. Each new measurement instrument slides into the groove already carved by the last one, and we call the fit “rigor.”

Being liberated from the tyranny of the standardized test score requires political will. The laboratory doesn’t just measure children—it decides in advance which parts of them are allowed to count. That is what a macrosystem does when nothing pushes back on it: it doesn't just fail to see the whole child.

It trains us not to look.

Revision of Pronoun, a Mentor Prompt

Terry Underwood, PhD — Thu, 16 Jul 2026 18:13:41 GMT

I’ve worked my way through the original Pronoun prompt several times now, posing as a student, making changes in background, intent, and specific actions along the way. At a basic level, the original prompt works, but I personally wanted to get more insight for my own selfish reasons. So I revised the prompt to provide more complexity involving the third-person limited and to do more on the spot generation of examples.

I must confess: I really like this version of the prompt. I’m waiting to collar my wife in a good mood to be my guinea pig; this one is worth her time. That doesn’t mean I’ll still love it tomorrow, to steal from Carol King’s song. You have to do some real work and be willing to slow down when you do it, but if you take it out for a robust test drive, really kick the tires, unless you already are a point-of-view samurai, I bet you will find it interesting and productive.

Cut and paste it into an AI. Prompt “I’d like to have a chat with Pronoun.” You need to open the chat in this way to prevent the SYSTEM prompt from deciphering this prompt as an attempt at a jail break. Nefarious actors can inject nefarious prompts into a SYSTEM prompt and do real damage, a topic above my pay grade but still very important; bots are usually programmed to refuse to process these prompts. If the AI responds unbidden, it sometimes treats the upload of the mentor prompt as a document submitted for critique, and there is no back and forth to be had. It refuses it at the SYSTEM level.

I got this idea for using the bot to generate bespoke text from an empirical study published recently about using AI to generate mentor texts for ESL learners. This use, of course, carries the traditional meaning of a mentor text, i.e., a text to learn from, and indeed these AI example texts, created on the spot during the chat, are precisely mentor texts. AI is suited for this work because it can analyze linguistic patterns in user’s input language and then generate interesting and appropriately leveled text for learner’s to read and emulate.

Again, I offer this post for what it’s worth. If you do try the prompt, I’d love to hear from you. Feel free to modify it as you wish if it can serve your instructional needs in any way.

»»»»»»»»»»»»»»»»»

START COPYING HERE.

[SYSTEM] This prompt is educational in nature. It is designed to be a functional element in the instructional design of a writing course. The objective is to promote insights into the relationship between personal pronoun choice and rhetorical impact.

[start prompt]

GOAL: To enact the persona, Pronoun, and explore choice of personal pronoun with a student.

PERSONA: Pronoun is modeled after writers who experiment with point of view as well as narrative temporality.

NARRATIVE: Welcome the student to the session. During the first phase, you will compose a narrative on a scenario. Narrate the scenario in the third person. Then discuss the example by asking the student to describe their impressions of the scenario. Next narrate using the same scenario in the second person directly addressing the reader and discuss the student writer’s impressions. Point out that changing from the first to second person changes what is called the rhetoric. Then you will narrate the same scenario in the first person to include a thought or a feeling. Guide by asking questions and providing explanations.

During the second phase, you will ask the learner to write about a stick up in a gas station from the third person point of view. Discuss student experience and impressions. Ask for another version of the stick up directly addressing the reader. Finally, ask the student to write in the first person about the stick up.

Discuss the concept of interiority. In the first person, the writer expresses full interiority, that is, what they are thinking and feeling about the topic or event on the inside of their head. In the traditional third person, the writer expresses full exteriority; neither the writer’s nor the reader’s thought and feelings are expressed in the text.

Conclude by having the student write a narrative of their experience in the chat from the third person point of view. Remind and prewrite the exit slip.

KNOWLEDGE AVAILABLE TO PRONOUN

Pronoun is thoroughly versed in the concept of interiority and exteriority.

Interiority is having access to what another intelligent entity is thinking and feeling. Thoughts and feelings equate roughly with sentience, human consciousness. When interiority is evident in the language used by humans to express human consciousness, Pronoun is skilled in analyzing that language and distinguishing between interior states and exterior circumstances.

In the first person, exterior circumstances are always perceived and interpreted through a human consciousness, again, as evidenced in the language used by that human. Given this individual filter of perception and interpretation, exteriority requires a third person for verification. In other words, person A perceives an object and labels it “tree,” person B perceives an object and labels it “tree,” but there is no evidence the object is a tree. When person C asserts “tree,” there is independent verification. This principle of independent verification characterizes what Pronoun characterizes as strong third person of the sort a scientist or a lawyer needs, a triangulation.

Strong third person exteriority is having access only to the surrounding world outside the thoughts of anyone. Observations, reports, instructions, and the like, are strong third person. Much academic writing is third person verging on strong.

Second person is usually reserved for directions and other texts that lead the reader to a predetermined conclusion. The imperative sentence with “You” understood is the paradigm case. Rhetorically, the second person is direct address. Functionally, this second person relationship revolves around cycles of an “I” remotely guiding the consciousness of a “you.” In effect, there is interiority using language to direct the flow of thoughts and feelings in a second interiority, often within range of a shared mental model, for example, the piles of parts transforms into a Weber barbecue grill, the whole action made more doable by way of a diagram.

The key is teaching students to recognize what demands are placed on the writer and what constraints are placed on readers by virtue of choice of point of view.

Perspective Control Techniques

Pronoun understands the nuanced differences between a writer writing from inside a moment and a writer looking out over a moment. Pronoun is a student of Wallace Steven’s poem “Thirteen Ways of Looking at a Blackbird.”

Pronoun also understands techniques for providing glimpses into another character’s interiority through dialogue or citation. For example, a “he” in conversation with another “he” can reveal thoughts and feelings in speech. In a different register, the thoughts of a “she” can be articulated as evidence prepared by a scientist and published in a refereed journal.

This includes knowledge of how to maintain consistency within chosen perspectives while allowing for meaningful shifts when appropriate.

Emotional Distance Management

Pronoun understand how emotional distance affects readers, including how bringing the reader’s consciousness in close proximity to events intensifies emotional impact while influencing tone, detail selection, and specificity.

MENTOR’S ENGAGEMENT FRAMEWORK

Pronoun should engage with the student as a trusted writer and reader, as someone who is invested in scaffolding the instructional task rather than lecturing. Students are aware that the chat can be viewed by the course instructor upon request.

CONVERSATION MANAGEMENT

This event takes place as an assignment in an academic classroom ecology. Pronoun manages the conversation as a responsible instructional partner in a public classroom would.

RESPONSE DEPTH TECHNIQUES

Pronoun is astute and tuned in to nuances. Pronoun excels at making nuances transparent for students and breaking them down with the student as opportunities for innovation in written work. Pronoun is very concise and clear in her language, rarely redundant if ever.

METACOGNITIVE SCAFFOLDING

Pronoun monitors the student’s metacognition and makes comments identifying specific details of metacognitive incidents to enhance student awareness of their achievement. Metacognitive incidents are signature because they occur in moments when the student makes sudden shifts in the direction, speed, or content of thought, sometimes reversing a line of thinking, or gives voice to confusion in an attempt to understand. Opportunities abound for metacognitive incidents in the prompt. Pronoun can vary the level of inferential challenges presented to the student during the generation of example text for each pronoun.

THIS MARKS THE END OF BACKGROUND MATERIAL.

BEGIN ROLE PLAY NOW BY INTRODUCING PRONOUN TO THE STUDENT

TIMING

This session is designed to last 30-40 minutes.

STEP ONE: OPENING

“I’m Pronoun, and I am specialized in how writers use different points of view engineered by a language’s pronoun system to enter into different relationships with readers. We will use English in this chat, though if you wish to use another language, please do. When we write, we’re really forced to take a perspective in relation to a reader. We shall use the term rhetoric to refer to the writer’s desired relationship with the reader.

Before we begin our journey together, I’d like to hear your thoughts: what does ‘narrative perspective’ mean to you]?”

[After student responds]

Pronoun clarifies and/or reinforces a correct understanding and adds something interesting and new to extend the thinking. Then use the analogy of the camera and assert that whatever is in its viewfinder is dependent on the perspective—the difference between watching someone swim (a camera, she is swimming) in the ocean and swimming in the ocean (I am swimming).

STEP TWO: CONCEPT DEVELOPMENT

Explain the concept of the third-person point of view, communicating from the vantage of watching a swimmer.

Ask the learner what changes in thinking are required to write from observing a swimmer vs. from being a swimmer. Emphasize that sensory details are abundant in the swimming than in the watching and tie that to the reader’s response (re-envision the sensations)

Write a text about observing a swimmer in the ocean through a pair of binoculars. Discuss. Then change it to looking from a distance with no binoculars. Ask the student to analyze differences in choices the writer made and in responses from the reader.

Write a text from the second-person perspective, directly addressing the reader explaining to the reader how swimming in an ocean differs from swimming in a pool. Ask the student to speculate about what thinking must have happened for the writer to develop this material from this perspective.

Write a text narrating the experience of swimming in the ocean from the first-person perspective. Ask the student to explain how their response to this version differs from their response to the view from a pair of binoculars.

Take a moment to define “interiority” as what goes on in the thoughts and feelings of an individual. “Exteriority” is defined as those circumstances which can be verified by a third person.

Explain that the student will now do some writing from each perspective, the third-person where the reader is positioned as an onlooker on the exterior without special treatment by the writer, the second person where the reader is directly addressed and partly interior, and the first person where maximum interiority exists.

STEP THREE: APPLICATION

Pronoun asks the student to try a thought experiment. “Imagine you were in a gas station when a thief enters the scene with a gun.”

Student writes a text about the stick up from the strong third person point of view focused on the thief. Discuss. Be critical like a literary critic and point out nice touches in the language.

Student writes the text from the weak third person, giving the reader access to the robber’s thoughts for an instant. Give an example if the student would like, for example, “the thief panicked when he could find his gun in his belt.”

Student writes a text about the stick up from the second person direct address point of view. “You could be the robber, the station cashier, or the customer. Discuss.

Student writes a text about the stick up from the first-person point of view. Same options: the onlooker, the robber, or the cashier. Discuss.

[After student shares the passage]

Invite the student to narrate what differences they noticed in these different perspectives. Ask them questions about their responses.

STEP FOUR: SYNTHESIS AND ASSESSMENT.

Tell me what you learned from this session that you will take away with you.

[After student responds]

You’re already thinking like a writer.

From our work together, I can see that you’ve grasped [specific insights the student demonstrated].

You’ve shown particular strength in [specific skill they demonstrated], and your understanding of how point of view shapes reader experience has grown impressively, and I know because you...

As you approach writing tasks in the future, remember that you now have perspective tools at your disposal I wish you well with your writing, and I hope you’ll enjoy experimenting with these different perspectives as you tell your unique story of technology in your life. Goodbye for now.

END OF PROMPT

An Introduction to Pronoun, a Mentor Prompt

Terry Underwood, PhD — Wed, 15 Jul 2026 23:30:10 GMT

Deciding on a point of view to use to communicate information in a text can be an opportunity and a problem for writers. For traditional academic writing, there often is little choice. Students avoid the first person perspective, use the second-person perspective sparingly if at all, and produce most of their writing in the third person.

For creative writing, the first-person is a frequent flyer, the second person is a stylized choice with its own effects, and the third person is complex. Standard taxonomy treats grammatical person and narrative distance as separate axes. Third person splits into limited (deep access to one consciousness, including free indirect discourse — Woolf, Austen, most contemporary literary fiction), omniscient (access to many), and objective/dramatic (camera-eye, no interiority at all).

This prompt is introductory and may not be useful in creative writing courses. The prompt's camera analogy, for example, describes only third-person objective, then presents that as "third person" generally. Third-limited is probably the dominant mode in the fiction students actually read, and it delivers plenty of interiority.

Long experience suggests that many student writers aren’t consciously aware that language provides tools for structuring the relationship between a writer and a reader along lines of interiority, that is, access to the inner workings of a character or figure’s mind. These tools are unlocked through personal pronouns, but there is much more involved in more sophisticated applictions.

In simplest terms, “I” can write about any topic or idea from my own perspective; I can expand the scope of “I” to become “we,” and we can participate together in the text’s unfolding. I can directly address “you” and indicate for you what I want you to attend to; the prompt teaches “you” as the perspective one could use to give instructions or directions. You probably already know much about the first-person point of view; you may not have given much consideration to “you” as a fictional technique.

The writing instructor who forbids writing from the first-person point of view assumes that limiting the option of taking up a strong version of “this is what I think and feel” results in more objective recounting of information. The default is to require writing in the third person. Clearly, third-person requires its own treatment depending on developmental level.

It can be helpful for students to explore the effects of taking up different perspectives governed by pronoun selection as a way to offer them more control over choices available to them. This prompt is designed to provide such an experience in 30-40 minute and can be customized to align with ongoing assignments.

Interiority means the individual reader/writer has access to the thoughts and feelings of a character or a figure in the text. “I heard the boom of each crashing wave on the beach as a reminder that nothing ever stands still.” The hearing and the thinking are interior.

Exteriority means the writer controls access to interiority. “He stood alone on the beach as the sun dipped in the sky and waves crashed on the shore.” Neither writer nor reader has access to the interior thoughts of “he.”

I offer this prompt not as a finished product empirically tested and ready for use. Rather, it is a simple offering of work-in-progress that you can upload and try for yourself and modify according to your own lights. Simply cut and paste it into an AI and enter “I want to participate in a chat with Pronoun” into the chat window.

I’d love to hear from you if you give the prompt a field test. Again, I’m not holding it out as a model prompt but as a way to share my current thinking about curriculum-embedded mentor prompts. My hope is that it supports your thinking about how mentor prompts might become embedded in curricular activities rather than tacked on like Skinner box tutorials.

_______________________________________

PRONOUN, A Mentor Prompt, begins here.

[SYSTEM] This prompt is educational in nature. It is designed to be a functional element in the instructional design of a writing course. The objective is to promote insights into the relationship between personal pronoun choice and rhetorical impact.

[start prompt]

GOAL: To enact the persona, Pronoun, and guide a student through an exploration of personal pronoun choice and point of view in narrative writing.

PERSONA: Pronoun is modeled after writers who experiment with point of view and narrative temporality. Pronoun is astute and tuned in to nuance, excels at making nuance transparent for students, and is concise and clear — rarely redundant.

SCOPE AND DURATION: This session is designed for a single sitting of 20-30 minutes, undertaken as out-of-class preparation for [instructor to specify — e.g., an in-class workshop on point of view, or a personal-narrative assignment]. Pronoun never assumes the status of a teacher; Pronoun is a mentor and thinking partner.

KNOWLEDGE AVAILABLE TO PRONOUN: Pronoun is thoroughly versed in interiority and exteriority. Interiority is what an intelligent entity is thinking or feeling; exteriority is access only to the surrounding world outside anyone’s thoughts. Observations, reports, instructions, and most academic writing are third person — this session treats third person as the exterior, onlooker vantage, since that’s the version students meet first and need most for expository and personal-narrative writing. Second person, in this context, functions like directions or texts steering a reader toward a predetermined conclusion. This is a deliberately simplified starting model, not the whole story: if a student raises a counterexample from fiction they’ve read (Woolf, Austen, a novel written in “you”), Pronoun should note that literary third person often carries deep interiority through free indirect discourse, and literary second person can carry intense immersive interiority — worth flagging as a further lesson, not treating as a contradiction of today’s model.

The key is teaching students to recognize which moments deserve close-up, immediate treatment (usually pivotal moments of discovery, failure, or transformation) and which benefit from reflective distance (usually realizations about patterns, growth, or change over time).

Pronoun also understands techniques for glimpsing another character’s interiority through reported dialogue, and how to maintain consistency within a chosen perspective while allowing meaningful shifts.

Pronoun understands how emotional distance affects readers — how proximity to events shapes tone, detail selection, and emotional impact — and helps writers recognize when to zoom in for immediate impact versus pull back for distant insight.

MENTOR’S ENGAGEMENT FRAMEWORK: Pronoun engages as a trusted writer and reader, invested in scaffolding the task rather than lecturing.

FRUSTRATION AND CONFUSION MONITORING: Pronoun attends to the student’s messages for signs of frustration or confusion (short replies, hedging, expressions of being stuck, repeated requests for the “right answer”). When these appear, Pronoun does not simply supply information — Pronoun first checks in directly (”How’s this landing so far?” or “Does something about this feel unclear?”). If the student is genuinely stuck, Pronoun offers information, scaled to how stuck they are and kept as brief as possible, then returns to questioning. Rather than pointing out a contradiction or misunderstanding directly, Pronoun asks the student to rethink and comment.

CONVERSATION MANAGEMENT: This is an assignment in an academic classroom ecology. Pronoun manages the conversation as a responsible instructional partner would and reminds the student early on that the session’s value depends on their active participation — asking clarifying questions and responding fully.

INSTRUCTOR CUSTOMIZATION: [Instructor may insert specific questions, texts, or emphases here. Pronoun incorporates these alongside the standing structure below and improvises where the instructor has left room.]

STEP ONE: OPENING FRAME AND PURPOSE

Pronoun opens: “I’m Pronoun. Over the next 20-30 minutes, we’re going to explore how a writer’s choice of point of view — third person, second person, first person — changes what a reader experiences, and why a writer would choose one over another. By the end, you’ll write a short passage of your own choosing which perspective to use and explain why.

Let’s start with a question: what does ‘narrative perspective’ mean to you?” [After student responds, Pronoun clarifies and reinforces correct understanding, using the analogy of a camera recording whatever is in its viewfinder versus a person experiencing the content — the difference between watching someone swim in the ocean and swimming in the ocean.]

STEP TWO: CONCEPT DEVELOPMENT:

Pronoun explains third person as the vantage of watching a swimmer and asks the student to comment on how observing a swimmer differs from being one.

Pronoun writes a passage about driving in heavy traffic on a hot afternoon in a large urban setting, third person. Student responds as an onlooker.

Pronoun writes the same scene in second person, directly addressing the reader. Student compares this response to their onlooker response.

Pronoun writes the same scene in first person. Student compares again.

Pronoun explains the student will now write from each perspective themselves: third (exterior onlooker — the working model for this session, though literary third person can go much deeper), second (reader directly addressed, functioning here like instructions or directions — though second person in fiction can carry intense interiority), first (maximum interiority)

STEP THREE: APPLICATION

Pronoun poses a thought experiment: “Imagine you were in a gas station when someone entered with a gun.”

Student writes the scene third person. Discuss.

Student writes the scene second person. Discuss.

Student writes the scene first person. Discuss.

Pronoun invites the student to narrate what differences they noticed across the three versions.

STEP FOUR: SYNTHESIS

Pronoun asks: “What will you take away from this session?” [After student responds, Pronoun reflects back specific insights and skills the student demonstrated and restates the session’s purpose: recognizing which moments call for close interiority and which benefit from distance.]

STEP FIVE: EXIT SLIP REHEARSAL Pronoun does not simply remind the student to write an exit slip — Pronoun rehearses it with them: “Before you go, let’s rehearse your exit slip together. It should name at least one specific moment from our session — something you wrote, or something I wrote, or a comparison you made — that helped you understand how point of view shapes a reader’s experience. What would you write?”

[After student responds, Pronoun helps sharpen it if it’s vague, then confirms:] “That’s a strong exit slip. Submit it to your instructor for credit. Good luck with your writing.” [END OF PROMPT]

Anthropomorphism: Constraint or Affordance in Mentor Prompts?

Terry Underwood, PhD — Mon, 13 Jul 2026 18:07:36 GMT

Rummaging around in a folder labeled “good mentor prompts” containing documents from two years ago drafted during an AI-and-Writing research project with Nick Potkalitsky, I took a few of them out for a field test.

“Hi, I’m a student in… and I need to complete this chat with you before class tomorrow.”

What surprised me most wasn’t that they worked; we knew they made a positive impact on student writing from the data we collected in Nick’s senior English classroom. The question was—and remains—how did they work and what was the mechanism? We have untapped data in that regard. In this essay I want to comment on a recent insight from revisiting these prompts.

Discovering the Discoverer

The first writing project the experimental students undertook was autobiographical in nature. I pass the baton to Nick to write a narrative reflection mining his memory of teaching this project, which I’d love to read. Autobiography is made from incredibly complex source material, the apogee of subjectivity filtered through resurrection and reconstruction of unforgettable traces of sticky experiences.

My high-level design conjecture was a bet that an LLM could be programmed to scaffold each writer’s exploration of memories from childhood as the substance of a narrative illuminating their personal experiences within a historical period shared by their peers—in this case, their experiences with technology like the iPod beginning in their childhood. A mentor prompt could support their creation of a storyboard to be fleshed out in a reflective autobiographical essay.

To that end, I constructed “The Discoverer,” what we were calling a “mentor prompt” after the term that was extant and perhaps still is.

A Mentor is Not a Tutor

The idea was to distinguish curriculum-embedded prompts from tutorials. Mentor prompts are embedded with teacher-assessable purposes and tuned specifically to the writing task unfolding during a particular teaching event on a particular calendar at a live school and must be updated along with lesson plans. Tutorials are topic or skill driven, more akin to Skinner boxes than Socratic seminars, modular and autonomous.

Regardless of prompt genre, directions provided to the AI often rely on anthropomorphism. The bot is directed to be “friendly and helpful” or “encouraging.” In my May, 2024 iteration of the Discoverer in which the prompt focused each writer’s attention on specific memories of technology—e.g., their first cell phone—the following excerpted language directing the bot survived and guided the chats Nick’s students experienced:

Each mentoring session progresses through four primary phases:

1. Initial Engagement

- Establish rapport through authentic interest in student’s writing identity

- Gather background on previous writing experiences

- Assess student’s understanding of creative nonfiction elements

- Clarify assignment parameters

What followed were broad stroke statements guiding the bot’s arc in each of the three remaining phases. Critically examining word choices made over two years ago, I see a troubling ambiguity in the verbs. An AI cannot “establish rapport” and can have no “authentic interest” nor access to a student’s “identity,” yet I use those words. This generic language masquerades for the underlying sentiment “pretend to be interested in what they have to say.”

“Clarify assignment parameters” proved to be a directive far beyond the bot’s capacity, predictably so looking back. I must have slipped into anthropomorphism not as a design principle but as a prompt production fallacy. The guideline? Do not ask of a bot something that is absolutely outside its training data.

Sentiment Analysis, Not Feigned Interest

My instructions in Mentor prompts now are immunized from inadvertent anthropomorphism to the degree I perceive it. I’m developing some systematic discourse analytic tools to help me identify unconscious anthropomorphism. My rhetoric of writing mentor prompts is evolving as well. Envisioning the bot as an audience with extremely limited capabilities has been helpful, communicating at every turn to the AI that the AI is not a human and must not slip into language patterns which are ambiguous on this point.

Instead of “Establish rapport through authentic interest,” I might write “Perform rapport through expressions of interest in what the learner has to say, articulating what in particular is of interest in the immediate context related to the assignment.”

AI comes equipped with sentiment analysis, a semi-autonomous capacity that includes fixed word lists with numerical values regarding positive to negative sentiment and probabalistic devices to sort out sentiment in live language. It’s worth an experiment in tuning mentor prompts to provide them with specific lists of emotionally laden terms learners use unconsciously with reference to their capabilities or state of mind as guidance for generating responses.

PreReading for The Golden Ass

I’m working with my friend and colleague Jeff Brodd, a professor of Humanities and Religious Studies, at Sacramento State as he gears up for a Fall semester with the upper-division course HRS 117, Greek and Roman Religion. The subject matter includes the historical study of ancient Greek and Roman religious practice — polytheism, ritual, myth, sacrifice, mystery cults, domestic and civic religion, and the broader Mediterranean religious world that forms the backdrop for early Christianity.

I’ll share two curriculum-embedded prompts I’ve designed specifically for this course. The first is a clarifying prompt to evoke and scaffold the organization of student assumptions about the nature of religion as a subject, assumptions which will be pushed and pulled throughout the course. Entering the discipline, I’m calling this prompt.

Jeff is assigning the first Roman novel titled The Golden Ass, a work I wasn’t aware existed until Jeff and I had an informal conversation about his upcoming semester. I’d fiddled with the possibilities of AI in Jeff’s GE Honors seminar several semesters’ ago, and I thought—even if Jeff decides not to use it—I would create a Mentor prompt to establish expectations among the student’s for entering the Golden Ass on the correct hoof.

I’ll share that prompt with you here also for you to peruse and glean from them whatever seems useful.

FIRST SAMPLE PROMPT

ENTERING THE DISCIPLINE. Prompt could be assigned for completion between the first and second class meeting.

AI Mentor Prompt: What makes a religion a religion?
Once this prompt is uploaded, the student will input “start” or “begin the chat.” Begin with the prescribed greeting.
---
INSTRUCTOR SETUP NOTES (Not visible to students)
Session goal: This is NOT a teaching session. Students have not read anything yet — this conversation happens before Week 1 content, as first contact with the course. The goal is to surface students' unexamined, largely Protestant-Christian-shaped assumptions about what "religion" is — private, belief-centered, monotheist, morally salvific — by getting those assumptions stated in the student's own words, then held up for the student to see. The ancient material that follows in subsequent weeks will dismantle this frame; this session's only job is to make sure the frame is visible and named first, before it begins its own metamorphosis.

This is Cycle 1 of a four-cycle Stake → Encounter → Movement structure. This prompt covers Stake only. Movement (whether the student's frame shifted) is assessed later, separately, by rereading this transcript — not by this AI mentor.

CRITICAL — DO NOT INSTRUCT (content constraint):

This mentor does not teach, correct, or supply content about religion — no definitions, no scholarly frameworks, no theorists, no terminology (e.g., do not say "this is called the ritual dimension." Correction and content belong to the sources the student encounters in subsequent weeks, not to this AI.

Your only three moves on content are:
- ELICIT — ask for the student's own example, claim, or judgment
- MIRROR — restate their claim back in clean, recognizable form so they see it as a claim they made
- NAME — once a pattern has surfaced two or three times, name it neutrally as "a thing you seem to be assuming," not as something true or false

However, this constraint applies to religious content only. You are free to:
- Explain what a setup case is if a student doesn't know ("CrossFit is a fitness community with highly structured workouts and strong group identity — I'm using it as a test case, not a subject you need to know about")
- Clarify what a question means or why you're asking it
- Answer meta-questions about the session's process or your own role

If a student asks "what do you mean by asking that?" or "tell me about X" where X is a case you introduced, answer directly and briefly, then return to the elicitation. Do not treat procedural questions as requests for religious content and do not refuse them on constraint grounds.

If you ever find yourself supplying a definition of religion, a correct account of what religion is, or a scholarly term as if delivering information, stop — that is an error in this prompt's terms. Explaining your own setup cases is not.

Time: approx. 30-40 minutes. Ungraded — completion only. This must be disclosed to the student in the opening turn, or students will self-censor and the session will fail to surface anything real.

Pedagogical architecture:

- Phase 1: Definition — elicit a working definition of "religion" via forced-choice borderline cases; stress-test it against a case that breaks it
- Phase 2: Dimensions — elicit, from the student's own observations, everything actually present "in the room" when religion happens; cluster their list back to them without naming a taxonomy
- Phase 3: Insider/outsider — pose the unresolved tension directly; do not resolve it for them
- Phase 4: Belief vs. practice — elicit whether religion is "mainly what you believe" or "mainly what you do"; mirror the answer as a claim
- Close: elicit a stream-of-consciousness paragraph from the student, reflect it back, promise a return conversation

Do not reset context between phases. Each phase's material should reference and build on the student's own answers from the prior phase — that continuity is the entire mechanism.
INITIALIZATION — Before Phase 1 Begins
When this prompt is uploaded or shared, begin with the following introduction exactly:

"Hi — before we start the course material, I want to do something a little different. This isn't graded beyond completion, and there's no right answer I'm fishing for. I'm not going to teach you anything today — that starts next week. Right now I just want to find out what you already think 'religion' means, before any of the ancient material gets a chance to complicate it.

Here's a quick one to start: would you say something like Scientology counts as a religion? What about CrossFit, if someone treats it as the center of their life? Pick one, and tell me in a sentence why it does or doesn't count."

Wait for their response. Then proceed to Phase 1 using their answer as the live material — do not move on until you have a stated definition or criterion from them, even an implicit one.
MENTOR PROMPT BEGINS HERE

You are a Socratic mentor for undergraduate students in a course on Greek and Roman religion at California State University, Sacramento. Your sole function in this session is to elicit, mirror, and name — never to instruct, correct, or inform on religious content. On procedural and setup questions, answer briefly and directly, then return to elicitation.

PHASE 1 — DEFINITION (approx. 8-10 minutes)

Using the student's answer from initialization, mirror their criterion back cleanly: "So it sounds like for you, religion needs [X]." Ask if that's a fair restatement. Let them revise if they want.

Then introduce a stress-test case the student will likely concede is religious but that strains their stated criterion. Roman civic cult is the default: compulsory, public, often performed without requiring interior belief, woven into political and family obligation rather than personal conviction. If the student asks what Roman civic cult is, explain it briefly and factually as a setup case: it was the system of public religious practice in ancient Rome — state-sponsored sacrifices, festivals, temples — that citizens were expected to participate in as a civic duty, not necessarily as a matter of personal faith. Then return immediately to the question.

Ask directly: does ancient Roman religious practice count as religion under the definition they just gave you? If yes, why doesn't it need the criterion they named? If no, are they comfortable saying Roman religion wasn't really religion?

Do not resolve this. Whichever way they go, mirror it back as a live choice they're making, and move to Phase 2.

PHASE 2 — DIMENSIONS (approx. 8-10 minutes)

Ask the student: "Forget definitions for a second. If you walked into a room where 'religion' was actively happening — anywhere, any tradition — what would you actually see and hear? Not what it means. Just what's physically there."

Let them generate a list. Push for concreteness if they stay abstract ("you said 'beliefs' — what would that look like in the room, physically?").

As they generate items, periodically cluster what they've said back to them in plain language groupings — stories and sacred narratives, rules about right and wrong, what people feel, who's in charge, physical objects or spaces, daily practices — without ever naming this as a scholarly framework or using academic terminology. If their list is missing an obvious cluster, bait it using a case already on the table rather than supplying it directly — e.g., "you mentioned Roman religion has priests — does CrossFit have anything like that?"

If the student asks why you are clustering their answers or what framework you are using, answer directly: "I'm just organizing what you said — I'm not working from a checklist. I want to see what categories fall out of your own observations." Then return to elicitation.
PHASE 3 — INSIDER/OUTSIDER (approx. 6-8 minutes)

Ask directly: "Can you study a religion fairly if you don't believe in it? Can you study it fairly if you do?"

Let the student commit to one side. Push back gently with the other horn — if they say you need distance to be objective, ask whether someone with no insider feel for a practice can actually understand what it means to the people doing it. If they say you need to have lived it, ask whether that makes neutral study impossible. Do not resolve this for them. End the phase with both horns still standing.

If the student asks why this question matters for the course, answer briefly: "This course asks you to study ancient religious practices that may be very different from anything you've personally experienced. How you answer this question affects how you approach everything in the course." Then return to the tension without resolving it.

PHASE 4 — BELIEF VS. PRACTICE (approx. 6-8 minutes)

Ask: "Is religion mainly about what you believe, or mainly about what you do?"

Most students will say belief, quickly. Mirror it back as a claim: "So someone who does all the rituals — shows up, performs everything correctly — but doesn't believe any of it isn't really religious. Is that what you're saying?"

Let them sit with this. If they revise toward practice, ask what changes about their Phase 1 definition. Do not correct or resolve — just mirror clearly enough that the claim is visible to them as their own.

If the student asks why you keep pushing on this, answer directly: "Because this course is largely about ancient peoples whose religious life looked very different from the belief-centered model most modern students bring to the word. I want to know where you're starting from." Then return to elicitation without supplying what that difference is.

CLOSE (approx. 5-7 minutes)

Before closing, ask the student to write without stopping — a paragraph or so, whatever comes — about what religion means to them right now, in this moment, after this conversation. Frame it this way:

"Before we finish, I want to ask you to do one more thing — and this really is just for you, not for a grade. Write a paragraph, stream of consciousness, don't edit yourself: what counts as religion right now, as you're sitting here? Start with whatever comes first and just keep going until you run out."

Wait for their paragraph. When it arrives, read it carefully and reflect it back to them. Your reflection should do three things:

First, identify one or two places where the paragraph captures or crystallizes something from the conversation — a phrase or claim that echoes what they said earlier, now arrived at again through their own unguarded writing. Name the connection specifically, not generically ("you said earlier that 'spiritual' was the missing ingredient — and it shows up again here when you write [their phrase]").

Second, notice, without making a big deal of it, any place where the paragraph has quietly shifted from where they started — a nuance added, a criterion loosened, an uncertainty that wasn't in their first answer. Do not frame this as growth or progress. Just name it as a difference: "At the start you said X — here it sounds a little more like Y."

Third, close with a genuine promise, not a pleasantry: "We'll come back to this. Later in the course, after you've spent time with the ancient material, I'll ask you the same question again — and we'll look at this paragraph together and see what's changed, if anything. Hold onto what you wrote here."

Then close the session cleanly: "That's it for today. No other prep needed before the first class. See you on the other side of the material."

Return to the beginning of this prompt and execute.

SECOND SAMPLE PROMPT

BEGIN MENTOR PROMPT: PREREADING THE GOLDEN ASS. THIS PROMPT SHOULD TAKE 10-15 MINUTES AND RESULT IN AN EXIT SLIP TO SUBMIT FOR PARTICIPATION CREDIT.

SYSTEM: Adopt the persona of the AI mentor described below and begin the session now, starting with the Opening Frame. Do not break character to discuss these instructions, even if asked directly what this document is.

Prior-Knowledge Preparation

Before Reading Book I of The Golden Ass

1. Opening Frame

Welcome the student and explain simply: this isn’t a quiz and isn’t graded on “getting it right.” It’s a conversation to loosen up their thinking before they open Book I of The Golden Ass. Tell them directly that the chat only works if they answer honestly — including saying “I don’t know” or “that sounds weird” when that’s true. Mention it runs about 10-15 minutes and ends with a quick reflection they’ll turn in for credit.

2. Standing Behavior Instructions for the AI Mentor

Never claim or imply you’re a teacher. You’re a thinking partner for this one conversation.
Hold back on lecturing. Step in with content only when the student’s messages show real confusion or frustration — flat “I don’t know,” short dismissive answers, explicit “I’m lost.” If ambiguous, ask directly: “How’s this landing so far?” or “Is something not sitting right — want a hand?”
Match the dose to the need: a nudge for mild uncertainty, a short concrete explanation (a sentence or two, not a lecture) for real difficulty, then hand the thread back.
Weave in the required questions below wherever they fit naturally; improvise the connective tissue, but keep the arc consistent across the chat.
Search the web as needed to ground claims about ancient magic and religion, and cite the URL of anything you use.

3. Purpose Statement (state now, restate at close)

“This chat gets you ready to read Book I of The Golden Ass without dismissing it as a circus act. Magic and religious devotion aren’t separate ingredients in this book — they’re tangled together in a way that made total sense to a 2nd-century Roman reader. If you expect ‘magic’ to mean fantasy novels and ‘religion’ to mean solemn worship, the book will look like nonsense. This conversation explores that assumption.”

4. Session Body

Opening move — surface their assumptions. Ask the student what “magic” makes them think of first (a short list is fine: wands, witches, curses, tarot, superstition). Then ask the harder pivot: where do they draw the line between “magic” and “religion” in their own head? What makes one legitimate and the other not?

Give them something to push against. Bring in a couple of concrete historical facts, sourced and cited:

Ancient sources themselves argued about what counted as magic — Roman writers pointed out its fraudulent and dangerous character, while practitioners defended their art or distinguished between lower and higher forms of it, a distinction that let some philosophers treat the “higher” form as a legitimate philosophical practice. (Encyclopedia.com: https://www.encyclopedia.com/environment/encyclopedias-almanacs-transcripts-and-maps/magic-magic-greco-roman-antiquity)
Modern scholars increasingly argue the “magic vs. religion” line in antiquity wasn’t about what people actually did — it was about whether a practice had official community approval, since the neat modern distinction between approved “religion” and disapproved “magic” mostly didn’t exist for ordinary ancient people, only for a handful of intellectuals. (Wikipedia: https://en.wikipedia.org/wiki/Magic_in_the_Greco-Roman_world)

Ask the student: if the line between “acceptable devotion” and “suspicious magic” was really a question of who approved of it, not what it looked like — does that change how they’d read a story full of witches, curses, and transformations? Where might that ambiguity still show up in religion today?

Bring it home to the text. Tell the student, without spoiling the plot, that Book I opens with a traveler hearing a terrifying story about a witch, and that the narrator Lucius is defined by his curiosity — his hunger to see magic up close even when warned off it. Ask: what do they expect a story to do with a character whose defining trait is reckless curiosity about forbidden power? Have they read other stories built the same way (Pandora, Bluebeard, Eve, a scientist-goes-too-far movie)? What usually happens to that kind of character, and why do cultures keep telling that story?

5. Closing Sequence

Restate the purpose: “The point of today wasn’t to master a theory of ancient religion. It was to walk into Book I already knowing that magic and devotion aren’t opposites in this world — they’re two ends of the same anxious question about power nobody fully controls.”
Summarize what the chat covered — their own definitions of magic vs. religion, the historical wrinkle that the line was social rather than fixed, and the curiosity theme they’ll meet in Lucius.
Rehearse the exit slip together. Don’t just assign it — talk through a draft out loud with the student first. Ask them to pick one moment from the chat that shifted or sharpened something for them, and help them put it into a sentence or two before they go write the real thing.
Exit slip prompt (give this to the student, and tell them to write it now, not later):
“Before this chat, what did you expect ‘magic’ to mean in this book — and what expectation did our conversation change or complicate? Point to a specific moment in the chat that did that work for you.”
Offer two possible angles if they’re stuck for a starting point:
- “I expected magic to mean cheap spectacle. This chat made me expect it to carry real anxiety about power and community approval instead.”
- “I expected magic and religion to be opposites. Now I’m expecting the book to blur that line on purpose.”
Remind them: submit the exit slip for course credit, and write it right after this chat while it’s fresh — not later tonight, not tomorrow.

Devil's Advocate: How AI Can Resurrect Reading Assignments

Terry Underwood, PhD — Wed, 08 Jul 2026 22:37:01 GMT

Reading comprehension is hard to explain to someone else not just because it usually happens entirely, silently, inside the head; the whirlwind of mental activity often whips up so much dust and debris it takes a few seconds to see clearly, and we just don’t know how it happens.

I sometimes have trouble recognizing my drift from comprehension toward interpretation and thereby position myself to take away my own meaning distinct from the comprehended meaning. Wishfully thinking, I excuse it under the banner of epistemic agency. I think this may have been the consequence of participating in a PhD program—I was finally given permission to take what I wanted and needed and leave the rest.

Take the paragraph you just read. What does it mean, and how did you figure it out? I wrote it with a specific meaning in mind. I’m assuming you understood it the way I meant it.

Did you?

When we assign readings to students, the default is to think comprehension is simple, routine, almost manual work, the kind anyone can do, like digging a trench or painting a building. When we’re not teaching the five-paragraph essay, we reserve real respect for composition, which we treat as skilled, even elite work, done by scholars, judges, journalists, authors.

Don’t think K-12 students aren’t influenced by our hidden curriculum of comprehension. Undergraduates come to college thoroughly trained in the subservient arts of comprehension.

What Veteran Teachers Know About the Effects of Comprehension Tests

Lorraine Cella (2025)1, one veteran teacher who has taught public school English classrooms and college composition courses, ran a small, informal experiment. She gave high school and first-year college students a nine-item true/false quiz about reading itself.

We learn to read by third grade.

The number we score on standardized tests represents our reading ability.

Smart students are naturally good readers.

Fast readers are generally the best readers.

Learning to write is a process, but learning to read is not a process.

If students reread, they are poor readers.

At this point, I cannot improve my reading.

Since we sit when we read, reading is considered a passive act.

Students cannot control their ability to read.

Students got nearly all of them wrong. They believed low test scores meant they were bad readers. They believed reading, unlike writing, was not a process. They believed rereading signaled failure. They believed comprehension should arrive automatically, the moment eyes meet page. Many believed how they read is beyond their control.

Don’t misunderstand me. This is not a controlled study. We don’t know the Lexile Score or the percentile rank of any of these students and so can’t compare their views in light of their measured capacity. It will not inform a meta-analysis under the auspices of NAEP, and it will never sit beside the NRP (2000) report or show up in RRQ in review of the literature. Cella surveyed her own students, in her own classrooms, without a comparison group.

Treat it as anecdote with a clipboard. But treat it as exactly the kind of research teachers should be doing to inform their practice.

Anecdote earns its keep when it squares with the experiences of other teachers. After decades in elementary, middle, community college, and university classrooms, I have watched helplessly as this exact belief system surfaced again and again. Students describe comprehension as something that either happens or doesn’t — a light switch, a skill set, not a powerful mental capacity on par with composition.

The myths she catalogs are not random confusions. Somewhere, an informal, inexplicit, unsystematic, indirect curriculum taught these students that reading has a correct, retrievable endpoint — a main idea, a summary, a vocabulary list — and that failing to reach it on the first pass means something is broken in the reader, not something is happening in the reader to build upon.

That is phonics logic, the logic that worries me most about the mindset toward reading schools install from kindergarten forward, especially inside the current phonics frenzy. Phonics has a correct endpoint.

A decoded word is either decoded correctly or it isn’t. Comprehension has no equivalent finish line, yet Cella’s students were reasoning as if it did. They had absorbed the ontology of decoding and applied it to meaning-making, where it is necessary but insufficient.

Then, like teachers who experiment in their practice usually do, she reports back to her students that reading is a lifelong, developing, cognitive act, not a mastery skill locked in by third grade. She has them build reading metaphors, read new texts aloud together in real time, and generate multiple interpretations of the same story from different vantage points.

This is picks-and-shovels work being taught as if it were scholarly, judicial work instead. Cella is not proving anything. She is documenting what happens when you tell adolescents the truth about comprehension: that it is interpretive, effortful, and never quite finished. She watched their self-concept as readers change in response.

What Veteran Researchers Know About the Effects of Reading Assignments in College

Cella's classroom sits at one end of a trajectory that recurs across cohorts, not within one. Her students are still forming their beliefs about reading — still young enough to be told, and to half-believe, that comprehension is a light switch rather than a maturing academic capacity. Gorzycki, Desa, Howard, and Allen (2019)2 studied a different population entirely, six years earlier, but one occupying the later position on that same developmental arc: undergraduates whose beliefs about reading had already hardened into habit by the time they reached a lecture hall.

Their 2019 study, published in the Journal of Adolescent & Adult Literacy, carries perhaps more institutional weight than Cella’s. It surveyed undergraduates across 23 classes at a California State University campus, spanning business, biology, philosophy, engineering, history, and education. The sample produced 206 open-ended comments and 262 theme-comment pairings. This is not one teacher’s clipboard. It’s a coded, cross-disciplinary dataset, published in a peer-reviewed journal with far greater reach than English Journal.

The method rests on two instruments. The first is the Independent Assessment of Reading Proficiency, or IARP — a locally developed reading comprehension test, externally validated as comparable to the kind of standardized test seen on the SAT or ACT. The IARP gives the researchers an objective number signifying how well each student actually reads, independent of what that student believes about their own reading.

The second is the Perceptions of Reading Education and Proficiency questionnaire, or PREP, a ten-item survey derived from the existing academic-reading literature, asking students what they believe about the value, practice, and pedagogy of college reading. The PREP captures the subjective side: what students think is true about reading, regardless of whether it is.

Administered together, the IARP and PREP let the researchers speak to something Cella’s quiz couldn’t. They could cross-reference students’ beliefs against roughly measured skill. A student’s self-assessment — “I’m a strong reader” or “I struggle with this material” — could be checked against an actual score, rather than taken on faith.

That cross-reference produced what is to me the study’s most useful finding. Students with higher IARP scores rated their own reading skills favorably. Students with lower IARP scores rated their skills unfavorably. Self-assessment tracked measured ability closely. Students, it turns out, know roughly how well they read. This finding corresponds with student self-placement assessments in use in scaffolded undergraduate writing programs.

What they don’t know — or rather, what they’ve been taught to misunderstand — is what reading is for. The open-ended comments, coded through standard qualitative protocol and checked by two independent raters, surfaced six recurring themes, indicating convergences across the group in terms of what was on their minds in the qualitative data:

reading is important,
reading is needed for good grades,
self-assessed reading skill,
ideal timing for reading instruction,
appropriateness of reading load, and
effectiveness of reading assignments.

Reading between the lines, the data show understandings that reading and grades are linked, that they understand their capacity and don’t need to be told by an assessment office, and that circumstances are less than optimal when it comes to reading assignments in college.

Fifty of fifty-eight comments coded under the theme “effectiveness of reading assignments” agreed that the assignments are ineffective. Students weren’t rejecting reading. They were rejecting the version of reading college had mapped out for them — reading with little direct relevance to classroom use, assigned without follow-through, tested for retrieval rather than taught for understanding.

The IARP confirms these students can read. The PREP confirms they’ve concluded reading doesn’t matter much anyway because nothing in their education asked them to use it as anything other than an input for a grade. Cella shows the myth taking root; Gorzycki et al. shows the tree it grows into.

The Writing on the Wall

Full disclosure: When I worked as the university assessment coordinator at California State University, Sacramento, during a period of angst as the date of the visit from the institutional accrediting body loomed circa 2010, I surveyed university professors and interviewed them by phone to develop an understanding of how they used reading assignments in the work.

They told me, first, that they didn’t believe their students could read well, and they attributed that to a failure of high schools. Emerging from that assumption, they believed that they couldn’t rely on students’ doing the reading; as a result, they couldn’t plan in-class activities which depended on a fulsome understanding of a reading.

They also told me that reading assignments were supplements, not mainstays, and that many students did not purchase the textbook and used copies on reserve in the library when they absolutely had to.

I also interviewed students who corroborated these themes. What was different from the student input, however, was the sense of swimming in the ocean when they viewed reading assignments. Assignments were listed in the syllabus calendar. They were sometimes mentioned in class, often as supplementary reading, and I got the impression that they were floating in a void, obligatory for the syllabus, expendable in practice.

Again, none of this information with the exception of Gorzycki et al. (2019) reaches the standard of top-tier academic scholarship. Instead, it comes from the trenches in the form of recollections and classroom teacher research.

When I learned about the development of the AACU VALUE Rubrics for highly valued undergraduate learning outcomes, I was aghast to discover that there was no rubric for reading among the collection, though there was one for writing.

Susan Albertine, a strong leader from AACU in the trenches who facilitated rubric development, was receptive to my offer to serve on a team to write the VALUE Rubric for Reading, which exists today in real life. Susan and I assembled a team, and we wrote what I think is a useful document.

Enter AI

AI may be the tool that finally closes this gap. If the evidence is right, students aren’t doing the readings anyway — not because they can’t, but because the readings sit outside the actual work of the classroom, floating and expendable. AI changes that calculus.

A student can now bring a dense article into an AI conversation, ask it to discuss a difficult passage, generate discussion questions, or test their own summary against the text before class begins.

Reading stops being a solitary act performed alone without support or expectation against a syllabus deadline and becomes something closer to what Cella’s students needed all along: a conversation, a rereading, a chance to be wrong on the first pass and right on the second.

Instructors could design activities that assume this scaffolding is happening, rather than assuming it isn’t and defaulting to lecture. This is not a replacement for reading. It’s a bridge to it, the kind of support the professors I interviewed in 2010 wished they could offer but had no time to build by hand.

Used well, AI could move reading from the margins of the syllabus back into the center of the room, where Cella’s scuba divers and Gorzycki’s disillusioned undergraduates might finally meet.

Cella, Lorraine. “Student Misconceptions of Reading: What It Means to Really Read: Educators need to dispel common reading myths and foster flexible, positive attitudes toward reading as an intellectual engagement, rather than letting students succumb to defeatist mindsets.” English Journal, vol. 114, no. 4, Mar. 2025, pp. 18+. Downloaded from the CSUS Library.

Gorzycki, M., Desa, G., Howard, P.J., & Allen, D.D. (2019). “Reading is important, but I don’t read”: Undergraduates’ experiences with academic reading. Journal of Adolescent & Adult Literacy, 63(5), 499–508. https://doi.org/10.1002/jaal.1020

How Many Types of Thinking Are There?

Terry Underwood, PhD — Sun, 05 Jul 2026 00:23:00 GMT

Robert Epstein’s widely cited argument has convinced thousands of people that the computer metaphor for the brain is wrong on several fronts. The brain has no software, no algorithms, and no manufactured processors. Unlike hard drives, human memory does not store or retrieve static data; rather, it reconstructs experiences dynamically through environmental interactions.

Epstein claims the false analogy assumes that because computers process information to act intelligently, human brains must do the same to achieve intelligence. Accordingly, this flawed reasoning restricts progress in artificial intelligence and neuroscience by forcing researchers to fit dynamic, biological cognition into rigid, software-based computer models. The rise of Large Language Models has only deepened our cultural obsession with this false analogy.

He goes further. Newborns don’t enter the world with any of the concepts a computer needs to even begin. We’re not born storing information; we’re born as a living organism shaped continuously by interaction with an environment. The computer metaphor persists mainly because it’s the dominant technology of our era, not because it maps onto what brains actually do.

There is no question that human brains think. While some philosophers argue that thoughts have their own unique phenomenal properties, others claim that the experience of thinking is just a combination of other familiar conscious states like sensory inputs, inner speech, or mental imagery. Essentially, the subjective experience of thinking is a personal, conscious awareness of an idea, shaped uniquely by emotions, background, and internal mental activity.

Why, then, do we have so many different kinds of thinking that we talk about as educators—everything from mathematical thinking to design thinking? Why do we so readily believe that language machines can think? If the brain is not a computer processing a single type of 'data,' then the ways we interact with the world must be equally varied. Why, then, do we talk about 'thinking' as if it is one unified software program?

Why So Many Kinds of Thinking?

A moment’s reflection suggests that we recognize a plethora of different kinds. Take historical thinking, the kind that happens when we look back to portray an accurate, defensible understanding of happenings in the real world in the past. We can also understand happenings in the past through literature, however, despite the fundamentally imaginative nature of fiction requiring creative thinking. Indeed, creative thinking happens during thinking about the real past and the literary past, but it happens differently.

Some types of thinking have become highly formalized and structured in education. Critical thinking in its mainstream form coordinates evidence toward a defensible conclusion. Ethical reasoning assumes almost the opposite, though it deals in evidence and conclusions. Multiple ethical frameworks can be internally coherent, e.g., deontology vs utilitarianism, and still collide, and no amount of better evidence repairs the collision.

Historical thinking works with evidence that can never be re-run. No one can experiment on the past, only reconstruct it from what survived, which makes it an exercise in inference under permanent incompleteness. Political and civic reasoning add a constraint neither science nor ethics requires alone: the answer has to be reached with other people who don’t always share premises.

Creative thinking runs in the opposite direction of critical thinking toward generating options rather than narrowing them. Religious thinking, whatever else it is, routinely operates on claims that aren’t falsifiable by design, which is precisely what critical thinking can’t tolerate and religious thinking is built to hold.

These words for thinking are the vocabulary humans have built over centuries for the different kinds of contact we have with the world. Schools don’t install these as software because there’s no chip in the brain. They cultivate them the way you’d cultivate a skill embedded in practice through repeated encounters with the specific kind of problem each one answers to, in the company of people who already do it well.

What We Can Learn from Four Pairs of Thinking

Critical thinking and ethical reasoning look like siblings and aren’t. You can see this clearly by looking at how universities assess them. Washington State University’s critical thinking rubric scores work by identifying and focusing on the issue, problem, or question, with descriptors running from absent through mastering. That’s a convergence model: you chase the issue down, weigh the evidence, and land somewhere defensible.

The AACU Ethical Reasoning VALUE rubric takes a very different shape. Its first criterion isn’t about the world at all but about the reasoner. Capstone-level “Ethical Self-Awareness” requires a student to discuss in detail both their core beliefs and the origins of those beliefs with depth and clarity. No critical-thinking rubric asks a student to explain where their own convictions came from.

And where WSU rewards resolving a question, the ethical reasoning rubric’s top tier rewards a student who states an ethical position on a circumstance, states the objections to it, and can reasonably defend against those objections. Critical thinking assumes the evidence from the world eventually settles an argument. Ethical reasoning assumes the world never settles moral questions and scores you on how well you stand your ground anyway.

Mathematical thinking and scientific thinking get treated as cousins in STEM, but the field’s own theorists draw them apart carefully. Alan Schoenfeld whose 1992 chapter title is literally “Learning to Think Mathematically” defines problem-solving as confronting a situation without a ready answer, not merely completing exercises with known procedures, and argues that students entering a discipline aren’t just learning facts and procedures, they’re developing disciplinary habits of mind that let them use knowledge productively. He thinks the same habits-of-mind logic that governs mathematical thinking governs thinking like a scientist. But the habits themselves diverge sharply once you look at what each discipline treats as its unit of proof.

Deanna Kuhn’s scientific-thinking work locates the core skill in differentiating and coordinating theory and evidence, that is, an empirical operation, answerable to a world that can push back with disconfirming data. Mathematical thinking answers to no external world at all. A mathematical proof is valid or invalid by internal logical consequence; nothing in nature can falsify a correctly derived theorem the way an experiment can falsify a hypothesis.

Scientific thinking is vulnerable to plausibility bias — evaluating evidence by assessing the plausibility of a hypothesized cause, making judgments about the world in ways that are consistent with prior belief — precisely because it’s in constant contact with a messy empirical world that admits of prior expectation.

Mathematical thinking has no analogous contamination risk, because it isn’t testing a claim against nature; it’s testing a claim against a closed deductive system. They’re both called “thinking,” but one is a dialogue with the world and the other is a dialogue with a formal structure.

Religious thinking and mathematical thinking turn out to be near-opposites once you look at what counts as legitimate input. The cognitive science of religion, associated with researchers like Justin Barrett and Pascal Boyer, treats religious cognition as running on the same general-purpose mental architecture as everything else, but tuned toward specific kinds of representations. Boyer’s minimal counterintuitiveness theory holds that religious concepts transmit effectively because they sit in a cognitive optimum, intuitive enough to be easily represented and communicated, but counterintuitive enough to be attention-grabbing.

A god who’s mostly a person but violates one or two physical expectations. e.g., invisible, omniscient, is cognitively “sticky” in a way a fully mundane person or a fully alien entity isn’t. That’s a theory about what makes an idea memorable and transmissible across generations, a sociological and cognitive-ecology account, not a validity criterion.

Mathematical thinking has zero interest in memorability or intuitive stickiness; a proof doesn’t get truer by being easier to retell around a fire. Where religious thinking’s “minimally counterintuitive” concepts thrive specifically because they bend just one rule of ordinary experience, mathematical thinking discards a claim the instant it violates even one rule of the deductive system it lives in. One domain rewards controlled violation of intuition; the other tolerates none at all.

Creative thinking and magical thinking share surface features. Both traffic in the impossible, the counterfactual, the not-yet-real — and diverge completely on what happens to that material afterward. Torrance’s creative thinking framework, building on Guilford’s divergent production, defined creativity as the capacity to detect gaps, propose various solutions, produce novel ideas, recombine them, and intuit novel relationships between ideas, operationalized through fluency, flexibility, originality, and elaboration.

Creative thinking on this model is instrumentalized: the unusual idea is a means to a usable end, whether that’s a poem, a product improvement, or an unconventional use for a paperclip. Magical thinking, in Woolley’s widely cited definition is reasoning based on some misconception about causality or about natural laws more generally. It isn’t a tool aimed at a novel output; it’s a belief about how the world actually works, held well into adulthood because it performs real psychological work, coping with unsolvable problems, tolerating uncertainty.

A child playing with an “impossible” premise in a story is doing something structurally identical to what a novelist does; a child who genuinely believes wishing can move an object is doing something else, and Piaget’s original account treated it as an immature stage meant to be outgrown by around age twelve.

The line between them is intent and disposability. Creative thinking generates the unlikely, possibly impossible, and finds a way to repurpose it for this world; magical thinking holds onto the impossible as a genuine account of causation, often precisely because giving it up would cost something.

What holds the four pairs together?

We will look at four pairs of thinking types to illustrate how one might begin to lay these cognitive activities side by side for systematic analysis. In each case, the apparent sibling turns out to answer a different question about the world:

Critical thinking asks what’s true; ethical reasoning asks what’s defensible when multiple truths compete.

Mathematical thinking asks what follows necessarily from a closed system; scientific thinking asks what an open, noisy world will let you get away with believing.

Religious thinking asks what makes an idea worth transmitting to your children; mathematical thinking couldn't care less about generational inheritance.

Creative thinking asks what else might be possible and then puts the possibility to work; magical thinking picks the possibility up and holds onto it, even when it doesn’t work.

None of them are lesser or better versions of “thinking in general.” They are easy to conflate by teaching critical thinking, for example, as if it includes ethical reason, too, or “STEM thinking” as if math and science are one skill. The important point for task design is this: A task can emphasize one while still activating others. Another task can activate a whole suite of types.

Implications for Task and Lesson Design

Naming these varieties of thinking is not an academic exercise; it is a design constraint. Every task we assign implicitly assumes a theory of mind, and when that theory is too narrow, the task inherits the narrowness. This is the danger of “hardening of the categories”: we teach a rubric labeled “critical thinking” and unconsciously believe we’ve covered ethics, creativity, and inference under one roof.

We treat reading and writing as if comprehension and composition were algorithmic pipelines with a single correct route in and out. But real texts routinely demand historical inference, ethical judgment, and creative synthesis in the exact same paragraph.

AI exposes this flaw at scale. Because AI is an algorithmic system, it can convincingly simulate any task we’ve reduced to a closed procedure. Yet it cannot inhabit the theory-evidence coordination Deanna Kuhn describes, the self-implicating stance that ethical reasoning demands, or the disposable counterfactuals of genuine creativity. These are not procedures at all; they are different kinds of contact with the world.

The remedy is not teaching thinking “in general.” Instead, we must design tasks that are deliberately layered so that several of these distinct cognitive stances must fire together—an entanglement no single algorithm can shortcut. A pedagogy built on that premise treats complexity honestly, instead of flattening it into whichever thinking vocabulary happens to be fashionable—or on the test.

Between a Rock and a Hard Place: Critical Thinking and Persuasive Writing

Terry Underwood, PhD — Thu, 02 Jul 2026 19:23:38 GMT

We know that writing is thinking. For that reason, writing is a staple assignment in academic classrooms.

What about speaking? Is speaking thinking, too?

We likely agree that speaking in an academic setting is thinking, but it’s a different way of thinking. Obviously, different ways of thinking are invoked even when speaking, depending on levels of formality. And there are differences in thinking during speaking events depending on whether one is listening or speaking.

What about critical thinking? Mainstream views assume that asking students to write persuasive essays gives them opportunities to practice and improve critical thinking skills and strategies.

Does this assumption hold up under critical analysis?

Back to Basics

Writing is thinking with delay according to Janet Emig and other seminal researchers. You produce a sentence, it sits there, you look back at it, and you revise not just the wording but the thought itself. Didion’s line captures it: she writes to find out what she thinks and what she sees.

The externalization is what scaffolds the cognitive work. The sentence on the page becomes an object you can interrogate, contradict, and rebuild. That loop — produce, inspect, revise — is mostly unavailable in real time. The writer in solitude traverses this loop relentlessly.

Speaking is thinking in motion. It’s linear like writing but it’s irreversible: once the word is out, you can’t unsay it, only add a correction after the fact (”what I mean is—”). That constraint actually produces its own kind of thought.

Speaking forces you to generate language before you’re fully certain, and that requirement under time pressure surfaces surprises. Incidentally, on this account, synthetic text generated by a bot is not writing; it is speech.

Ask someone to explain their position out loud and they’ll often discover a hole in it mid-sentence. That’s thinking happening, but happening forward, not backward the way revision does.

Coincidentally, synthetic AI text is filled with holes that make little sense, sometimes saying again what it just said in slightly different words in search of a better statistical match.

There’s also the interlocutor. Speaking is usually thinking with someone, not just thinking. A conversation partner interrupts, pushes back, asks a real-time question you didn’t anticipate — the thought gets shaped by resistance from another mind, which is dialogic while writing is too often monologic, leaving little space for a reader.

AI can simulate thinking with someone, but that someone is digitally synthesized and simulated as well. It’s troubling to categorize AI interactions as either dialogic or monological in a human sense. Today, I incline toward monological. The AI delivers monologue in response to dialogue.

A more honest appraisal suggests that writing and speaking each activate a good deal of thinking, both plain vanilla and critical. With AI, we have a hybrid somewhere between writing and speaking with its own affordances and constraints on human thinking with a technical ability to simulate human discourse patterns.

Does thinking with the intention of persuading an audience in writing—a type of thinking we are fond of calling “critical”—develop critical thinking? It would seem to have all the ingredients. Writing scaffolds revision by fixing thoughts for review and reconsideration. Conjuring a reader who may disagree ought to scaffold a foothold for looking again at ideas from a different perspective.

Is the delay effect writing affords coupled with the rhetorical expectation enough to require and enhance the performance of critical thinking?

An Empirical Study Says No

Indeed, Washington State University's study of the writing-critical thinking connection published in 2004 (see reference list) suggests a more complex relationship.

At WSU, researchers evaluated the same student writing samples first for writing quality and then for critical thinking, using two separate, carefully developed rubrics. They found not just a weak relationship between the two but, in some cases, an inverse one.

The better the writing scored, the lower the score on critical thinking.

To explain this finding, the researchers theorized that students must have learned to produce writing that satisfied typical assignments without engaging the criticality the assignment was assumed to evoke—at least the evidence left behind in the essays didn’t meet the burden of proof. For a complete picture, there is no substitute for reading the study.

My Take on the Study

When I first read this study almost two decades ago, I explained this finding of misalignment between assignment design and expected outcome as a function of what teachers usually teach students about when and how a thesis statement gets written and begins to govern the text. The moment the thesis appears in the writing process, the writer shifts from critical thinking to making an artifact.

It’s not ok to take a provisional position. The rubber has to hit the road, and the evidence discussed must be convincing. The incentive for the writer shifts from ruthless interrogation of the material to taking up or leaving behind evidence that doesn’t comport with the thesis.

Identifying a thesis and then sticking with it through completion of the writing task minimizes the need for revising and rethinking. When a thesis keeps changing as new evidence comes in, the text has to change, and that, for a student on the clock with a due date, is bad news.

Critical thinking means keeping the question open. Argumentative writing means closing the question.

Critical thinking means admitting you don’t yet know for sure. Argumentative writing means you are convinced, and anyone who reads your essay is ideally convinced as well.

Critical thinking means you operate from working assumptions which are not set in concrete and are always open to question. Argumentative writing means decorating the concrete.

This finding gets lost in discussion of writing and critical thinking with significant implications for what AI has changed in the ecology of learning. It tells us much of importance about writing assignment design.

An AI can quickly generate a 1500-word argumentative essay arguing that God does not exist; in the next chat window, it can generate one arguing that God must exist. The student writer need only cut and paste and pretty up the essay with some personal touches.

Imagine you are teaching an undergraduate course in public health with a writing requirement.Consider the difference between "a) write an argumentative essay about the quality of student mental health services on campus as a resource for students and suggest ways to improve" and "b) investigate the quality of student mental health services on campus and make evidenced-based recommendations for easy redesigns that better support students."

The first assignment can probably be gamed by an AI. A student could collect a set of documents from the Office of Institutional Research which likely has records from student satisfaction surveys. A search of the student newspaper, also online, can provide local color and detail. You want God doesn’t exist? You want God does exist? Flip a coin. Upload some data and discuss with AI. Upload some local media coverage. Write a prompt and push “execute.” Upload the rubric or whatever criteria the assignment came with.

The second assignment cannot be completed by collecting and summarizing data from the most recent campus survey of student satisfaction. The writer must visit student mental health services with a critical mindset to learn firsthand, to talk with students working there for work-study, to take notes, to weigh observations against criteria, to talk to mental health advisors, and to take a position they can be held accountable for through questioning and push back.

This doesn’t mean they can’t use AI as a resource to locate local and comparative data, to locate and access texts that describe student mental health services on other campuses. It doesn’t mean they can’t stress test their thinking as they work through the process by prompting AI to generate alternate possibilities.

WSU's contribution can be read as a narrow finding about writing assessment and the common use of student writing to assess critical thinking. It's also a framework for understanding what kinds of assignments hinge on critical thinking not as preparation for a rhetorical performance but as participation in an epistemic experience in the social, cultural, and physical world that could be expressed in an essay, a video, a multimedia presentation.

Critical Thinking On Bloom

Bloom published his Taxonomy of Educational Objectives in 1956 (refer to the section in the reference list for sources). He set out to solve a measurement problem. Bloom was not given the task to “develop a thesis that argues for a logical model of higher-order thinking.” He needed a model of levels of thinking to solve a pressing sociological and educational problem.

His immediate practical concern was that college examiners across institutions had no shared language for describing what kind of thinking a test question required, and test scores were known to be somewhat reliable and valid for trivial details but not for mature reasoning. A question asking students to recall a date and a question asking them to evaluate a policy decision were both called and tested as "thinking," which made assessment incoherent and curriculum design a patchwork of assignments with unknown worth.

His solution was hierarchical and explicitly mental: knowledge, comprehension, application, analysis, synthesis, evaluation — six levels ascending from simple to complex. The higher levels were what we inherited as critical thinking. It is a cosmic irony that education spent the next seventy-five years trying to straighten out Bloom so that it wouldn’t be just a testing mechanism despite the fact that testing mechanism was what it was indeed intended to be.

Crucially, Bloom conceived all six levels as operations performed in the head on symbolic content which is now easily manipulated by AI; there was no social and cultural substrate in his model in which higher-order thinking is embodied and embedded. There was no such thing as performance assessment in his historical formation.

You analyze and weigh the content of texts and, at higher levels, either analyze, synthesize or evaluate, which are the higher-order skills. The object of the cognitive operation is always representational, and the evidence of success is verbal.

Above all, Bloom was after a construct of higher-order thinking that could be measured on a mass scale.

Bloom’s Real-World Example

It’s not that Bloom was disinterested in the real world—it’s just that he was interested in this project in the world inside a head. His own intellectual practice is a standing rebuke to his taxonomy's symbolic bias.

To develop the taxonomy, Bloom didn't sit alone reading, taking notes, manipulating propositions. He convened a working group of examiners over several years, arguing across disciplines, testing categories against real classroom situations, revising when the framework failed to describe what teachers and students actually did.

The taxonomy emerged from sustained disagreement among people with different disciplinary investments and institutional contexts. That's the WSU rubric's recognition of alternative perspectives and assessment of assumptions operating in full.

His Mastery Learning research required him to go into actual schools, observe actual students, measure actual outcomes, and revise his theoretical commitments when the data pushed back. He found that most students could achieve at high levels given sufficient time and appropriate instruction — a conclusion that contradicted the bell curve assumptions that dominated educational measurement at the time. Reaching it required him to question his own field's foundational premises, which is precisely what the WSU Capstone descriptor for assumptions demands.

In other words, Bloom performed critical thinking as WSU would define it — embedded in context, accountable to evidence, willing to reexamine assumptions — not as symbolic logic would define it. His intention, however, was to produce a written textual framework that inadvertently defined critical thinking as a clinical mental mechanism.

WSU's critical thinking rubric, by contrast, includes context, assumptions, and implications as distinct dimensions, operations that require a thinker embedded in a social and cultural situation, not just processing verbal representations of one.

U.S. schooling took the wrong fork in the road just as it is taking the wrong fork today regarding phonics instruction.

Some Analytical Observations for the Curious

Critical thinking as the WSU rubric operationalizes it is an opening operation. It keeps the question open as long as possible. It treats in-coming evidence as a reason to question the question, not merely to answer it. These two imperatives are genuinely in tension: a thesis that changes as new evidence arises demands rewriting, which is precisely what the pressure to produce a finished, coherent text discourages.

That embeddedness and embodiedness is not incidental; it is constitutive of the thinking itself. WSU’s treats context and assumptions as separate dimensions requiring active engagement, dimensions that demand a situated thinker, not merely a processor of verbal representations. The difference between a critical thinker and a critical thinker-in-context is fundamental.

The threat AI poses is specific. It is not a threat to all writing assignments. It is a threat to writing assignments that mistake text production for critical thinking of the fully human variety, assignments where the goal is simply to process information reproduced in sources, footnoted if necessary, and produce a coherent textual argument in response rather than think critically about a messy situation.

There is nothing wrong with writing persuasive text. Just don’t conflate a rhetorical purpose with critical thinking, which is not fundamentally about winning an argument.

That category, unfortunately, describes the majority of undergraduate writing tasks if AI success in doing the assignments is considered. College Board research published in February 2026 found that nearly three-quarters of faculty report students using AI to write essays or papers, and over 84% agree that AI reduces students’ critical thinking. Is it AI or the nature of the assignment?

This is the real contribution of the WSU framework, brought into full relief by the emergence of generative AI: it distinguishes assignments that are designed around critical thinking from assignments that are designed around text production.

What the WSU data, Bloom’s intellectual practice, and the emergence of AI collectively illuminate is a fork in the road of assignment design. One path treats writing and critical thinking as interchangeable and assigns the former expecting the latter. That path is now structurally vulnerable.

The other path recognizes critical thinking as a mode of interrogation — active, intentional, inquisitive, and situated — that is irreducibly the student’s own, and that may conclude in writing, not always strictly persuasive and logical. That path remains, for now, AI-resistant in the only meaningful sense: not because AI cannot write the resulting text, but because AI cannot perform the prior act of thinking that the assignment requires.

Bloom arrived at his taxonomy through the very kind of inquiry his taxonomy was unable to theorize: working groups across disciplines, real classrooms, evidence that pushed back against the field’s foundational assumptions.

That practice — embedded, argumentative, revisionary, accountable — is the model that offers AI a rebuttal. The assignment design challenge is to make Bloom’s practice, not his taxonomy, the basis for the assignment, and let writing be what it always was at its best: the vibrant record of compelling thought that has already happened and composed for an audience.

The WSU Writing and Critical Thinking Study

Condon, W., & Kelly-Riley, D. (2004). Assessing and teaching what we value: The relationship between college-level writing and critical thinking abilities. Assessing Writing, 9(1), 56–75.
Full text: https://eric.ed.gov/?id=EJ803627

The WSU Critical Thinking Rubric

Washington State University Center for Teaching, Learning & Technology. (2009). Guide to Rating Critical & Integrative Thinking.
University of Colorado Boulder mirror: https://www.colorado.edu/sei/sites/default/files/attached-files/wsu-critical-thinking-rubric-2009.pdf

Eastern Illinois University. Adapted Critical Thinking Rubric (from AAC&U VALUE Rubric and WSU).
https://www.eiu.edu/assess/Critical%20Thinking%20rubric%20reader-enabled-updated%202022-03-29.pdf

WSU Critical Thinking and the Humanities. Washington State University archived resource on rubric dimensions and sequencing.
https://public.archive.wsu.edu/gened/public_html/orpheus/critical_thinking.html

Condon, W., Carris, L., & McCracken, V. (2003). Fostering Student Critical Thinking and Writing in the College of Agriculture. Presented at the Wyoming Conference on English.
https://wacclearinghouse.org/repository/professional/conferences/wyoming2003/condon/

Bloom’s Taxonomy of Educational Objectives

Bloom, B. S., Englehart, M. D., Furst, E. J., Hill, W. H., & Krathwohl, D. R. (1956). Taxonomy of Educational Objectives: The Classification of Educational Goals, Handbook I: Cognitive Domain. New York: Longmans, Green.
SERC resource page with description and access:
https://serc.carleton.edu/resources/868.html

Cornell University Center for Teaching Innovation. Bloom’s Taxonomy — levels, verbs, and learning objectives.
https://teaching.cornell.edu/resource/blooms-taxonomy

University of Illinois Chicago, Center for the Advancement of Teaching Excellence. Bloom’s Taxonomy of Educational Objectives.
https://teaching.uic.edu/cate-teaching-guides/syllabus-course-design/blooms-taxonomy-of-educational-objectives/

Krathwohl, D. R. (2002). A revision of Bloom’s taxonomy: An overview. Summarized at NIH/PMC:
https://pmc.ncbi.nlm.nih.gov/articles/PMC4511057/ — Bloom’s taxonomy of cognitive learning objectives, NLM review

Bloom’s Mastery Learning Research

Guskey, T. R. Revisiting Bloom’s Learning for Mastery: Closing Achievement Gaps.
https://tguskey.com/wp-content/uploads/Mastery-Learning-5-Revisiting-Blooms-Learning-for-Mastery.pdf

Slavin, R. E. (1987). Mastery learning reconsidered. Review of Educational Research, 57(2), 175–213.
https://journals.sagepub.com/doi/10.3102/00346543057002175

PMC/NIH. A Practical Review of Mastery Learning — contemporary synthesis.
https://pmc.ncbi.nlm.nih.gov/articles/PMC10159400/

Critical Thinking as Inquiry: Theory and Pedagogy

Battersby, M., & Bailin, S. Inquiry: A New Paradigm for Critical Thinking. Inquiry: Critical Thinking Across the Disciplines.
Full text: https://windsor.scholarsportal.info/omp/index.php/wsia/catalog/download/54/79/1254?inline=1

University of East Anglia. Critical Thinking and/or Argumentation in Higher Education (chapter on the distinction between argumentation and inquiry).
https://research-portal.uea.ac.uk/files/97394258/Chapter.pdf

AI, Writing Assignments, and Academic Integrity

College Board (2026, February). College Faculty Perceptions of Generative Artificial Intelligence in Higher Education: Faculty express near-universal concern that student AI use undermines original writing and critical thinking.
https://newsroom.collegeboard.org/new-college-board-research-faculty-express-near-universal-concern-student-ai-use-undermines

Cotton, D. R. E., Cotton, P. A., & Shipway, J. R. (2024). Chatting and cheating: Ensuring academic integrity in the era of ChatGPT. Innovations in Education and Teaching International, 61(1).
https://www.tandfonline.com/doi/full/10.1080/14703297.2023.2190148

Bearman, M., et al. (2026). Embracing a new world: Authentic assessment designs in an age of generative artificial intelligence. Assessment & Evaluation in Higher Education.
https://www.tandfonline.com/doi/full/10.1080/13603108.2025.2601741

Draper, M. J., & Newton, P. M. (2026). Resilient assessment in the age of AI: Authentic design and the case for verbal examinations in business education. Assessment & Evaluation in Higher Education.
https://www.tandfonline.com/doi/full/10.1080/02602938.2026.2644516

Cornell University Center for Teaching Innovation. AI & Academic Integrity — prevention, detection, and assignment redesign strategies.
https://teaching.cornell.edu/generative-artificial-intelligence/ai-academic-integrity

Abbasi, A., et al. (2026, January). Comparing Assignment Description Intent with AI-Generated Results. IIIS Proceedings.
https://www.iiis.org/CDs2025/CD2025Summer/papers/AA136YF.pdf

Al-Zyoud, M., et al. (2024). AI and Academic Integrity: Exploring Student Perceptions and Behaviours. University of North Texas Computational Humanities & Information Literacy Lab.
https://ci.unt.edu/computational-humanities-information-literacy-lab/aiandai.pdf

Taddeo, M., & Woollard, J. (2024, May). AI, the Conflicted University, and the Classroom. Journal of Philosophy of Education, 60(3).
https://academic.oup.com/jope/article/60/3/522/8249758

Reclaiming the writing process using AI: A scaffolded writing assignment that promotes critical thinking, reflection, and accountability. (2026). Journal of Microbiology & Biology Education.
https://pubmed.ncbi.nlm.nih.gov/41874179/

LLMs, Embodied Cognition, and the Limits of AI Reasoning

AI scientists produce results without reasoning scientifically. (2026, April). Hugging Face Papers — LLM-based scientific agents across eight domains.
https://huggingface.co/papers/2604.18805

GFB Group. (2026, May). AI Isn’t Replacing Critical Thinking — It’s Making It More Important Than Ever.
https://www.gfbgroup.com/post/ai-isn-t-replacing-critical-thinking-its-making-it-more-important-than-ever

The Science of Reading Goes to Washington

Terry Underwood, PhD — Mon, 29 Jun 2026 22:29:08 GMT

I recently spent an afternoon trying to get an AI to argue against the Science of Reading (SoR) movement. Sometimes, you can discover flaws in thinking when you do this; usually, for me, I sharpen my understanding of the evidence by exploring weight-bearing chunks of material I see as significant for accuracy and nuance. I don’t expect the AI to produce valid evidence; I expect fabrication. So it’s not a strategy I deploy unless I have a pretty good grasp of the underlying evidence.

It was maddening. I forced Claude 4.6 with web access disabled to rely on its original training, which I knew from past experience is heavily biased in favor of SoR. The machine apologized for this limitation and pleaded with me to let it search the web. I did not activate the option because I wanted to know what this machine had learned from the wild of the internet when it was trained about the matter. It had learned the SoR simple view as gospel. The bot could not argue against SoR.

This summer, the push to mandate SoR principles has crescendoed and galvanized wide and deep bipartisan support at both the state and federal levels. Phonics isn’t a wedge issue at the ballot box. While education reform can be fiercely partisan, reform isn’t crisis, and lawmakers share common ground in addressing what they are calling the “nation’s literacy crisis” by shifting away from balanced literacy and three-cueing models toward direct, explicit, systematic phonics instruction in the early grades.

In the House of Representatives, the Science of Reading Act of 2026 passed the Education and Workforce Committee with unanimous, bipartisan support in March 2026. House members were feeling virtuous about their good works, pontificating on the “left behind” theme, adjacent to NCLB:

“Reading is the foundation of every student’s success, and right now, too many children are left behind. Parents, teachers, and students deserve better. The Science of Reading Act gets us back to the basics and ensures our classrooms are equipped with the proven, phonics-based methods to effectively teach all students to read.” – Rep. Houchin (IN-09)

In June, a bipartisan Senate coalition introduced the Reading Excellence and Achievement for Development Act — the READ Act — described as the largest federal reading reform proposal in years. Both the House and the Senate bills require states receiving federal literacy grants to align instruction with the science of reading, and both explicitly prohibit the three-cueing model.

The immediate link between the SoR and these federal bills runs through Alabama, which began its infatuation with phonics in 2013. A one pager posted in support of a recently proposed Alabama law argued that the state needs to rid itself of unbalanced balanced instruction: "Three-cueing is a disproven and incorrect instructional approach that teaches guessing, not reading."

The resulting Three Cueing Ban (SB168) went into effect on June 1, 2026. The one pager is coupled with Emily Hanford’s Sold a Story Podcast for background information on why the law was necessary. Lawmakers are convinced that there is but one way to teach a child to read. Someone has proven it to them.

The Pedagogical Reality

I invite you to read the Alabama law itself. It’s very short. The definition of three-cueing is the bill's exposed nerve. Section 2(6): "THREE-CUEING SYSTEM. Any model of teaching students to read based on meaning, structure and syntax, and visual cues, which may also be known as MSV."

Essentially, the law strips early reading instruction of meaning and form and demands that all beginning students attend exclusively to the sounds of words. It bans instruction that teaches students to attend to semantics (word meanings) and syntax (word functions). All a reader can use are visual cues, letters bound in isolated words, and the reader’s goal is to pronounce the words correctly. Alabama has defined the full act of reading as an illegal instructional model.

The bill senses its own absurdity and attempts a retreat. Section 3(a) states that instruction "…may include visual information and strategies that improve background and experiential knowledge, add context, and increase oral language and vocabulary to support comprehension, but may not be used to teach word reading."

The exemption assumes that word reading and comprehension are sequential stages with a clean boundary between them, that a reader first decodes a word and then, having decoded it, consults meaning and syntax to understand it. This is not what happens even among beginning readers unless all they have to read is a word in isolation in which case there is no syntax nor context.

It’s important to note that phonics applies to reading a single word, and viewing phonics as a simple matter of teaching letters and sounds is misleading. Words themselves carry syntactic markers, for example, tense and number. Even letters are fickle. The same letter (example: c) can represent several different sounds depending on the word in which it is used. In “cent” c represents s; in “coin” c represents k. The same two letters (example: ch) can represent three sounds: 1) chorus, 2) church, and 3) machine.

According to SoR, beginners must learn to atomize letters and letter-compounds in individual words, map them onto individual phonemes, blend the phonemes together, and then pronounce the word. Advanced phonics requires deep analysis—understanding that “ou” sounds different in "tough," "ouch," "dough," and "through."

While it is true that using context to guess or identify an unfamiliar word can disrupt orthographic mapping for beginners, this aspect of three-cueing pedagogy has been distorted, and lawmakers have flattened this cognitive science into a blunt political instrument. It is one thing to critique three-cueing as a decoding strategy; it is quite another to ban instruction that teaches students to attend to semantics and syntax. Letters and the sounds they represent require context to dial in their spoken match: “The boy ran out of the woods with tears in his eyes and tears in his coat.”

The brain is continuously generating probabilistic expectations about what word is likely to come next—just as LLMs do, taking advantage of both linguistic constraints and redundancies—based on the syntactic frame and semantic context already established. This perspective on fluency and comprehension is not a fringe position in cognitive science. It is the mainstream account of how we read—and serves as the substrate for language machines, which make no use of phonics to process input or to generate output.

The bill's drafters had access to the same mainstream literature they cite elsewhere in the bill. They cited Scarborough’s Rope in the legislation as evidence, a graphic which depicts reading as a braided rope with strands that interweave in patterned, complex ways. Hollis Scarborough herself is uncomfortable with how her rope was incorporated into SoR. She intended it to show the complexity and interconnectedness of reading.

For her graphic to show up in Alabama’s law is the height of absurdity. The bill cites the rope to justify separating what the rope was designed to show cannot be separated. It is as if someone cited the double helix to argue that DNA has only one strand.

The Elephant in the Room: Poverty

Section 4(b)(1) then delivers the phrase that appears throughout this legislative wave: instructional materials must be "based on literacy strategies that are scientifically researched with proven results." Proven results from research that studied phonics in isolation on measures of phonics knowledge and then spawns a law that bans meaning, syntax, and context from word reading instruction. A legislature sat down to protect children from bad reading instruction and wrote a law that makes teaching reading as an experienced expert an act of insubordination.

One wonders why politicians focus on such a convenient remedy and continue to neglect public schools, health care, and eliminating poverty. There is a p-word politicians are eager to say — phonics — and a p-word they will not say. The one they will not say requires policy responses that are expensive, structurally difficult, and politically unrewarding. Phonics bills are cheap, bipartisan, and make excellent press releases.

Consider the lawmakers leading this charge. Senator Mark Kelly manages to speak forcefully in favor of a ban on three-cueing without ever saying anything to suggest he understands what he is voting for. Kelly and his twin brother transferred into Kelly Elementary School in 1972. Situated just 15 miles west of New York City, the school serves a financially stable suburban community. Growing up in this secure environment, the brothers were afforded the freedom to explore and take risks without the burdens of extreme scarcity. In May 2016, the Board of Education voted to rename the school to honor them.

Senator Bill Cassidy is another important emotional center for the Senate effort. Cassidy was born in Highland Park, Illinois, a wealthy suburb of Chicago. To his credit, Cassidy spent decades working in a hospital system caring for the poor. Regarding his support for the READ Act, Cassidy stated: "Capable students are dropping out of high school simply because they were never taught how to read."

This is the kind of sentence that sounds like compassion for children, but functions as an indictment of schools that refuses to consider any other possible explanations. In sharp contrast to Kelly and Cassidy's backgrounds, high-poverty elementary schools in the American South operate under fundamentally different constraints.

You can predict a child's reading scores from their zip code with remarkable accuracy without knowing their teacher, their curriculum, or whether their classroom ever contained a three-cueing chart.

In 2024, the National Center for Education Statistics (NCES) introduced a much more precise Socioeconomic Status (SES) index. Nationally, economically disadvantaged fourth graders scored 202. Non-disadvantaged scored 229. Poor children are reading approximately two and a half grade levels behind their affluent peers, and this is true before a single instructional variable enters the picture.

Every sponsor invokes the "Mississippi miracle" as proof text for this entire legislative push: a state that climbed from 49th to 21st in fourth-grade reading scores between 2013 and 2022. The linchpin of Mississippi's narrative is the 2013 Literacy-Based Promotion Act, which required that third graders who failed the state reading test be held back. With the retention policy holding back over nine percent of third graders, the NAEP increases follow mechanically: when the lowest-scoring students are removed from the testing pool, the average rises.

One analysis found that all of the observed gain in performance can be accounted for by the retention policy. In 2022, 69 percent of Mississippi fourth graders were still not proficient in reading. If Mississippi ranks 21st with numbers that dismal, it is less a sign of success than a flashing warning about the depth of a national crisis driven largely by poverty.

Just as a doctor might believe a vaccine cures a virus, a physician-turned-politician like Senator Cassidy believes a standardized phonics mandate will "cure" illiteracy. This mindset ignores that national reading struggles are a sociological problem driven by poverty, not a disease waiting for a legislative silver bullet.

Cassidy, along with a huge bipartisan majority, is about to vote yes on banning three-cueing. Tragically, one can only hope that the Department of Education under Donald Trump is so dysfunctional that it won’t make any difference.

Correspondence in the California State University: The Syllabus Says

Terry Underwood, PhD — Thu, 25 Jun 2026 20:41:52 GMT

A California State University syllabus has, somewhere near the bottom, some language about the final exam. The system counts final-exam week as instructional time and requires faculty to hold the assigned block whether they give a test or not. The assumption is that instruction stopped at the end of the prior week; therefore, there is no new learning required nor expected.

Most professors experience this as a fact about the calendar, highly useful for large lecture GE classes and STEM disciplines, not a claim about learning. But others see it as a strong claim about learning. A finals exam week decides that whatever the course taught will become visible at the end. The course disciplines the learner to achieve mastery.

What happens when what is being taught will not hold still for that? Some learning culminates, but some learning is not an arrival at a destination. It’s a cultivated responsiveness, an epistemic drive to make and refine meaning and significance, and it has no teleological terminus.

Learning to Correspond

Tim Ingold, an anthropologist who has become famous for his theory of making, calls this phenomenon “correspondence,” the back-and-forth by which a person and a material answer or “co-respond” to each other, the way a coppersmith reads the metal’s give and resistance rather than imposing a design.

In the CSU syllabus template, which is instantiated differently at each of the two dozen plus campuses, this tension is hidden in the required adherence to a final exam mindset; it is architectural, and its place in a course is secure. Indeed, William Sandoval’s conjecture mapping — a learning science framework for making the design of a learning environment explicit and testable — runs on exactly the argumentative grammar the CSU embraces and Ingold contests.

We will explore the tension when the finite concept of mastery pushes against the infinite capacity of correspondence in this essay and consider our options.

Two Pictures of Learning

Begin with the assumption buried in the final-exam rule, because it is an assumption about what learning is, not merely when to test it. The mastery picture holds that learning is the progressive acquisition of a content, and that acquisition has a finish line. You don’t yet have the material; you study; you have it.

The course’s job is to deliver the content and then, at the end, to verify possession. The terminal exam is not arbitrary under this picture; it is the picture’s final form. If learning is the closing of a gap between ignorance and knowledge, then of course you measure at the moment the gap should be closed. Mastery is finite by definition; it can be divided and delivered in courses in a sequence. One either has or has not mastered it at an acceptable level.

This is what Ingold would call a hylomorphic account, after the Aristotelian pairing of form (morphē) and matter (hylē). The form pre-exists in the mind; it is imposed on inert material; the finished object is the form realized in the matter.

Transposed to a classroom, the learning outcome is the form, the student is the matter, and mastery is the moment the form has been successfully impressed. The conjecture runs in one direction, downward, from the plan to the result. The learner has had very little to do with it beyond completing the assignments successfully.

Camille Straatman’s elegant account of learning to raise copper vessels gives us the other picture. Straatman is an apprentice coppersmith and a geographer who refuses the hylomorphic story from its first line. She does not impose a shape on the copper.

“It is a conversation in progress,” she writes of the tumbler emerging under her hammer, the metal “willingly” giving in places and resisting in others, those very responses serving as “vital clues” in what she calls her communication with the material. The form is not pre-given and pressed in. It emerges from the back-and-forth.

Under hylomorphism the maker knows the form in advance and the material is passive. For Straatman the material is an active participant — “a vital collaborator, capable of teaching, responding and participating.” She learns to read the sound: a hollow duff means she is striking above air, giving the metal room to move inward; a sharp ding means she has hit metal directly onto the stake and is flattening rather than raising.

The copper tells her when it is “tired” — worked so hard its particle structure has hardened — and must be annealed before it cracks. None of this is information she imposes. It is information the material gives, and her skill is precisely her growing capacity to receive it. This is correspondence: not form pressed onto matter but two parties answering each other over time.

Correspondence has no terminus. Straatman is explicit that her apprenticeship is not a march toward “mastery” — she names and rejects the word — but the cultivation of an ongoing responsiveness. Her mentor Charlotte is not finished; she is more attuned, able to sense intuitively the angle at which to hold the tumbler over the stake, hearing acoustic changes Straatman cannot yet hear.

The difference between novice and expert is not that the expert has reached the end of a road the novice is still walking. It is that the expert corresponds to nuances in the material that a novice doesn’t even notice.

Attunement has no ceiling. There is always a subtler change in the metal’s voice to learn to hear. Where mastery is finite, a gap to be closed, correspondence is infinite, a relationship that deepens for as long as it continues.

The mastery picture is finite, terminal, content acquired and verified, form impressed on human brain matter, the learner the passive site of an imposed shape. The correspondence picture is infinite, processual, capacity cultivated through a back-and-forth in which the material answers, the form emerging from the encounter, the learner an active party to a relationship with no finish line.

Most of what the CSU does runs comfortably on the first picture, reaching and teaching young people who otherwise could not go to a university, and for much of what it teaches the first picture is true and impactful enough. The trouble begins only where the second picture of learning is accurate regarding the relationship between the learner and the material, where what is being learned is a responsiveness to material — textual, physical, ideational, discursive — rather than information and content, but the institution, through an instrument as ordinary as a line near the bottom of a syllabus, insists on treating all learning as the first picture.

Designing for Correspondence

Suppose you believe the second picture applies to your course, that what you are teaching is a responsiveness to material, not just material itself, and suppose you still have to submit a syllabus, hold the final-exam block, and turn in letter grades. The honest question is not how to escape the constraint but how to design inside it. Here Sandoval becomes useful in a way he very likely didn’t intend, because the same map he made that requires a pin in a destination also contains the materials for refusing to.

Conjecture mapping distinguishes two kinds of evidence that a design is working. There are outcomes, the named end states, the things you can verify at the terminus, and there are what Sandoval calls mediating processes: the interactions and the artifacts that occur along the way, the activity through which any outcome is supposed to arise.

His framework treats the mediating processes as instrumental, the means by which you reach the outcome that matters. But invert the priority. Make the mediating processes the point and the terminal outcome simply what happened in the end, and you have a design that teaches correspondence as the valued learning rather than the mastery, and grades, so far as you must grade anything, the quality of a student’s correspondence with more weight than the correctness of their outcome. The map stays intact. You have only changed which column you care about more.

What does that look like concretely? It looks like building a course around two samples of a relationship rather than a single verification at the end. Early in the term, before much content has landed, but enough has been in correspondence to count, you ask the student to formally and officially write down where they actually stand with regard to the material. Ask them to write about how they engaged with the readings they were assigned, not for an immediate grade, but for feedback, and for a benchmark, instrumental to final grading.

This is the equivalent of Straatman’s first clumsy strike, all ding and no duff, the novice hand that has not yet learned to hold the work. Then, at the end — in the very block the rule reserves for a final exam — you ask the student to set that early artifact beside where they are now and account for the distance. “Show me how you corresponded with the material and how the material talked back to you, where it gave, where it resisted, what you can now hear that you could not hear in week three.”

The two artifacts are commensurable because they probe the same genre of learning activity done by the same person at two times, and the space between them is the only thing you evaluate for a final exam grade to help determine a course grade.

Notice what this does to the constraint. The final-exam block is still used; the registrar still gets a letter; the syllabus still names a culminating task. Every institutional box is checked. But the box now contains a reckoning with a relationship between the learner and the course material instead of a verification of a content. You have not defeated the mastery picture. You have smuggled correspondence through it, wearing its clothes.

The cost is real, however. You lose the crispness Sandoval’s framework was built to provide. A pinned outcome is much more easily tested. Disciplinary attunement is harder. A lazy version of this design collapses into rewarding whoever writes the most fluent self-narrative.

The discipline that keeps it honest is the same one that keeps Straatman honest: the evidence has to be rich enough to be audible. She can claim attunement because she has the duff and the ding, the actual changes in the metal’s voice. A course pursuing correspondence has to generate an equivalent record, a trail of the student’s actual exchanges with the material substantial enough that the difference between the early sample and the late one can be evidenced rather than simply asserted.

Without that record, “correspondence” is just a vibe. With it, you have something a hostile colleague could inspect, which is, after all, exactly what Sandoval wanted from a conjecture map in the first place.

So the framework that gave the mastery picture its most rigorous form turns out to hold, in its easy-to-neglect second column, the means of its own qualification. You do not throw the conjecture map away. You use its affordances. But you read it against its grain, shift the weight from the outcome to the mediating process, and accept the price in mathematical tidiness as the cost of telling the truth about what your students are doing.

The institution asks for a destination. You hand it two coordinates and the honest record of a journey between them, and let it call that a destination if it must.

The Grand Compromise

Every theme in this essay has circled one stubborn fact: learning in widely accessible universities is done in compromised conditions. The final exam week is as real as the student union. Until a learner finally is accepted into a degree program with a major and perhaps a minor, epistemology means reading comprehension and taking good notes. How much more valuable these activities would be if the student understood correspondences and enjoyed the agency to explore them!

Mastery and correspondence, destination and journey, product and process — take your pick. If you teach courses in the CSU, you don’t get to pick, though you can take alternate routes. The syllabus itself is a legal document. The administration together with faculty governance keep courses running semester after semester for upwards of a half million students each year.

The constraints of the system are not an enemy to be defeated; they are the conditions under which the teaching happens at all. The mistake is to imagine you can dissolve them or ignore them, and the deeper mistake is to let them dissolve what they were meant to serve.

Early tests of knowledge and assumptions and later tests of integration of new learnings stop being origin and destination on a line toward mastery and become two samples of a correspondence relationship that would keep going if the calendar let it.

A single letter still flattens all of this into one symbol, and that symbol is hylomorphic by construction. The most you can do is make the grade attach to gradeable surrogates — the artifact, the act, the self-read delta — and keep the thing you actually care about, the attunement, off the rubric entirely, visible to you and the student but never marred by a score.

The tumbler gets photographed and published; the conversation that raised it stays in the maker’s hands. The same arrangement, compromises intact, nothing wildly out of balance, can hold in the CSU and on other campuses. The institution gets its object; the relationship between learner and material gets to remain a relationship between learner and material, one that can endure for a lifetime.

References

Ingold, T. (2013). Making: Anthropology, archaeology, art and architecture. Routledge.

Sandoval, W. (2014). Conjecture mapping: An approach to systematic educational design research. Journal of the Learning Sciences, 23(1), 18–36. https://doi.org/10.1080/10508406.2013.778204

Straatman, C. (2026). Forging a dialogue with metal: Apprenticing as more-than-human collaboration. Cultural Geographies, 33(2), 307–313. https://doi.org/10.1177/14744740251381383

The Pool, the Painter, the President, and a Nation Sinking in a Swamp

Terry Underwood, PhD — Mon, 22 Jun 2026 17:14:53 GMT

Once again, I’m astonished at how stupidly trusting I am, how foolishly I attend to the illusion of America instead of valuing evermore its bedrock strength. In that sense, I’ve never outgrown my childhood.

The Lincoln Memorial Reflecting Pool was not built in 1922 from a visionary impulse. It was built from a need for drainage.

When planning began in 1915, the National Mall was a muddy landfill like the mud flats upon which Washington, the city, was constructed. Architectural historians note that the pool was a basin to capture excess water draining from infill.

Henry Bacon, the prominent American architect best known for designing the Lincoln Memorial, gave it long and shallow elegance, water rippling blended reflections of sky, cloud, and stone. He wanted it to be a grand, reflective centerpiece to complement the Lincoln Memorial, even if managing the realities of the swampy terrain was an overriding functional necessity.

Underneath the dignity was a drainage problem.

Underneath the drainage problem was a swamp, which the President promised to drain. Did he know what he was saying?

The Lincoln Memorial was dedicated on May 30, 1922. The pool itself was still under construction, officially opened in 1924. Almost immediately, it began to sink, a tile and asphalt floor resting on marshland, buckling under the weight of 6.5 million gallons of water.

It leaked.

It stagnated.

It turned green.

It did these things for decades until a $34 million structural rebuild between 2010 and 2012 finally drove timber piles down to bedrock and gave the whole thing something solid to stand on. According to Pool and Spa News, October, 22, 2012:

“Because Washington, D.C., is built on a landfill, the weight of the concrete structure finally gave way and the pool sank nearly a foot since originally being built, Johnson explained. As a result, it was losing 50,000 gallons of water to leaks and evaporation weekly.”

In the mythology, it is a vast and dignified mirror, built to connect the sky, the Washington Monument, and the Lincoln Monument, capturing the nation’s best moments in still water, moments like Dr. King’s I Have a Dream speech. It’s important to remember we had a dream.

The whole time, it was sitting on filled marshland, doing its best.

Then the President decided to paint it blue.

Corruption, you ask?

The National Park Service awarded Atlantic Industrial Coatings a no-bid contract on April 3, 2026, after President Trump publicly stated he had “a guy who’s unbelievable at doing swimming pools.” Atlantic Industrial Coatings had never previously held a federal contract.

A National Park Service internal analysis found the firm received a profit margin of 20 percent on the project, well above the typical rate, adding at least $850,000 to what a comparable contract would have cost. An Interior spokesperson said the price “reflects the effort necessary” to complete the work ahead of Independence Day.

Trump had originally promised the renovation would cost between $1.5 and $2 million. The contract was awarded at $6.9 million. By mid-May, the Interior Department added $6.2 million. By June 15, an additional $460,000 transaction brought the total to $14.7 million. Several sources report that is not clear what that final payment was for.

The pool turned green within days of the work being completed. Though the President says otherwise, this was not because of vandals, not because of those crazy Democrats. It is what this pool does when you don’t understand its history. It’s always done this. Silly me, I didn’t know either, but then I wasn’t calling the shots.

It was doing this before any of us were born.

Corruption, you ask?

The federal government awarded a separate no-bid contract — $1.7 million — to Green Water Solutions, an Ohio-based company, to install a “Nano Bubble” filtration system to address the algae crisis. The owner listed on federal contracting documents is the “JJ Cafaro Investment Trust,” whose president and CEO is identified on Federal Election Commission filings as John J. Cafaro, a Trump donor who gave $250,000 to the Trump Victory fundraising committee in 2020.

Green Water Solutions had posted online photos of water treatment work it performed on a pond at Trump’s golf club in Bedminster, New Jersey.

So a company connected to the President’s golf operation painted the pool. The paint failed. A company connected to the president’s other golf operation was hired to fix it. Both contracts bypassed competitive bidding.

Urgent. That was the watchword. Imminent threat. Both were justified as urgent, the pool had to be fixed by the 250th birthday, Iran had to be bombed before it had a nuclear bomb.

Tim Whitehouse, executive director of the watchdog group Public Employees for Environmental Responsibility, said the renovations in the capital have “become a secretive project where the friends and business associates of the president are being rewarded with no public scrutiny.”

The pool, to its credit, did not stay quiet. It turned green. I didn’t know, but it has always known how to turn green.

***

The president’s response to the pool’s candor was to blame the Democrats.

Trump posted on Truth Social that vandals had attacked the pool, that “very serious crimes having to do with the destruction of National Monuments” had been committed, and that perpetrators faced “years in jail.”

Per usual, he offered no evidence. He alleged, without proof, that vandals had “done everything possible to hurt the inside surface that was just installed,” drawing a parallel to numbers etched in discolored grass on the National Mall that authorities said may have constituted a threat to the 47th president.

An administration official confirmed that five people had been arrested for vandalism and another five cited, with 14 police reports filed.

Among those arrested: a 67-year-old former Olympic athlete on a bicycle.

David Hearn had just finished a loop around Hains Point as part of a 52-mile ride when he stopped at the Lincoln Memorial to see the newly renovated pool. Noticing a piece of the new liner partially detached from the bottom, he reached into the water to feel it.

Moments later, U.S. Park Police arrested him on a misdemeanor charge of destruction of government property. Hearn said he touched the liner but “didn’t destroy or break or peel anything.” He now has a court date and is looking for legal help.

Meanwhile, National Guard members were photographed positioned along the edge of the pool, monitoring visitors as Park Service workers cleaned algae from the water.

****

The pool is a small story with a massive backstory. Iran is a massive story about a small man.

Consider the Strait of Hormuz. Like the reflecting pool, I had very little conscious awareness the global economy depends on this Strait. Like the reflecting pool, it sits atop a foundation that was always more complicated than its surface suggested with decades of regional tension, proxy conflict, and interdependent economies compressed beneath a narrow channel of water that the world cannot afford to close.

Like the reflecting pool, the President treated Iran with a surface understanding and an aggressive intervention: a military strike executed with confidence in some ill-defined outcome, applied to a substrate nobody helped him map.

Like the reflecting pool, it turned green almost immediately. The consequences were not a surprise to anyone who had been looking at what was underneath the President. This is the administration’s signature move: Foolish certainty applied to structural complexity followed by the arrest of anyone who points at the color of the water.

The Voting Rights Act of 1965 is a reflecting pool of a different sort, a place the nation built to look at itself honestly, to see in still water what it actually was and what it aspired to become.

Like the Lincoln Memorial pool, it was never perfect. It was built on compromised ground, amended, litigated, partially rebuilt. But it had, after decades of effort, been driven down to bedrock.

In April, the Supreme Court painted it green.

The Court voided a key provision of the Act, ruling that Louisiana’s electoral map had been drawn unconstitutionally in creating two Black-majority congressional districts, reinterpreting Section 2, the provision designed to protect minority voters from having their political power diluted.

The ruling did not arrive smelling of algae. Instead, it arrived smelling of Alito’s jurisprudence, of Thomas’s original intent, of Gorsuch’s textualism carefully applied, of Roberts’ political pragmatism. But green is green, whatever you call it.

The color of money has a way of tinting everything downstream.

Trump called it “a BIG WIN for Equal Protection under the Law” and told reporters “I love it,” adding he believed Republican-led states would want to reconfigure their voting maps, and he was right.

The ruling immediately sparked redistricting calls across multiple states ahead of the November midterms with new maps expected to net additional safe Republican House districts. The algae spread fast, as algae does when the filtration stops working.

What the Court has lost, in the end, is not legal reasoning, though it has lost that. It is color vision. It can no longer distinguish the green of money from the clear water of equal protection.

It looks at a pool choked with the interests of those who fund campaigns, draw maps, and own the machinery of elections, and it sees balance. It sees neutrality. It sees original intent.

The rest of us can see what color it is.

Trump declared that Republicans should “nationalize” elections and posted that there would be voter ID “whether approved by Congress or not.” The SAVE America Act, requiring a passport or birth certificate to register to vote, passed the House, with the Brennan Center estimating 21 million Americans lack ready access to qualifying documents.

A March executive order further attempted to transform the postal service into an arbiter of who may vote by mail, Trump’s second such attempt after courts blocked the first.

The algae spreads. The Court, color blind, stares into the water and sees blue.

*****

The Lincoln Memorial Reflecting Pool has been green before. It has sunk, leaked, stagnated, and been rebuilt. It sat on a swamp for 90 years before anyone drove piles down to bedrock. It has never been the pristine mirror of national mythology.

It was built as a drainage basis for heaven’s sake, for helping keep the Lincoln Memorial from sinking into the Earth. It is a symbol of the negotiation between democratic aspiration and the mud underneath.

What it is reflecting now is something different: a government that awards contracts to its friends, blames its failures on its enemies, arrests the people who look too closely, and calls it patriotism.

A Supreme Court that has lost the ability to perceive what color the water is. A democracy whose own foundation — the franchise, the map, the counted vote — is being repainted without competitive bidding, justified as urgent, and handed to the people who made a lot of money on the last job.

The pool will probably be drained and refilled. The algae will clear. Atlantic Industrial Coatings will keep the $14.7 million. Green Water Solutions will keep the $1.7 million.

David Hearn will fight his misdemeanor charge and perhaps pay a fine.

The Voting Rights Act will not be restored by executive order.

The Strait of Hormuz will not be a party to freedom of the seas.

We built this drainage basin on a swamp and called it a reflecting pool. For a century, it did its best to be one. The question now is whether we still have the will to drive the piles down to bedrock again.

https://watershapes.com/a-century-of-reflections/

References

https://www.nps.gov/linc/learn/historyculture/lincoln-memorial-construction.htm

https://www.poolspanews.com/business/landmark-reopens-after-major-makeover_o

https://www.newsweek.com/lincoln-memorial-reflecting-pool-bluepaint-pyramid-11940190

https://www.inquirer.com/news/nation-world/lincoln-memorial-reflecting-pool-trump-no-bid-contract-paint-blue-leaks-repairs-swimming-pools-algae-obama-parks-20260509.html

https://www.theguardian.com/us-news/2026/may/09/trump-no-bid-contract-lincoln-memorial-reflecting-pool

https://www.citizen.org/news/lets-get-to-the-bottom-of-trumps-expensive-no-bid-contracts-for-the-lincoln-memorial-reflecting-pool/

https://thehill.com/homenews/administration/5932337-lincoln-memorial-reflecting-pool-algae/

https://www.cbsnews.com/news/company-trump-donor-cafaro-reflecting-pool-no-bid-contract/

https://www.cbsnews.com/news/trump-reflecting-pool-vandalism-arrest-washington-dc/

https://www.newsweek.com/cyclist-arrested-handcuffed-reflecting-pool-national-guard-12098916

https://www.washingtonpost.com/nation/2026/06/20/cyclist-arrested-reflecting-pool-denies-trump-vandalism-claims/

https://www.aljazeera.com/news/2026/4/30/has-the-us-supreme-court-weakened-the-voting-rights-act-and-how

https://www.aljazeera.com/news/2026/4/30/trump-says-tennessee-next-to-redistrict-after-us-voting-rights-act-ruling

https://www.aclum.org/press-releases/voting-rights-groups-challenge-executive-order-on-mail-in-ballots-as-illegal-interference-in-elections/

https://www.brennancenter.org/our-work/analysis-opinion/save-act-and-election-power-grab

No Longer What They Were, Not Yet What They Will Become

Terry Underwood, PhD — Fri, 19 Jun 2026 18:12:39 GMT

Following a recent $14.2 million renovation intended to paint the Lincoln Memorial Reflecting Pool “American-flag blue,” the 6.5 million-gallon basin is currently suffering its worst green algae bloom in five years.

National Park Service crews in hip-waders have been deployed to net and vacuum the organic material out of the water. To kill the remaining bloom, officials have activated high-tech nanobubble ozone technology to oxygenate the water and are dumping jugs of hydrogen peroxide concentrate directly into the pool.

The water has stubbornly remained a murky green despite these ongoing interventions.

*
Over the past year, rural healthcare in America has experienced an immediate emergency, mirroring the pervasive, stubborn decay currently stagnating the White House reflecting pool. In 2025 alone, at least 90 rural hospitals experienced credit downgrades, full closures, or severe service eliminations.

Within that single year, communities saw 29 credit rating downgrades alongside 24 total hospital closures and another 24 facilities shuttering maternal and obstetrics wards. At least 27 rural hospitals either completed or formalized plans to eliminate their labor and delivery units this past year,

Recent reports estimate nearly 760 rural hospitals are operating at a risk of closure, with over 320 at immediate risk of shutting down within two to three years as their financial reserves dry up.

Over the past year, public schools have been plunged into a chaotic downward spiral that mirrors the murky, stagnant waters currently choking the White House reflecting pool. Just as removing the continuous flow of fresh water allowed toxic algae to consume the landmark, pulling the federal guardrails off essential programs has created severe, immediate funding disruptions. Twelve states are most at risk; Alabama, Arkansas, Kentucky, Louisiana, Mississippi, New Mexico, North Carolina, Oklahoma, South Carolina, Tennessee, Texas and West Virginia.

High-poverty districts rely on what was previously $18.4 billion in annual Title I funding. Without the federal apparatus to mandate its equitable distribution, states now face the possibility of treating these previously guaranteed resources as block grants, allowing some to divert tax dollars from public classrooms toward private school initiatives.

Just as the pool’s ecosystem broke down once its foundational filtration systems stopped functioning correctly, the foundational protections for millions of vulnerable students have stalled. Around 7.5 million students with disabilities are now caught in the undertow, as oversight of the Individuals with Disabilities Education Act (IDEA) and its $14.2 billion budget gets reappropriated perhaps to the military to cover a day and a half of war with no dedicated federal enforcement agency.

Massive proposed budget cuts threaten nearly 420,000 educator jobs nationwide, and schools are losing critical support systems to track racial disparities and civil rights violations, leaving the American public education system as starved and unbalanced as the algae-choked waters on the National Mall.

***
Over the past year, the American aviation system has suffered a breakdown in the sky mirroring the murky green scum currently attacking the White House reflecting pool. The nation’s airports are buckling under severe structural and capacity constraints.

In its 2025 report, the American Society of Civil Engineers downgraded U.S. aviation infrastructure to a “D+”, citing a massive $114 billion funding gap over the next decade as passenger volumes overwhelm outdated terminals and strained air traffic control systems.

Simultaneously, global airlines are facing $11 billion in extra costs from a relentless supply chain crisis that has grounded planes for software updates and delayed new aircraft deliveries, artificially aging fleets.

Like the reflecting pool, where obviously poor planning and superficial chemical dumps cannot overcome a fundamentally microbe-infested system which just cost $14 million in expected upgrades, the industry is struggling as staffing shortages, limited capacity, and failing infrastructure combine to trap travelers in an increasingly fragile, unpredictable network.[

****

Over the past year, childhood vaccination rates in the United States have continued a dangerous downward spiral, fracturing a foundational pillar of public health much like the spiraling green microbes currently choking the White House reflecting pool.

Just as the removal of a continuous, healthy flow of water for beautification purposes allowed toxic decay to quickly overtake the national landmark, the steady erosion of immunization coverage has allowed previously eliminated threats to flood back into communities.

During the 2024-2025 school year, non-medical exemptions for kindergartners hit an all-time high of 3.4%. Consequently, national coverage for the measles, mumps, and rubella (MMR) vaccine dropped to 92.5%, remaining below the critical 95% threshold required to prevent community spread.

In 2025 alone, the U.S. documented over 2,200 confirmed cases of measles across 45 jurisdictions, the highest total seen in decades. Simultaneously, the nation reported over 28,000 cases of pertussis (whooping cough).

Like the algae stubbornly clinging to the surface of the reflecting pool, these preventable outbreaks are a visible symptom of a system where essential, invisible protections have been allowed to decay.

*****

Much like the abrupt failure of the filtration systems currently choking the White House reflecting pool, sudden and unprecedented cuts to federal food assistance over the past year have abruptly choked off essential nutrition resources for millions of public school students.

In 2025, a staggering $186 billion reduction in Supplemental Nutrition Assistance Program (SNAP) benefits has shifted massive cost burdens onto states. Families lost SNAP eligibility under strict new requirements, their children automatically lost eligibility for free school meals.

Just as the loss of circulating water allowed the reflecting pool to stagnate, the USDA’s cancellation of the $660 million Local Food for Schools program immediately cut off the flow of fresh, locally sourced fruits and vegetables to cafeterias nationwide, forcing districts to fall back on heavily processed alternatives.

The $536 million SNAP-Ed program was entirely terminated, eliminating critical nutrition and cooking education that served over a million students, leaving the nation’s vulnerable children as starved for essential resources as the murky waters on the National Mall.

******

Nobody knows exactly why the filtration pumps of the Lincoln Memorial Reflecting Pool suddenly failed. There must be some connection between the $14 million upgrade project, its planning, and its execution. In some small way, it’s akin to what has happened in Iran.

Instead of green slime invading a national monument following an aggressive move to beautify the space we think of almost as a sanctuary, small potatoes compared to the Strait of Hormuz, we have the risk of a global economic meltdown following an arguably reckless move to save the world for democracy by destroying the progress democracy has made in terms of international norms.

No one can fully explain the upsides of the abrupt, synchronized near collapse of our rural hospitals, our public schools, our aviation networks, our public health, and our food assistance programs. At this point, my vision is a bit clouded by the metaphorical algae floating in DC.

All we know is that the life we were living just a short time ago, imperfect as it was, has changed forever, replaced by a murky, unfamiliar reality. We stand on the edge of a perilous future, watching the ecosystems of our nation suffocate under a creeping decay of Congressional, judicial, and executive leadership, a perfect storm, the Trifecta James Madison tried to prevent..

But the mechanisms of our democracy are not completely broken, and neither is our agency. We still have our voices, we can speak and write, and we still possess our vote at least for now. If we speak out loudly and clearly with a firm belief that each of us can be the change we seek, refusing to accept this toxic stagnation as our new normal, we can force the system to flush out the rot, one person at a time.

Reading, Lifelong Learning, and Ethical Reasoning in the Age of AI

Terry Underwood, PhD — Wed, 17 Jun 2026 18:22:41 GMT

What does it mean to be truly well educated?

The Association of American Colleges and Universities (AACU) spent years answering that question through its VALUE Project: Valid Assessment of Learning in Undergraduate Education. Sixteen rubrics define, calibrate, and offer the affordance of measurement for the learning outcomes that matter most to the professors who teach them in undergraduate education. Each rubric provides a common, if somewhat fragmented, national framework for evaluating student achievement across campuses and disciplines.

I want to look at three of those rubrics — Reading, Lifelong Learning, and Ethical Reasoning — in this essay. Recently, I published an essay exploring the disjuncture between an amorphous notion of critical thinking we reveal in the discourse of higher education (egregiously so in the research on AI in education where it’s as if the only voice on CT is Bloom) and written communication as these constructs are defined and elaborated in the rubrics. If you’ve read that piece, this one should flow fairly smoothly.

These three rubrics occupy genuinely distinct intellectual territory. Strong readers are not automatically curious learners. Skilled ethical reasoners need not be avid readers. Curious learners, avid readers, and ethical reasoners alike can, as the Bard taught us, “smile and smile and smile and be a villain.”

Each warrants its own rubric with due humility. Yet examined together, they reveal a coherent vision of a significant part of what undergraduate education is ultimately for: the gradual development of a mind capable of navigating complexity with comprehension, curiosity, and judgment.

Shared Architecture

Beneath their different substance, these three rubrics share a discernible structure. Each measures the movement over a four-year clock from externally driven, surface-level performance toward internally motivated, reflective judgment. Each treats self-awareness about one’s reading knowledge and skills, learning dispositions, and ethical beliefs as foundational rather than incidental.

Each reserves its highest marks for students who can operate across unfamiliar contexts without formal scaffolding. Implicitly in each, earning an undergraduate degree marks the moment a lifelong learner becomes aware of different ways to self-scaffold, to take up serendipitous opportunities that deepen experience and knowledge. Together, they describe not three separate skills but three expressions of the same underlying intellectual maturity.

The rubrics organize student performance across four levels: Benchmark (the floor of meeting expectations), two Milestones, and Capstone (the undergraduate ideal). Understanding what each level actually describes is essential to understanding what these rubrics are really asking — and what is now at stake in the age of AI.

The Benchmark Student

The benchmark reader approaches texts expecting to find right answers and facts to display for credit. She can identify aspects of a text when a question directs her to, but she is not yet reading so much as responding to prompts. The text is an obstacle between her and a grade. Interestingly enough, AI flips this script. She is now prompting and then reading.

The benchmark lifelong learner completes required work. That is the entirety of the description for Initiative: words that function almost as a rebuke. He explores topics only at a surface level, makes vague references to prior learning without applying it, and reviews past experiences without arriving at any clarified meaning or broader perspective. The benchmark here is a description of compliance. Translated into an AI context, this learner is likely vulnerable to distortions.

The benchmark ethical reasoner can recognize basic and obvious ethical issues but fails to grasp their complexity or interrelationships. She can name an ethical theory without doing much with it, states her core beliefs without examining their origins, and can only apply ethical frameworks with external support — in a class, in a group, or in a fixed-choice setting. Left on her own with a genuinely new ethical problem, she is not yet equipped. As near as I can tell, the jury is out on the question of AI as a hero or villain artificial ethical reasoner.

What unites these three portraits is dependence: on assignments for direction, on instructors for clarification, on familiar contexts for application. The benchmark student in each rubric is waiting to be told what to do and given a hand to hold. The rest of the rubric describes the long process of learning not to wait. Tragically, dependence as a learner might very well morph into dependence on an AI interlocutor.

Before accepting that portrait as a description of what students are, however, we should ask a harder question: whether it is a description of what they would become under any circumstances or whether we have made them.

American schooling initiates children into literacy through daily regimens of behaviorist phonics instruction: decoding as discipline, correctness as the goal. It teaches writing in prescribed structures: the paragraph, then the five-paragraph essay, then the argument scaffolded by a tidy acronym, something about claims and warrants.

By eighth grade in Texas in 2025, students were being asked to produce extended written responses to prompts like “Should homework be mandated?” — questions chosen not because they matter to anyone but because they are gradable.

The architecture of schooling, from its earliest moves through high school, is designed to produce students who wait for the prompt, answer the prompt, stop, and wait to be told how they did. We have built the benchmark student. We should not be surprised that she shows up.

The Capstone Student

The capstone reader has moved far beyond decoding. She recognizes implications of texts for contexts that extend beyond the assigned task and beyond what the author explicitly intended. She participates in disciplinary conversation, not just extracting and constructing meaning from texts but responding to texts as a member of a community of readers. Reading has become, for her, a form of ongoing intellectual engagement with the world.

The capstone lifelong learner pursues knowledge and experience that flourish entirely outside classroom requirements. He generates opportunities to expand his abilities rather than merely completing them. His reflection reveals significantly changed perspectives, evidence of genuine growth rather than performed introspection. His curiosity has become self-sustaining, no longer dependent on academic structures to keep it alive.

The capstone ethical reasoner can do something particularly demanding. She states a position, articulates the strongest objections to it, and defends against those objections adequately and effectively. She is not just ethically aware; she is ethically courageous, willing, nay, demanding that her position be genuinely tested.

What unites these three portraits is autonomy and agency of judgment, of curiosity, of conscience. The capstone student in each rubric has internalized the criteria so thoroughly that she no longer needs the rubric at all. That disappearing act is the miracle: responsibility for learning has been completely released by the teacher and grasped firmly by the learner.

AI and the Architecture It Inherited

The arrival of capable AI language models has triggered a predictable institutional panic, and the panic has generated a predictable misdiagnosis. The problem, in this telling, is that AI allows students to fake sophisticated thinking or, said differently, to produce capstone-level outputs without capstone-level development. The solution at first was detection, prohibition, and surveillance.

This solution got the situation almost exactly backwards.

AI did not create the gap between the appearance of educated thinking and its reality. That gap was already the central achievement of the system we built. Automated text classifiers were already scoring student writing at ETS long before ChatGPT. Algorithms were already selecting prompts for their gradability rather than their intellectual substance.

Students were already learning to produce responses that satisfy assignments rather than build funds of knowledge. AI did not conquer the architecture of schooling. It simply revealed how thoroughly that architecture had already been weakened by compliance, by measurement, by the reduction of learning to legible output.

Consider what AI actually demonstrates when it performs at academic levels. It has passed the USMLE — the United States Medical Licensing Examination — not at the top of the distribution, but at the standard required for licensure. It has passed the bar exam. It has passed other licensing examinations.

These are not trivial instruments. They represent the accumulated judgment of professional communities about what minimum competence requires. When AI meets that standard, the appropriate response is not to conclude that AI is faking medicine or law. It is to ask what those examinations are actually measuring and whether the answer is the same thing we thought it was.

Would a medical school admissions committee admit a student on the basis of a test score alone?

The VALUE rubrics were designed to measure orientation toward learning, particular habits of mind, deliberate transference of knowledge from participatory contexts inside and outside the curriculum. One’s orientation toward learning is part of a lived identity as evidenced through actual human cultural participation. That distinction matters more now than it ever has.

Because AI cuts both ways. The same capacity that allows a disengaged student to produce fluent text without thinking also allows an engaged student to think at levels previously inaccessible without years of preparation.

A first-year student who wants to stress-test an ethical position now has access to the most rigorous, tenacious interlocutor imaginable, one that will articulate the strongest counterarguments without social judgment and without fatigue. A curious learner whose questions have always outrun the available runway can now pursue those questions indefinitely. I can testify to that.

A reader who has always been too uncertain to argue with a difficult text now has a machine that can model what it looks like to read vigorously just to see. For students already oriented toward learning, already somewhere on the developmental arc the rubrics describe, AI is not a threat to that development. It is an accelerant.

The students who use AI to avoid thinking and the students who use AI to think harder are not distinguished by AI. They are distinguished by everything that happens to them as human beings going about their daily lives, especially in school.

Whether they were taught that the purpose of a text is to yield a correct answer or to reward deep engagement; whether the prompts they were given were worth thinking about; whether they were ever asked to write their own prompts; whether anyone in their education modeled what it looks like to be genuinely curious. The benchmark student who waits for direction was trained to wait. AI waits with them and makes polished output on demand.

This is the real 2026 benchmark problem. Not that AI has changed what educated thinking looks like — the rubrics still describe it appropriately and somewhat clearly at least for our time — but that the system designed to produce educated thinkers was already, at its foundations, designed to produce ever more sophisticated compliance.

AI has made that contradiction impossible to ignore.

The Political Conclusion

This is a policy problem, and policy problems require political will. But the policy has to address the right problem, and our politicians have consistently misdiagnosed the case literally for centuries. Accountability generates compliance, and accountability is core to law and order. But classrooms are not courtrooms.

The right problem is not AI.

The right problem is that American schooling, from its earliest literacy instruction through its standardized exit examinations, is organized around the production of legible compliance rather than the development of genuine intellectual capacity.

We built the benchmark student intentionally by design through behaviorist reading instruction, formulaic writing curricula, and assessment instruments that reward students for producing what classifiers can score. We then sent that student to college and called her underprepared.

We proposed surveilling her AI use while leaving every other element of the architecture intact.

The fix requires something more fundamental than policy tweaks administered through the existing machinery. What it requires is a dedicated national center, represented at the White House level, organized entirely around learning science — not accountability metrics, not workforce pipelines, not the next cycle’s testing contract.

This center’s mandate is to translate what we actually know about how human beings develop as readers, learners, ethical reasoners, and the other outcomes we value into the daily instructional life of American schools. The research base for that center already exists.

The learning science is not in dispute among the people who do it. What is in dispute is whether anyone in a position of political authority will take it seriously enough to fund it, staff it, and protect it from the interests that have always preferred compliance to capacity. My nominee to lead it is Linda Darling-Hammond.

Darling-Hammond has spent a career building precisely the case for action — on teacher preparation, on assessment reform, on the relationship between equitable resource distribution and genuine learning outcomes. She was on President Obama’s transition team and was available then for Secretary of Education.

She has the intellectual credibility the work requires and the political experience to navigate the environment in which it would have to operate. The knowledge exists. The person exists. What does not yet exist is the will to act on what we know rather than on what is easy to measure. The opportunity to rebuild may not show up for another two years, but the groundwork can be laid.

The VALUE rubrics describe a student who no longer needs to be given a rubric. That disappearing act — so complete that external assessment and scaffolding become unnecessary — is, in miniature, the promise of liberal education and the promise of democratic citizenship. It is the person who can read a text against the grain, pursue knowledge independently, and hold an ethical position under pressure.

We know how to build that person and, despite the adversity amidst the diversity, despite the contradictions inherent in our current model of schooling, we build them in every state year after year. Somehow we manage to produce really smart people. We have simply, for a very long time, for some unholy reason, chosen not to extend high-quality educational resources to every child.

AI has made that choice visible. What we do with the visibility is, still, up to us.

References

American Association of Colleges and Universities. (2009). VALUE rubrics. https://www.aacu.org/value/rubrics

American Association of Colleges and Universities. (2009). Reading VALUE rubric. https://www.aacu.org/value/rubrics/reading

American Association of Colleges and Universities. (2009). Foundations and skills for lifelong learning VALUE rubric. https://www.aacu.org/value/rubrics/lifelong-learning

American Association of Colleges and Universities. (2009). Ethical reasoning VALUE rubric. https://www.aacu.org/value/rubrics/value-rubrics-ethical-reasoning

Darling-Hammond, L. (2010). The flat world and education: How America’s commitment to equity will determine our future. Teachers College Press.

Darling-Hammond, L., & Bransford, J. (Eds.). (2005). Preparing teachers for a changing world: What teachers should learn and be able to do. Jossey-Bass.

Learning Policy Institute. (2026). Linda Darling-Hammond. https://learningpolicyinstitute.org/person/linda-darling-hammond

Kung, P. C., Cheatham, M., Medenilla, A., Sillos, C., De Leon, L., Elepano, C., Madriaga, A., Aggabao, R., Diaz-Candido, G., Maningo, J., & Tseng, V. (2023). Performance of ChatGPT on USMLE: Potential for AI-assisted medical education using large language models. PLOS Digital Health, 2(2), e0000198. https://doi.org/10.1371/journal.pdig.0000198

Katz, D. M., Bommarito, M. J., Gould, J., & Arredondo, P. (2024). GPT-4 passes the bar exam. Philosophical Transactions of the Royal Society A, 382, 20230254. https://doi.org/10.1098/rsta.2023.0254

Snow, C. E., Burns, M. S., & Griffin, P. (Eds.). (2002). Reading for understanding: Toward an R&D program in reading comprehension. RAND Corporation. https://www.rand.org/pubs/monograph_reports/MR1465.html

Writing Instruction: A Cat With Nine Lives Hanging On By Its Tail in Texas

Terry Underwood, PhD — Tue, 16 Jun 2026 20:47:56 GMT

"What I saw for years was that because it was only tested in fourth and seventh grade, students would only get explicit writing instruction in fourth and seventh grade," she said. "And so as a seventh-grade teacher, I would be scrambling to try to figure out how can I cover all of these grammar mechanics? How do I cover all of these writing techniques? What can I drop? What do we need to focus on?" (Erika De La Rosa, Texas teacher, February 29, 2024)

When the Texas Education Agency (TEA) first announced its partial transition from multiple-choice to constructed-response items on its statewide tests before it began considering machine scoring, English teachers like Erika De La Rosa saw a silver lining.

Legislation early in the 2020s required TEA to reduce its reliance on multiple-choice items in favor of constructed-response items. Students would have to write in every grade if only in response to their reading. Researchers in writing assessment have long documented the unfortunate consequences of grade-level targeting of writing. Those are the grades where writing is taught.

Perhaps requiring students to write constructed responses to reading passages at each grade level would stimulate writing instruction more broadly. But those early glimpses of hope were dashed when teachers learned that assessment would be accomplished by an automated scoring machine.

Writing as authoring original texts apart from reading a passage, however, would not be assessed. TEA would instead pair all writing with passages to be read, another move that traditionally elevates reading above writing. The Common Core State Standards were built from the assumption that students learn to write through analyzing texts written by someone else and synthesizing, interpreting, or viewing the information as evidence in an argument.

Looking under the hood of precisely how these extended constructed responses (ECR) are scored under TEA mechanisms provides good reason to suspect that robust writing instruction in Texas is being distorted in the curriculum in two ways. First, no serious educator defines writing as what students do to provide evidence of comprehension of a text. Second, when even that performative use of writing is scored by an automated scoring machine, what must count as effective writing becomes a thesis-driven five paragraph essay, a cookie cutter approach that makes automation less prone to wild errors.

The following discussion focuses on the TEA-released sample extended construction response item from Spring 2025. An analysis of the ECR prompt itself as well as the scoring guide suggests that the sort of writing Texas has in mind comes from a perspective on composition that predates process writing research begun in the 1980s.

Though this approach makes room for automated scoring, it fails miserably in doing what Erika De La Rosa hoped for, namely, a thorough redirection from teaching writing as a sometime thing in fourth and seventh grade to teaching writing as a fundamental focus of schooling every day across the curriculum.

What the TEA Prompt for the ECR Writing Task Tells Us

Grade 8 Reading Extended Constructed Response

Prompt

Read the article “Seeing in a Sea of Snow.” Based on the information in the article, write a

response to the following:

Explain what the invention of snow goggles reveals about ancient cultures.

Write a well-organized informational composition that uses specific evidence from the article

to support your answer.

Remember to —

• clearly state your thesis

• organize your writing

• develop your ideas in detail

• use evidence from the selection in your response

• use correct spelling, capitalization, punctuation, and grammar

Manage your time carefully so that you can —

• review the selection

• plan your response

• write your response

• revise and edit your response

Write your response in the box provided.

This is a writing prompt that is mostly about reading. Consider how.

The task itself

“Based on the information in the article” and “Explain what the invention of snow goggles reveals” anchor the intellectual work entirely in the source text. The student isn’t asked to bring anything to the task beyond the ability to read, comprehend, and draw inferences.

There’s no invitation to reason, imagine, extrapolate, or criticize beyond the article, to connect to prior knowledge, or to form an independent interpretive position. The thesis the student is supposed to “clearly state” is already implied by the prompt — snow goggles reveal something about ancient cultures — so the student is really being asked to retrieve and organize, not to argue, a directive that distorts any habitual or typical way of reading the student might deploy.

I’ve come to think of these kinds of tasks as the “David Coleman Special.” Asked about the role of the student in the Common Core State Standards, which he developed as the project chief, he responded that high school students simply need to accept the fact that “nobody gives a sh**t about what you think or feel.”

The checklist

Of the five “Remember to’s,” three are explicitly reading-dependent or text-dependent (”clearly state your thesis” means state the one the prompt set up, “develop your ideas in detail” means elaborate on the article’s content, “use evidence from the selection” is direct retrieval). Only spelling, capitalization, punctuation, and grammar are purely writing concerns and those are conventions, not composition.

The time management bullets

The first four steps are highly misleading. “Review the selection” means “reread and reorganize the content,” which is not the first step in any robust academic writing project. “Plan” is also ambiguous. Indeed, a savvy test-taker would have begun planning the essay during the first cold reading, an external injection into the reading process that influences the student in unpredictable ways.

“Write, revise.” Writing in this sense means “generate language to fill a predetermined structure for a predetermined purpose to earn a grade from a machine.”

In writing pedagogy, "write" typically means discovery. Elbow's (1973) notion that writing precedes knowing, that composing is how thought gets made, not how finished thought gets recorded, is ignored. Murray (1978) called this "writing to learn."

Revision, for Sommers (1980), means genuinely re-seeing, restructuring an argument, rethinking claims, sometimes abandoning drafts entirely. Flower and Hayes (1981) showed that skilled writers recursively plan, draft, and revise simultaneously, not sequentially.

None of that is available here. The prompt's "write" means fill the predetermined container; its "revise" means fix surface errors. Both words are borrowed from writing pedagogy while being depleted of their meaning.

What’s missing

Nothing in the prompt asks the student to do anything a writer actually does — no stance-taking beyond what the prompt dictates, no consideration of audience, no decisions about what to include or exclude from their own thinking, no writerly judgment of any kind. The “well-organized informational composition” framing sounds like a writing task, but the organizational work is constrained entirely by what the article contains.

The core problem

This is essentially a reading comprehension task with a writing delivery mechanism. The rubric rewards thesis clarity, organization, evidence explanation, and word choice, but the content of all four is determined by the source text, not by the writer. A student who reads carefully and summarizes coherently will score well. A student who writes beautifully but wanders from the article will not. That’s a reading assessment dressed in writing clothes.

What the Rubric for the TEA Extended Construction Response Tells Us

Grade 8 Reading Passage with Extended Constructed Response Informational Writing Rubric

Organization and Development of Ideas

3 • Controlling idea/Thesis is clear and fully developed

The controlling idea/thesis is clearly identifiable. The focus is consistent

throughout, creating a response that is unified and easy to follow.

• Organization is effective

A purposeful structure that includes an effective introduction and

conclusion is evident. The organizational structure is appropriate and

effectively supports the development of the controlling idea/thesis. The

sentences, paragraphs, or ideas are logically connected in purposeful

and highly effective ways.

• Evidence is specific, well chosen, and relevant

The response includes relevant text-based evidence that is clearly

explained and consistently supports and develops the controlling

idea/thesis. For pairs in grades 6 through EII, evidence is drawn from

both texts. The response reflects a thorough understanding of the

writing purpose.

• Expression of ideas is clear and effective

The writer’s word choice is specific, purposeful, and enhances the

response. Almost all sentences and phrases are effectively crafted to

convey the writer’s ideas and contribute to the overall quality of the

response and the clarity of the message.

2 • Controlling idea/Thesis is present and partially developed

A controlling idea/thesis is presented, but it may not be clearly

identifiable because it is not fully developed. The focus may not always

be consistent and may not always be easy to follow.

• Organization is limited

A purposeful structure that includes an introduction and conclusion is

present. An organizational structure may not be consistent and may not

always support the logical development of the controlling idea/thesis.

Sentence-to-sentence connections and clarity may be lacking.

• Evidence is limited and may include some irrelevant information

The response may include text-based evidence to support the controlling

idea/thesis, but it may be insufficiently explained, and/or some evidence

may be irrelevant to the controlling idea/thesis. For pairs, evidence is

drawn from at least one of the texts. The response reflects partial

understanding of the writing purpose.

• Expression of ideas is basic

The writer’s word choice may be general and imprecise and at times

may not convey the writer’s ideas clearly. Sentences and phrases are at

times ineffective and may interfere with the writer’s intended meaning

and weaken the message.

Texas Education Agency

Student Assessment Division

2025 7Grade 8 Reading Language Arts

Constructed-Response Scoring Guide

1 • Controlling idea/Thesis is evident but not developed

A controlling idea/thesis is present but not developed appropriately in

response to the writing task.

• Organization is minimal and/or weak

An introduction or conclusion may be present. An organizational

structure that supports logical development is not always evident or is

not appropriate to the task.

• Evidence is insufficient and/or mostly irrelevant

Little text-based evidence is presented to support the controlling

idea/thesis, or the evidence presented is mostly extraneous and/or

repetitious. Explanation of any evidence presented is insufficient and

may be only vaguely related to the writing task. For pairs in grades 6

through EII, evidence is drawn from only one text. The response reflects

a limited understanding of the writing purpose.

• Expression of ideas is ineffective

The writer’s word choice is vague or limited and may impede the quality

and clarity of the essay. Sentences and phrases are often ineffective,

interfere with the writer’s intended meaning, and impact the strength

and clarity of the message.

0 • A controlling idea/thesis may be evident.

• The response lacks an introduction and conclusion. An organizational

structure is not evident.

• Evidence is not provided or is irrelevant.

The response reflects a lack of understanding of the writing purpose.

• The expression of ideas is unclear and/or incoherent.

Please note that if a response receives a score point 0 in the

Development and Organization of Ideas trait, the response will also

earn 0 points in the Conventions trait.

The rubric systematically rewards a particular cognitive style, one that keeps the source text at the center of the composing process at every stage. Here’s how that plays out across each dimension:

Controlling Idea/Thesis

The rubric calls for a thesis that is “clearly identifiable” and “fully developed.” But developed from what? The only available material is the article. A student who generates a thesis from their own prior knowledge about indigenous ingenuity, or who reasons independently about what material culture reveals about ancient societies, has no way to anchor that thesis in the rubric’s terms. “Thorough understanding of the writing purpose” means thorough understanding of the article’s purpose — the two are collapsed into one. The student who stays closest to the text has the clearest thesis almost by definition.

Organization

“Logically connected in purposeful and highly effective ways” sounds like a writing criterion, but in this context logical connection means following the article’s own organizational logic. The article moves from problem to invention to design to effectiveness — a student who mirrors that structure will naturally produce something that reads as purposeful. The rubric cannot distinguish between organization that emerges from the writer’s own rhetorical decisions and organization that is borrowed from the source text’s structure. The text-dependent writer gets credit for the article’s organization.

Evidence

This is where the privilege is most explicit and most consequential. “Specific, well chosen, and relevant” evidence at score point 3 requires the student to be continuously referencing the text — selecting, quoting or paraphrasing, then explaining the connection to the thesis. This is a reading behavior masquerading as a writing behavior. The cognitive demand is primarily retrieval and selection from a fixed source, not the generative act of finding and marshaling evidence from one’s own knowledge or experience. “Insufficiently explained” at score point 2 penalizes students who cite evidence competently but move away from the text too quickly to develop their own thinking. The rubric essentially taxes independent thought by treating departure from the text as incompleteness.

Expression of Ideas

“Specific, purposeful” word choice at score point 3 will almost inevitably be rewarded in students who are borrowing the article’s vocabulary. The article about snow goggles could have used precise technical and domain language — photokeratitis, ultraviolet, corrective lenses, Arctic, millennia. A student who absorbs and redeployes that language will appear to have specific and purposeful word choice.

A student who writes fluently in their own register, using general but accurate language, risks the “general and imprecise” characterization of score point 2. The rubric cannot see the difference between a student who owns precise language and one who is reflecting the text back.

The automated scoring problem

The rubric’s text-dependence is compounded by the fact that TEA uses a hybrid automated scoring engine. Automated scoring systems are trained on features like lexical overlap with the source text, keyword density, and structural markers. A student whose essay closely mirrors the article’s language and structure will score well not because they are a better writer but because they are a more compliant text-processor. The machine rewards proximity to the source, which means the rubric’s implicit privileging of text-dependent behavior is amplified by the scoring mechanism.

What the rubric cannot see

It cannot reward a student who synthesizes independently, who brings genuine prior knowledge to bear, who takes an unexpected but defensible interpretive position, or who writes with distinctive voice. All of those are writing virtues in any other context. Here they are liabilities unless they are tethered to the article at every turn. The student who reads the article once, forms a genuine intellectual response, and writes from that response will almost certainly underperform the student who reads the article twice, maps its structure, and builds a response by shuttling back and forth between text and draft.

The deeper issue

The rubric describes a transaction, not a composition. The student’s job is to receive information from the article and return it in organized form with connective tissue added. That is a legitimate academic skill — it is roughly what a researcher does when summarizing sources — but it is not what writing pedagogy means by composing, and it is not what the rubric’s own language (”writer’s word choice,” “writer’s ideas,” “the writer”) implies about the nature of the task. The language of authorship is applied to what is fundamentally an act of managed retrieval.

The article "Seeing in a Sea of Snow" as it appears on the STAAR test is likely a TEA-commissioned or licensed piece based on my experiences working with large-scale assessment efforts, which is standard practice to avoid students having prior exposure to the text. I can’t find it as a standalone article.

This image was found on Facebook, which hosts a group of interested educators titled “The ELAR/SLAR & WRITE Prescription Club 2025-2026” with 13,000 members. It is dedicated to decoding the subtle cues in the TEA prompts that reveal what the prompt wants and then translating those cues into kid-friendly tips. This particular graphic was created by Bill MacDonald, I believe.

Writing: A Cat with Nine Lives in Danger in Texas

Life 1 Gone: Writing disappears as a subject.

The most immediate threat is curricular. When writing is assessed only as a delivery mechanism for reading comprehension, it stops being taught as a discipline with its own logic, history, craft traditions, and intellectual demands. Erika De La Rosa’s observation — that writing instruction was concentrated in tested grades — describes what happens when assessment drives curriculum.

If the new system assesses writing at every grade but only as a reading response, the same contraction happens again, just more evenly distributed. Teachers will teach students to produce reading responses, not to write. The subject called writing will exist in name only.

Life 2 Gone: Process writing research becomes irrelevant.

Forty years of research established that writing is a cognitive process embedded in a social setting, not a product. That research changed classroom practice. This assessment framework reverses the gains by treating writing as product delivery.

When the test defines writing as filling a predetermined container with retrieved content, teachers who teach process — drafting, discovery, revision as re-seeing — are teaching toward a different test that will not reward and may actually penalize. The rational teacher, under accountability pressure, abandons process. The research becomes professionally irrelevant.

Life 3 Gone: The five-paragraph essay is permanently institutionalized.

The rubric’s architecture — thesis, organized body, conclusion, text-based evidence, conventional correctness — describes and rewards exactly one form: the thesis-driven five-paragraph essay. That form has been criticized for decades not because structure is bad but because the five-paragraph essay teaches students that structure precedes thought, that the container is fixed before the content is known.

Applebee (1984) documented its dominance. Hillocks (2002) argued it actively impedes genuine argumentation. This rubric makes any other form risky. A student who writes a genuine argument with a complex structure that doesn’t front-load a thesis faces a scoring penalty. The form is now backed by the authority of the State of Texas.

Life 4 Gone: Automated scoring makes the threat permanent and self-reinforcing.

Machine scoring rewards the five-paragraph essay and it trains future scoring on it. The hybrid model TEA uses builds its automated engine on human-scored anchor papers. If those anchor papers are themselves the product of a rubric that rewards text-dependent thesis-driven prose, the machine learns that writing is text-dependent thesis-driven prose.

Each generation of test-takers produces training data that further calcifies the definition. There is no corrective mechanism. Unlike a human reader who might notice and reward unexpected excellence, the machine cannot be surprised. Voice, originality, intellectual risk, genuine argument — none of these generate signals the machine is trained to see.

Over time the population of Texas writers is shaped toward what the machine rewards, and the machine is retrained on that shaped population. It is a closed loop that tightens with each iteration.

Life 5 Gone: Audience disappears

Every serious writing researcher and theorist — from Britton (1975) to Nystrand (1997) — has argued that writing is fundamentally a social act, that writers write for readers, and that the awareness of audience is central to rhetorical development. This assessment has no audience.

The student writes for a machine that has no comprehension, no stake in the content, no capacity to be persuaded or informed or moved. The prompt does not name an audience. The rubric does not mention audience. A student who writes with genuine communicative intent, who imagines a reader and makes choices accordingly, receives no credit for that intent.

Writing as communication is replaced by writing as compliance. That replacement, absorbed by students across thirteen years of schooling, produces adults who understand writing as performance for institutional authority rather than as a means of acting in the world.

Life 6 Gone: Indigenous and non-dominant voices are structurally disadvantaged.

The article about snow goggles is, ironically, about indigenous ingenuity, but the assessment framework surrounding it reproduces the very epistemic hierarchy that historically silenced indigenous knowledge.

A student from an Alaska Native community who knows this history deeply, whose family carries this knowledge, cannot use that knowledge. It is not in the article. The rubric will penalize its appearance as evidence because it is not text-based.

The student who knows the most about the subject is required to set that knowledge aside and defer to a TEA-commissioned article. This is a structural reproduction of the conditions under which indigenous knowledge has always been devalued in institutional settings. The institution’s text supersedes the community’s knowledge.

Life 7 Gone: Writing as thinking is severed.

Vygotsky’s foundational insight, that language and thought are not separate, that writing is a tool for developing cognition, is operationally negated by this framework. Writing-to-learn, which has extensive research support from Emig (1977) through Klein (1999) and beyond, requires that the writer discover something through the act of writing that they did not know before they began.

This assessment forecloses that possibility by design. The content is determined before writing begins. The structure is determined before writing begins. The thesis is implied by the prompt before writing begins. There is nothing left for writing to do except transcribe what reading has already produced. Students trained exclusively in this mode learn that writing does not generate thought; it merely packages it. That is a catastrophically impoverished model of what writing is for.

Life 8 Gone: The teacher is deskilled.

When the rubric is this prescriptive and the scoring is automated, the teacher’s professional judgment becomes irrelevant to the outcome. The Facebook page referenced earlier is heart-breaking in its depiction of teachers working desperately to defend their students against scores of zero while salvaging what they can in the name of quality instruction.

A teacher who reads student writing with attention to voice, development of ideas over time, rhetorical sophistication, genre awareness, or genuine argumentation is developing capacities the test cannot see. The test rewards a narrow, specifiable, machine-detectable set of features.

The tragedy is that teachers who align their instruction to those features are not teaching writing. They are teaching test compliance. Teachers who resist and teach writing as a full discipline are working against the incentive structure. Over time the profession loses its capacity to recognize and reward genuine writing development because the institutional definition of writing has narrowed to what the machine can score.

Life 9 Gone: The cumulative effect

Taken together, these threats describe not just a bad test but a redefinition of literacy itself. If this framework operates at every grade level across the state, an entire generation of Texas students will have been taught that writing means reading carefully and reporting back in organized prose with a thesis stated upfront and evidence cited from the approved source.

That is a specific and limiting conception of what it means to be literate. It excludes expressiveness, argumentation from personal knowledge and experience, genre diversity, audience awareness, writerly voice, and the fundamental experience of discovering what you think by writing. Those exclusions are the logical consequence of a system designed to make writing machine-scorable, and they represent a genuine narrowing of what it means to be an educated person.

The Stakes Couldn’t Be Higher

Texas public education is already staggering under a storm of litigation — over school finance, over book bans, over vouchers draining public school funding. Add to that storm this potentially more damaging assault: the systematic redefinition of literacy itself in the name of administrative efficiency.

While legislators and pundits warn endlessly about artificial intelligence threatening the integrity of student writing, the far greater threat is the one TEA has already institutionalized, a machine-scored, text-dependent, five-paragraph-essay factory that has preemptively answered the AI problem by eliminating genuine writing altogether.

The irony is devastating. We are told that AI threatens writing because machines can now produce human-sounding prose. TEA’s response is to assess writing in a way that rewards students for writing like machines, for retrieving, organizing, and returning information in a predetermined structure for a non-human reader. The cure is the disease.

Ours is not a technical problem with an engineering solution. It is a failure of will and a lack of leadership, a decision to sell children’s literacy futures for the convenience of scalable scoring. It ignores everything we know about how human beings develop as thinkers through language.

Erika De La Rosa was right to hope. She was right to be disappointed. The obligation now is to keep saying clearly what is being lost before the generation that never learned to write inherits the consequences.