Depth and Clarity: A Thought Exercise in Essay Writing

The Thought Exercise

Depth and Clarity was conceived around the idea of qualitative superintelligence. We use that phrase as our name for a thought experiment, not as an established technical category: what might an intelligence with a substantially greater capacity for understanding notice, connect, or question that we overlook? We brought that question to essay writing and developed seven stages through which to explore it. The entire framework remains a provisional concept. We are testing it through a demonstration essay, examining where it helps, where it breaks down, and what needs to change.

Its guiding principle is: Seek deeper understanding, then make that understanding clear enough for a reader to examine and challenge.

Why Begin With Superintelligence?

Used as a buzzword, superintelligence can suggest extraordinary capability without explaining what better thinking would involve. Our exercise asks us to make that promise concrete. For this purpose, the word qualitative directs attention to differences in understanding: noticing a hidden assumption, finding a distinction that changes the question, or connecting evidence in a way that makes an explanation more revealing. What would such a difference look like in an essay, and how could a reader examine it?

That question is the framework’s conceptual starting point. It gives us a reason to look beyond the first plausible answer and to ask what our present understanding leaves out. The seven stages translate that ambition into work we can attempt and check: define the problem, compare explanations, examine support, invite correction, and make the resulting reasoning clear. Deeper thinking, dependable evidence, clear explanation, and stronger independent writing ability are the qualities we want to develop and examine.

We cannot know what a hypothetical superintelligence would conclude by imagining one. Every proposed insight still needs evidence or reasoning. The thought experiment supplies questions, not authority, and makes no claim that current AI has achieved superintelligence. Nor are planning, criticism, or feedback unique to superintelligence; the writing research discussed below helps us examine these familiar practices on their own merits.

We can explore the question through explanatory, interpretive, evaluative, and argumentative essays. The stages below propose ways of directing attention, and their value remains open to examination. The exercise can be undertaken with or without AI.

What Supports the Concept

Research on writing and learning, together with established approaches to argument, gives us reasons to investigate several parts of this framework. The citations below identify those connections and their limits. They were added while reviewing the concept; they should not be mistaken for the sources from which the original seven stages were derived.

The seven-stage arrangement, its prompts, and its application to AI-assisted writing remain our proposal. The cited work does not test Depth and Clarity as a whole. It can also challenge our design: evidence favouring flexible writing processes, for example, would count against treating our stages as a fixed sequence.

Exploring the Seven Stages

Imagine taking an essay question through each stage. What would become clearer? What might challenge the initial answer? Brief notes can make those changes visible; the exercise does not require seven separate documents.

The stages describe a useful sequence of questions, with permission to return to earlier work. Read enough to orient yourself before proposing an answer. Use exploratory writing whenever it helps you discover what you think. If drafting exposes a missing inference, return to the evidence; if research changes the problem, revisit the question.

Flower and Hayes’s cognitive theory treats composing as interacting, goal-directed processes rather than discrete steps completed once. Their account gives us a theoretical reason to make movement between our stages explicit; it does not establish that our particular stages are the right ones. (Flower and Hayes 366–67)

Each stage suggests an output and a question with which to examine it. Writing the demonstration essay will test whether these proposed stages and checks are useful, clear, and sufficient. Problems encountered during that work will inform revisions to the framework itself.

Choosing how much work to do

For a first pass, keep one short note covering the question, a tentative answer, its strongest relevant alternative, the evidence needed, and the main uncertainty. Then draft enough to expose a problem and revisit the stage that can help resolve it. Give the central claims priority when checking evidence.

Expand the notes when an argument depends on a disputed source, a difficult interpretation, or several competing explanations. The amount of documentation can vary; the need to support the central argument remains. If time is limited, narrow the question or conclusion rather than conceal an evidence gap.

Making the feedback loop explicit

The working loop is: propose → test → identify a problem → revise → test again. Evidence, a resistant passage, an objection, or a reader’s misunderstanding can provide feedback. Record what revealed the problem, what changed, and whether the change addressed it. Repeated approval alone supplies no such check.

The demonstration essay will also test the framework itself. When a stage is unclear, duplicates effort, or misses an important question, record the difficulty, revise the instruction, and retry that part of the essay. This is how the working concept can develop through use. Neither loop implies improvement of the underlying AI’s intelligence.

1. Establish the Purpose

State what you want to answer, who you are writing for, and what the essay should help that reader understand. Decide whether the main task is explanation, interpretation, evaluation, or persuasion.

Record the relevant constraints: length, deadline, expected background knowledge, source requirements, and rules governing assistance. Identify what falls outside the essay.

In Writing Next, Graham and Perin report positive effects for explicit writing-strategy instruction and specific product goals in the adolescent studies they reviewed. This offers empirical support for exploring deliberate planning and revision. The interventions involved teaching, however; reading our framework is not equivalent to receiving that instruction. (Graham and Perin 15–17, 42)

Write a brief starting position before asking AI to supply one: “At present, I think … because …; I am uncertain about ….” If you know too little to take a position, record what you need to find out. A baseline should preserve your actual understanding.

Output: A focused question, intended reader, purpose, constraints, and starting note.

Check: Can you explain what would count as answering the question within the available space and time?

2. Examine the Frame

Investigate how the wording of the question shapes possible answers. Define terms whose meanings could change the conclusion. Identify assumptions, false choices, and relevant perspectives that the question leaves out.

Distinguish the kind of question being asked. “What happens?”, “What does this passage mean?”, and “What should we do?” call for different kinds of support. An evaluative essay needs explicit criteria; evidence about consequences does not by itself establish which consequences should matter most.

Reframe the question when doing so reveals a more precise problem. Keep track of the original task so that reframing does not become an excuse to answer a different assignment.

Identify what the claim is about: the features of a work, its maker’s intentions, or its effect on an audience. A close reading can support an interpretation without establishing either intention or actual reader response. Choose evidence that can answer the kind of question being asked.

Return to the thought experiment: what distinction might a better-informed thinker notice in the question? State the proposed distinction plainly, then examine whether it changes the possible answers. Imagining greater knowledge does not supply the missing knowledge.

Output: A refined question, key definitions, assumptions to examine, and any evaluative criteria.

Check: Would someone who disagrees recognize the question as a fair statement of the issue?

3. Develop Competing Explanations

Consider plausible answers before settling on a thesis. For an explanatory essay, compare mechanisms. For a literary essay, compare readings of the text. For an evaluative essay, compare judgements and the criteria behind them.

Develop the strongest relevant alternative, including evidence that could favour it. Ask whether the alternatives disagree outright or apply under different conditions. Do not invent an opposing position merely to fill a slot, or give weak claims equal weight for the appearance of balance.

Lord, Lepper, and Preston found that inducing participants to consider an opposing possibility reduced bias more than general appeals to fairness in two social-judgement experiments. That gives us a reason to test a specific alternative rather than merely promise open-mindedness. Applying this finding to essay development is our inference; the experiments did not test essay quality. (Lord, Lepper, and Preston 1231)

Where appropriate, name the observation, passage, or finding that would help distinguish the alternatives. In interpretive work, ask which reading accounts for more relevant detail with fewer strained assumptions.

A connection to another field should explain something specific. An interesting analogy can suggest a question; it cannot establish the answer.

Output: A provisional thesis or explanatory focus, a credible alternative where one exists, and a way to compare them.

Check: Have you described the alternative fairly enough that it could change your position?

4. Ground the Argument

For each important claim, record its wording, supporting material, exact locator, and the reasoning connecting the two. Explain the limits of the support and note relevant counterevidence.

This distinction has a precedent in the Toulmin model as explained by Purdue OWL: an argument includes a claim, supporting grounds, and a warrant connecting them. We use that distinction to ask why particular evidence supports a conclusion. Purdue’s account is an instructional explanation of an argument model, not an experimental validation of our evidence map. (Purdue Online Writing Lab)

Keep these distinctions visible in your notes:

• Supported finding: Check the underlying source and represent what it establishes within its actual scope.
• Interpretation: Identify the evidence and explain the reasoning from that evidence to your reading.
• Hypothesis: State what would support or weaken the proposed explanation.
• Speculation: Label the possibility and explain why considering it is useful.
• Unresolved question: Preserve the gap rather than inventing an answer.

These labels describe how you are using a claim; they are not scores for certainty. An interpretation can be strongly supported. A reported finding can remain contested. Where a sentence combines observation and interpretation, separate the parts in your notes. For a value judgement or recommendation, also state the criterion or premise on which it depends.

Ask: Does this evidence support this exact claim, or merely something nearby? A passage may support a reading of one scene without supporting a claim about every work by its author. A study of an immediate outcome may leave a long-term claim unresolved.

Check source quality, context, methods where relevant, and access. Several articles repeating the same study are several reports of one evidentiary basis; trace the relationship rather than counting links as independent confirmation. Record whether you inspected the original or relied on an intermediary. Verify quotations and keep quotation, paraphrase, and your own analysis distinct.

For literary or cultural interpretation, identify the work and edition or version, locate the passage or scene, and explain which details support the reading. Check whether the source’s scale, completeness, or quality lets you examine the relevant detail; a small reproduction or isolated excerpt may conceal something consequential. Examine details that resist it as closely as those that fit. Distinguish a narrator’s or character’s statement from a claim about the author. When historical context matters, source that context and explain its bearing on the work; context alone does not settle an interpretation.

Output: A claim-and-evidence map with locators, reasoning, support limits, counterevidence, and gaps.

Check: Can a reader trace each central claim to appropriate support and see what remains uncertain?

5. Challenge and Refine

Test the argument before investing heavily in polish. Address its strongest relevant objection, a difficult counterexample, and any assumption carrying more weight than you initially recognized.

For causal arguments, examine whether the evidence distinguishes your proposed cause from plausible alternatives. For interpretations, examine passages that resist the reading. For evaluations, test whether you apply your stated criteria consistently.

State what would change your mind. Record your response to criticism: accept it, accept part of it, or explain why it does not warrant a change. A confident critique still needs reasons.

Ask what a thinker with knowledge we currently lack might challenge. Turn that imagined challenge into a question we can investigate, such as whether an omitted case changes the thesis. Keep it unresolved until evidence or reasoning warrants an answer.

Where competing positions explain different parts of the problem, investigate whether a precise distinction can account for both. Do not force synthesis when the evidence favours one position or leaves a real disagreement.

Output: A revised thesis with clear boundaries, a response to the strongest objection, and a brief record of consequential changes.

Check: Has the objection been answered with evidence or reasoning, rather than softened wording alone?

6. Build the Explanation

Organize the essay around what the reader needs to understand. Establish essential context and definitions before asking the reader to rely on them. Order the sections so that each prepares the next step in the argument.

Across each major section, make the claim, support, reasoning, and relationship to the central question clear. Treat these as functions to fulfil, not a compulsory four-part paragraph formula. An example can clarify a claim without proving that it is generally true.

Draft in your own words from the argument and source notes. Keep unresolved evidence visibly marked in the working draft. If explaining a step proves difficult, check whether the problem is missing reasoning before replacing the words.

After drafting, summarize each paragraph’s function in the margin. Use this reverse outline to identify repetitions, gaps, and passages that do not advance the purpose. Recheck citations and qualifications after substantial revision.

Output: An outline and a draft whose reasoning the intended reader can follow.

Check: Can the reader see why the conclusion follows, including its limits?

7. Evaluate and Learn

Review both the essay and your understanding:

• Accuracy: Are facts, quotations, and source representations dependable?
• Depth: Does the essay explain relevant mechanisms, assumptions, or meaningful distinctions?
• Coherence: Do the conclusions follow from the reasons?
• Fairness: Are credible alternatives represented accurately and given appropriate weight?
• Clarity: Can the intended reader understand and assess the argument?
• Learning: Can you explain the thesis, its strongest support, its limits, and an important revision without relying on AI wording?

Use a written explanation, annotated outline, or conversation for the learning check. Consult sources for exact quotations and details as needed; independent understanding does not require memorizing every citation. Difficulty explaining a point identifies something to revisit and is not, by itself, evidence of misconduct.

Record who performed the research, reasoning, and composition. If AI supplied those parts, a finished essay can demonstrate the workflow, but it does not establish that the human writer can perform them independently. Keep that learning question open until the writer makes a separate attempt.

In their 2013 review, Dunlosky and colleagues rated self-explanation as having moderate utility, while identifying gaps concerning durability, classroom application, and time demands. This supports investigating explanation as part of learning. Our proposed check of a writer’s understanding is an adaptation, not a validated assessment of independent writing ability. (Dunlosky et al. 14, sec. 2.5)

Compare the finished work with your starting note. Record one change in your understanding and one practice to carry into the next essay. A later attempt on a new question with less assistance can offer a further check of what you can do independently. One successful essay cannot establish that the framework caused lasting improvement.

Revise the weakest consequential element first. Stop polishing when a pass produces no identifiable improvement. If a central claim still lacks necessary support, investigate, narrow it, remove it, or explicitly frame the essay around the unresolved question. A deadline or a lack of new AI suggestions does not resolve that gap.

Output: An essay ready for its intended use, with remaining limitations visible, and one specific lesson for the next task.

Check: Have substantive problems been resolved or honestly bounded, and can you account for the argument yourself?

Working With AI

Use AI to propose alternatives, ask questions, compare structures, or identify possible weaknesses when the assignment or publication permits it. Keep responsibility for the thesis, interpretation, source checks, and retained wording with the writer.

Check suggested sources directly. Multiple AI critiques may prompt different questions, but agreement among them is not independent verification. Repeatedly revising an essay with AI changes the work; it is not recursive improvement of the AI itself.

Record substantive assistance and disclose it as required. Before accepting a revision, explain what problem it solves and why the change is justified. Where independent writing is the goal, decide which reasoning and composition decisions you will practise yourself.

Questions to Carry Forward

The seven stages leave us with seven prompts for considering how an argument develops:

1. Purpose: My question, reader, constraints, and starting position are …
2. Frame: The terms, assumptions, and criteria that matter are …
3. Alternatives: My provisional answer and strongest relevant alternative are …; I can compare them by …
4. Evidence: My central claims are supported by … at …; the reasoning is …; the limits are …
5. Challenge: The strongest objection is …; it changes or does not change my position because …
6. Explanation: The reader needs to understand these steps in this order …
7. Learning: I initially missed …; noticing it changed …; what remains uncertain is …; next time I will …

Keep returning to the question that prompted the framework: What are we failing to notice, and how would noticing it change the argument?

About This Exploration

This article develops a concept first explored in a conversation between the author and AI. AI assisted with its structure, critique, wording, and subsequent source research. Our first demonstration, “Before the First Page: What a Book Cover Asks Us to Believe”, tests the framework through the original Gatsby jacket. Its account of the seven stages records how the argument changed and why stages 2, 4, and 7 were expanded. The essay and these refinements remain under review. A single essay can reveal problems and suggest improvements; it cannot establish general effectiveness.