Structure: from reader stasis to earned closure
The essay moves through this sequence:
reader stasis → instability → consequence → evidence/mechanism → countercase/qualification → better model → earned closure
Evidence: the smallest sufficient source set
Include enough evidence to carry the claim, including a serious countercase.
Opening: enter the reader's problem quickly
Use a scene, a contradiction, or a direct claim, depending on genre.
Nut: make the route clear
Make the reader contract visible enough that the route is clear. See Module 7: The nut graf.
Ending: what the reader leaves with
End with recognition, a decision, a boundary, or a sharper unresolved question.
Main risks
Watch for these failures:
- trying to preserve the whole Thinking Essay
- writing for multiple readers
- using examples as proof without mechanism
- a dramatic claim with its caveat buried later
- a conclusion stronger than the evidence
Analytical memo variant: make the reasoning inspectable
The Howard Marks material suggests a distinct variant of the argumentative essay: a memo in which the reader is allowed to watch a judgment being formed and tested.
Use it when the subject is uncertain, mechanism-heavy, and consequential for a reader who must decide what to believe or do.
A useful architecture runs in ten steps:
- Occasion: what changed, repeated, or failed enough to deserve explanation?
- Live question: what uncertainty is the memo actually resolving?
- Provisional view: what do you currently think, stated strongly enough to test?
- Question ladder: what must the reader establish first, and what question follows from each answer?
- Visible logic: claim → mechanism → evidence → implication.
- Countercase: what serious objection changes or narrows the view?
- Calibration: what is fact, inference, judgment, expectation, and unknown?
- Compression: after difficult sections, restate the conclusion at higher compression.
- Practical return: what problem is still sitting on the reader's desk after the analysis?
- Ending: return to the governing question with a better model, not merely a summary.
The surface can stay conversational because the underlying structure is strict.
Do not imitate Marks's rhetorical questions, verbal tics, analogies, or aphorisms. The transferable craft is the visibility of the reasoning.
AI role: test the essay, then edit locally
Reader simulator, architecture critic, evidence challenger, final local editor.