Most writers are taught a hierarchy: first you determine your argument, then you outline your structure, then you write sentences that serve that structure. Sentences are containers for ideas. The structure is primary. The sentences fill it.
Dan Wang inverts this relationship. He argues that it is "not at all an irrational thing to construct an entire essay around a single beautiful sentence."1 In fact, he recommends it. If you have one sentence that is genuinely excellent — that carries weight, that is precisely shaped, that says something important in an unforgettable way — then the essay can be built around that sentence. The structure emerges from the sentence's requirements, not the reverse.
This is radical because it means sentences are not servants of structure. They are generative. A single refined sentence can determine the direction, the pace, the emphasis, and the argument of an entire essay. The sentence is the anchor point. Everything else orbits it.
A truly excellent sentence does structural work that a merely adequate sentence cannot. It establishes tone. It signals the level of formality. It contains a compression of meaning — multiple levels of significance in a small space. It sets up expectations about what will follow.
When Wang says "not at all an irrational thing," he is not being poetic. He is being precise. If you have captured one sentence that is exceptionally good, that sentence has already solved several structural problems simultaneously. It has established the register (how formal, how playful, how serious). It has suggested what scale of analysis will follow. It has created a rhythm that the following sentences should match or deliberately contradict.
Building an essay around such a sentence means the rest of the writing has clear criteria: every subsequent sentence must justify its existence relative to that anchor. Does it clarify the anchor sentence? Complicate it? Extend it? Contradict it in productive ways? Sentences that do none of these are noise.
This is why capturing sentences throughout the year (the note-capture practice) becomes structurally important. You are not capturing random observations. You are hunting for sentences that are dense enough, precise enough, and generative enough that an entire essay could orbit them.1 A sentence that can do this kind of work has specific qualities: economy (no wasted words), precision (the exact right word, not approximately right), and depth (multiple meanings operating simultaneously).
The process Wang describes is concrete:
Throughout the year, capture sentences as they arrive. These might be completely disconnected from current projects. They are captured because they are good sentences, not because they serve an existing essay.
By late December, you have an accumulation of raw material. Many of these sentences are standalone. They do not yet belong to anything.
In the assembly phase, you begin connecting sentences to one another. You discover that three disconnected sentences actually resonate with each other. They are about different topics but they operate at the same tonal register. They have similar rhythmic qualities.
Suddenly, a cluster of sentences suggests a shape. An essay begins to form not from outline but from the sentences themselves demanding relationship with each other.
The gaps between captured sentences become the places where new writing is required. But the new writing is constrained: it must honor the tone, pace, and density of the captured sentences. It must justify their presence.
This is the opposite of starting with structure. Structure emerges from sentence selection and combination. The essay is not a container that sentences fill. The sentences are the skeleton around which other material accumulates.
Voice Cultivation Through Stylistic Models — The sentence-as-anchor approach directly requires the stylistic study that comes from deliberate influence-absorption. Without that study, you don't develop the capacity to recognize when a sentence is genuinely excellent versus merely competent. The ability to hear the difference is itself developed through close study of master sentences.
Note Capture and Refinement Practice — Sentence-anchoring is the structural payoff of disciplined capture. You capture sentences specifically so you will have material excellent enough to build around.
Prose as Transmission — Sentences operating at Level 5+ (narrator present, ornamental, functional flourish) are the ones that can serve as anchors. Level 2-3 sentences (bare transmission) cannot carry structural weight alone.
Identify your best sentence: Look back at captured material. Which one sentence is genuinely excellent? Not interesting. Excellent. Precise, economical, carrying multiple meanings.
Ask what it requires: This sentence contains certain claims or suggestions. What would need to be true for this sentence to be fully justified? What argument does it presuppose?
Find resonating sentences: Which other captured sentences operate at the same tonal register? Which are asking related questions? Start clustering.
Identify gaps: Where do the captured sentences create jumps that need bridging? These gaps are where you write new material. But the new material has clear constraints: it must honor the tone and density of the excellent sentences.
Test structure: Does the sentence collection, with bridging material, create an argument? Or are they still disconnected? If disconnected, you may not have enough material. Or you may have the wrong anchor sentence.
This is slow work. But it produces writing where nothing is wasted. Every sentence justifies its presence either because it is captured excellence or because it serves captured excellence.
The Sharpest Implication: If essays can be built around single sentences, then the bottleneck in writing is not outlining ability or argumentative skill — it is sentence-level craft. The ability to write a sentence worth building around is more valuable than the ability to organize arguments. This inverts the hierarchy most writing instruction maintains. It says: get your sentences right first. The structure will follow.
Generative Questions: