Gawande's New Yorker editor Henry Finder taught him a craft distinction Gawande still uses. He'd literally talk about, you know, this uh this one is an O, meaning where starting is where it ends. It's going to it's going to it's going to start where it ends.1 Other pieces: this one's more like a W. We got three peaks and we're going to work our way through and we're going to advance from left to right.2
Two opening-shapes. The O opens with where the piece will end, then circles back to it through the body. The W opens at a starting position and traverses three peaks before landing somewhere new. Different pieces want different shapes. Knowing which shape your piece is governs how you write the opening.
The opening discipline is structured by two biases Gawande names: humans have a against boring bias3 (we attend to things that aren't boring) and we also have a bias not only for not boring, but also for not predictable4 (we attend to things that surprise us). The opening has to satisfy both. Predictable opens fail even when not boring. Boring opens fail even when unpredictable.
O-vs-W is the opening-shape diagnostic for long-form pieces. The shape choice determines how the opening relates to the rest of the piece:
O-shape: The piece begins at the destination, then circles through territory to return there. The opening states or implies the ending; the body shows the path that produces the ending. Useful when the destination is a specific argument or insight that gains weight from being foreshadowed.
W-shape: The piece begins somewhere distinct from where it ends. The body traverses multiple peaks (three turns, multiple discoveries) before arriving somewhere new. Useful when the journey itself is the content, when the writer wants the reader to traverse the discovery alongside them.
The against-boring + against-predictable biases govern both shapes. The opening must surprise (against predictable) and engage (against boring). Predictable-but-engaging openings fail. Engaging-but-predictable openings also fail. Both biases must be satisfied.
The need fires for writers whose openings consistently land flat. Either the opening is predictable (every chapter opens with the patient in the emergency room) or boring (background context first, then engagement). The diagnosis is missing opening-shape awareness — the writer is defaulting to a single shape without considering what the specific piece requires.
Three mechanisms:
Against-boring is a universal cognitive bias. Audiences attend to novelty. Openings have seconds to demonstrate novelty. Failed openings produce reader-bounce.
Against-predictable is an additional layer. Audiences also detect predictability — the opening with a dramatic medical case pattern is recognizable, and recognition produces predictability-detection, which produces disengagement.
Shape-fit reduces both failure modes. An O-shape applied to W-content reads as both circular and inert. A W-shape applied to O-content reads as wandering. The shape has to fit the piece's structure, not be applied as default.
O-vs-W pairs with find-a-home opening craft (Roth) — the threshold-discipline operates regardless of shape, but shape determines how the threshold gets crossed. It pairs with hook-as-opening-compression — the hook is the first sentence of either shape; the shape governs what the hook should accomplish.
O-shape example (Gawande on hand-washing): The piece could open with Two million people get hospital infections per year because someone didn't wash their hands. The number is the destination. The body shows how a single infection-control specialist navigates a hospital trying to change the hand-washing behavior. The piece returns to two million with new weight by the end. The reader started with the number, now they feel what the number costs.
W-shape example (a different Gawande piece): The piece could open with a specific patient's story. The body traverses multiple cases, multiple framings, multiple turns. The ending lands on an insight the opening didn't preview. The reader follows discovery rather than circling.
Both shapes work. The choice depends on what the piece is. The discipline is recognizing which shape this specific piece needs, not defaulting to one shape across all pieces.
You're starting a piece. Before writing the opening, ask: is this an O or a W?
Is the destination already clear in your head? If yes, you may be writing an O — the opening can foreshadow the destination. Is the destination something you'll discover through the writing? If yes, you may be writing a W — the opening should establish the starting position without claiming to know the destination.
Does the piece have one main insight? O-shape often. Does the piece have multiple discoveries that compound? W-shape often.
Write a candidate opening. Test it against both biases. Is it boring? (Cut.) Is it predictable? (Restructure.) Only opens that pass both tests should survive.
Iterate the opening more times than you iterate other parts. The opening is where the most reader-loss happens. Disproportionate iteration on the opening produces disproportionate returns.
You'll know openings have failed when readers describe pieces as good but I almost stopped reading. They got past the opening but only barely. The fix is opening-craft, not body-craft.
You'll know shape-mismatch has occurred when the body of the piece doesn't fulfill what the opening promised structurally. The opening was O-shaped but the body is W-shaped, or vice versa. The mismatch produces reader-confusion.
Roth's I always know the beginning and the ending is structurally aware of opening shape — Roth's beginning is the O-shape end-foreshadow or the W-shape starting-position. Same underlying discipline at different vocabulary. Bilton's narrative-nonfiction methodology operates similarly.
The compound insight: O-vs-W openings are attention-economy strategy + expectation-violation cognition applied to long-form opening craft. The principle has empirical grounding.
The Sharpest Implication
If your openings consistently fail, the diagnosis is usually missing shape-awareness. The fix is asking which shape this specific piece is and writing the opening accordingly, rather than defaulting to one shape across all pieces.
Generative Questions