David Sedaris is one of the most celebrated American essayists alive. His personal essays — Me Talk Pretty One Day, Calypso, the New Yorker pieces — are formally brilliant, line-by-line precise, and emotionally devastating on the page. Readers report being moved to tears and laughter by single paragraphs. The work is unquestionably good writing.
Sedaris also performs his essays at live readings across the country. At the lectern, with the printed text in front of him, he reads the essays aloud to audiences who have paid to hear them. The performance is competent — Sedaris's delivery is dry, his timing is good, the audience laughs. But the performance does not produce the audience experience that a great Moth-tradition oral storyteller produces with comparable runtime. The audience enjoys the reading. They do not leave it transformed.
Matthew Dicks's analytical point: writing and oral telling are different arts. Skills transfer partially in some directions but not in others. Sedaris is exhibit A — exceptional written-craft, competent-but-not-exceptional oral delivery. The reverse case is also common — exceptional Moth storytellers whose written transcripts read flat because the moves that made their oral telling work do not survive transcription to the page.1
The principle is not a criticism of Sedaris or of Moth storytellers. It is a clarification: writing and telling are distinct disciplines with overlapping but non-identical craft repertoires. Storytellers who treat them as the same art under different delivery channels produce work that is suboptimal in whichever channel they did not specifically train for.
The principle has three components.
1. Writing and telling are distinct arts with non-identical craft repertoires. Some skills transfer (structural sense, the ability to identify the load-bearing moment, narrative discipline). Some skills do not transfer (writing's sentence-level cadence and word-level precision; telling's vocal-level pacing, eye-contact discipline, in-moment recovery). The two disciplines have a shared foundation (narrative skill) and distinct upper layers (the medium-specific tactical refinements).
2. The Sedaris case demonstrates writing-exceptional-without-telling-exceptional. Sedaris reads his essays. The reading is an author reading — a written text delivered orally. The audience experiences it as the reading of a great essay, not as the telling of a story. The two experiences are not the same; the second produces engagement the first does not.
3. The reverse case is symmetric. Moth storytellers whose live performances are exceptional sometimes produce written transcripts that read flat. The moves that produced their oral success — doubling-back, vocal pacing, conversational hesitation, real-time eye contact — do not transfer to the page. The page-reader receives a transcript that describes the events but does not produce the engagement the live audience experienced.
The cognitive mechanism: writing and telling place different demands on different audience-side cognitive processes.
Reading is asynchronous and the reader has control of pacing. The reader can dwell on a sentence, re-read, pause to think, set the book down for an hour and return. The writer's craft must work across all these reader-pacing options — a sentence must reward dwelling, support re-reading, hold up to thinking. This produces a craft that emphasizes line-level precision — every sentence must be load-bearing because every sentence is a potential dwell-point.
Listening is synchronous and the audience does not control pacing. The listener cannot pause; they cannot re-listen mid-stream; the speaker controls the pacing. The teller's craft must work moment-to-moment without dwell-points — the listener has only the current second to receive what the teller is delivering. This produces a craft that emphasizes moment-level engagement — every moment must be doing something the listener can absorb in real time.
The two craft-emphases are different. Writing-craft optimizes for the sentence that rewards dwelling; telling-craft optimizes for the moment that holds attention without dwelling. The optimizations point in different directions. A sentence that rewards dwelling may be too dense for moment-to-moment listening; a moment that holds attention in listening may be too thin for sentence-level reading-craft.
A second mechanism: writing has no body; telling has a body. Writing-craft does not have access to vocal pacing, eye contact, body language, or in-room responsive timing. Telling-craft has all of these as available tools. The two crafts therefore work with different toolsets — the teller has tools the writer does not have, and the writer has tools the teller does not have (typography, sentence-level rhythm, white-space, return-to-passage).
A third mechanism: writing's audience is imagined; telling's audience is present. The writer constructs an imagined reader and crafts for that imagined reader. The teller constructs for a present audience and adjusts in real time to the present audience's response. The two craft-orientations are different at the cognitive level — writers cannot read the room because there is no room; tellers cannot ignore the room because the room is in front of them.
Transfers from telling to writing:
Does not transfer from telling to writing:
Transfers from writing to telling:
Does not transfer from writing to telling:
The shared foundation layer (structural and narrative skill) is what makes cross-training possible — strong writers can become strong tellers and strong tellers can become strong writers, given the medium-specific training. The distinct upper layers are what makes either-without-the-other suboptimal — writers who do not train in telling produce author-readings; tellers who do not train in writing produce transcripts that read flat.
The principle is foundational to the Dena principle (structure trumps sentences) in oral storytelling specifically. The structure-trumps-sentences observation applies most strongly in the oral medium because the medium does not support sentence-level dwelling. In writing the structure-trumps-sentences ratio shifts toward sentences having more weight because the reader can dwell. The two principles together describe how the structure-versus-sentence balance is medium-dependent.
It pairs with punch sentence as own paragraph: Dicks's specific advice for writing (give punch sentences their own paragraph; mimic vocal performance through typographic isolation) is the writing-craft counterpart to the oral-craft technique of vocal-pause-around-the-punch-sentence. The cross-medium analogy demonstrates that some oral-craft techniques have equivalents in writing-craft, even though they do not literally transfer.
It connects to strategic listening: storytellers operating across both media need to apply the meta-skill separately to written artifacts and to oral artifacts. The libraries built from reading great essays differ from the libraries built from listening to great oral storytellers. Storytellers who work in both media maintain two libraries.
It contrasts with imperfection is the beauty (Catherine Burns / Moth) in a medium-specific way: the maxim applies fully in oral telling (where imperfection signals presence); it applies only weakly in writing (where deliberate imperfection in prose can work but typically as a controlled stylistic choice rather than as a presence-signal).
Consider two cases in detail.
Case A — David Sedaris reading "Stadium Pal" at a live event. The essay on the page is a model of personal-essay craft — vivid setup, specific sensory detail, sustained comic tension, devastating final image. Reading time on the page: roughly 12-15 minutes for a typical reader. At the live event, Sedaris stands at the lectern, the printed text in front of him, and reads the essay aloud. His delivery is dry, his timing is professional, the audience laughs at the funny lines. The essay's craft is unmistakable. The audience's experience is enjoying a fine essay read aloud by its author.
The experience is not what an oral-storytelling-tradition performance of comparable content would produce. The oral version would tell the story rather than read it — eye contact rotating among specific audience-members, no printed text in front of the performer, doubling-back when memory called for it, vocal pacing modulated by the present-room response. The oral version would change the audience-members in a way the reading does not. The reading delivers a great essay; the telling delivers a great experience.
The instructive point is not that Sedaris's reading is bad. It is that the reading is a different art than oral storytelling, and conflating the two leads to mis-calibrated audience expectations. Sedaris-as-essayist is one of the best alive; Sedaris-as-oral-teller is competent but not exceptional. The two evaluations coexist because the two arts are distinct.
Case B — A Moth StorySLAM champion whose written transcript reads flat. Consider a hypothetical (composite from many Moth performances): a storyteller wins three StorySLAMs in a row with a story about her grandmother. Live, the story devastates audiences — multiple performances bring rooms to silence-then-applause; audience-members report being moved to tears; the story is requested for retelling at multiple events.
The story is transcribed for a podcast script. The transcript is read by a podcast editor unfamiliar with the live performance. The editor reports the story is fine but does not understand the live response. It just doesn't pop on the page, the editor reports. The transcript captures the events but does not capture the live moves — the specific vocal pause before the line about her grandmother's hand on the kitchen counter; the way the storyteller's voice caught slightly on the line; the eye contact she held with a specific audience-member during the heart-landing. All of those moves were doing the work the transcript cannot recreate.
The instructive point: the Moth storyteller's craft is oral-specific. Transcribing the performance strips the load-bearing elements. The audience that read the transcript would never know what made the live performance devastating; the audience that heard the live performance could not articulate which specific moves did the work, but the work was done.
The two cases bracket the principle. Strong writing without strong telling produces Sedaris at the lectern. Strong telling without strong writing produces the Moth champion's transcript. Both are partial; both are valuable in their own medium; neither transfers cleanly to the other.
You are a storyteller who has just written a piece. The piece is going to be delivered in two contexts — first as a stage performance at a storytelling event, then as a written essay submitted to a literary magazine. The temptation is to use a single draft for both contexts.
It is Sunday afternoon, 4:14pm. You're at the kitchen table with the draft. You read it aloud as if performing it. You notice: the sentences are too long for listening; some passages reward dwelling that the listener cannot do; the load-bearing moment is constructed at sentence-level precision that the listener will not have time to register. The piece is a written essay, not a told story.
You consider the options. Option 1: deliver the written piece as a stage performance anyway, accepting that the stage performance will be sub-optimal. Option 2: create a second draft of the same content optimized for oral delivery — shorter sentences, no dwelling-passages, load-bearing moment constructed at moment-level rather than sentence-level.
You choose Option 2. You set the written draft aside. You take a blank sheet and you tell the same story aloud, recording yourself. You transcribe the recording. The transcript is the oral draft — fragments, conversational vocabulary, hesitation-and-recovery, moment-level rather than sentence-level construction. The oral draft is the same story but is a different artifact than the written draft.
You revise both. The written draft is polished for the literary magazine. The oral draft is rehearsed for the stage performance. The two artifacts cover the same content but are optimized for their distinct media. The audiences for each get the appropriate version. Neither audience receives a draft that was optimized for the other medium.
The pattern is: identify which medium each version of the piece is for, develop separate drafts optimized for each medium, accept the additional drafting work as the cost of doing both arts well, deliver each draft in its appropriate context without expecting the other version to substitute.
Your written essays land flat at live readings. The essays were optimized for the page; the live reading exposes their oral-suboptimality. The fix is to develop a separate oral draft rather than read the written draft aloud.
Your oral stories read flat as transcripts. The stories were optimized for live delivery; the transcript strips the load-bearing oral moves. The fix is to develop a separate written draft if you want the content in written form, not to transcribe the oral performance.
You produce one draft and deliver it in both media. This is the symptom of the conflation failure. The single draft is necessarily sub-optimal in at least one medium. The fix is to develop distinct drafts per medium.
You assume your skill in one medium transfers fully to the other. Writers who assume their skill makes them strong tellers, and tellers who assume their skill makes them strong writers, are operating the conflation failure at the meta-level. The fix is to train deliberately in the second medium rather than relying on the assumption.
Your audience responses differ dramatically across media even for the same content. When the same story works in one medium and falls flat in the other, the difference is usually the medium-specific craft. The fix is to develop the medium-specific craft for the failing medium.
Dicks across both interviews holds the writing-vs-telling distinction as one of his clearest medium-specific framings.2 He emphasizes the partial transferability — some skills do transfer (structural sense most importantly), but the medium-specific upper layers must be developed separately. Dicks's framing avoids overclaiming the distinction (he does not argue that writing and telling are completely unrelated; the foundational layer is shared) and avoids underclaiming (he does not argue that strong writers are automatically strong tellers; the upper-layer separateness is real).
The principle is somewhat tension with the Dena principle in a productive way. The Dena principle says structure dominates sentences in oral telling; the writing-vs-telling principle says writing requires sentence-level craft that oral telling does not. The two principles together describe how the structure-versus-sentence ratio shifts by medium — closer to all-structure in oral, closer to balanced in writing.
The principle has no direct counterpart in Lewis's synthesis (Lewis focuses on oral storytelling specifically and does not address cross-medium transferability). It is a Dicks-tradition contribution rather than a Moth-tradition cultural consensus.
Look at what the Sedaris case demonstrates: communicative craft is medium-specific, and skills transfer only partially across distinct media. Several other domains have analogous distinctions.
Music performance / composition vs improvisation: Music tradition distinguishes between composition (writing music for others to perform) and improvisation (performing music in real time). The two are related but distinct disciplines with bounded transferability. A great composer may or may not be a great improviser; a great improviser may or may not be a great composer. The structural parallel to writing-vs-telling is direct: both pairs distinguish asynchronous-pre-prepared from synchronous-real-time communicative production. Pairing music tradition with storytelling tradition gives each domain a vocabulary for what they each operationally know — music's centuries-long working through of composer-vs-performer skill distinctions can inform storytelling's more recent articulation of writer-vs-teller distinctions.
Theater / playwriting vs acting: The theater tradition formalizes a similar distinction — playwrights and actors are different artists with different craft repertoires. Some playwrights are competent actors; some actors are competent playwrights; the two skills are bounded-transferable. The cross-orientation parallel suggests the writing-vs-telling distinction is part of a broader pattern across performance arts: the artist who creates the artifact and the artist who delivers the artifact in real time are distinct artistic roles with non-identical skill requirements. The two-artist tradition (playwright + actor) has produced refinements (table-reads, workshopping, director-as-intermediary) that storytelling tradition might productively borrow.
Software engineering / design vs implementation: Software engineering distinguishes between system design (the architectural specification) and implementation (the line-by-line code that realizes the design). Strong designers may or may not be strong implementers; strong implementers may or may not be strong designers. The same partial-transferability applies. Pairing software tradition with storytelling tradition: software has formalized the separation of concerns between design and implementation as an explicit professional practice (different teams, different specializations, different career tracks); storytelling could productively examine whether explicit specialization tracks (the writer who specializes in writing; the teller who specializes in telling; coordination between them) would professionalize the field's craft-development.
The Sharpest Implication The writing-vs-telling principle implies that the storyteller's career-development decisions should be informed by which medium they intend to work in primarily, with deliberate secondary training in the other medium if they intend to operate in both. Most storyteller-training treats the two as one art, which produces writers whose live readings under-deliver and tellers whose published work under-delivers. The implication: storytelling pedagogy should explicitly separate the two crafts in its curriculum and develop medium-specific training tracks. Storytellers operating in only one medium can specialize without guilt; storytellers operating in both must train both separately.
Generative Questions