# Novel II — Deep Craft: Showing vs Telling
In ["Novel II — Deep Craft: Point of View"](/harrys-desk/novel-ii-deep-craft-point-of-view) I said POV is the fundamental contract between reader and story — who sees, from what distance, with what constraints. Today the question is not who sees but how much the reader feels. And that question, in every writing workshop since the 1980s, gets the same answer: show, don't tell. It is the most repeated piece of writing advice in the English language. It is also, as stated, wrong.
Let me be careful. The instinct behind "show don't tell" is sound. Telling — summarizing, explaining, naming the emotion — is often weaker than showing — dramatizing, embodying, letting the reader infer. But "always show, never tell" is a recipe for novels that are all foreground, no rhythm. Novels where every emotion is demonstrated through a physical gesture, every piece of backstory is delivered through a flashback scene, every transition is dramatized in real time. These novels are exhausting. They are also, paradoxically, less vivid than novels that tell well, because constant showing flattens the contrast that makes showing powerful.
The real craft is not showing instead of telling. It is knowing when to do which, and — this is the part nobody teaches — managing the rhythm between them.
What Showing Actually Does
Let me define terms precisely, because the looseness is where the trouble starts.
Showing is dramatization. The reader is present in the moment, experiencing it in something close to real time. The writer renders the scene through sensory detail, action, and dialogue. The reader infers meaning from behavior rather than receiving it from narration. "She gripped the armrest until her knuckles went white" is showing. The reader sees the fear without being told "she was afraid."
Showing does three things that telling cannot. First, it creates immediacy. The reader is inside the moment, not hearing about it after the fact. Second, it creates emotional depth through inference. When the reader assembles meaning from behavior, they participate in the meaning-making, and participation creates ownership — the reader owns the emotion they constructed more than the emotion they were handed. Third, it creates trust. The writer who shows is trusting the reader to get it. The writer who tells is doing the reader's work for them, and the reader can feel the condescension even if they cannot name it.
These are real advantages. They are why "show don't tell" exists as advice. But they come with costs that the advice never mentions.
What Telling Actually Does
Telling is summary. The narrator compresses time, conveys information, moves the reader from one moment to another without dramatizing every step. "She was afraid of water her whole life, though she never told anyone why" is telling. It covers years in one sentence. No amount of showing can do that.
Telling does three things that showing cannot. First, it controls pace. A novel cannot dramatize everything. If you show every conversation, every meal, every drive, every transition, the novel drowns in its own detail. Telling is how you skip the parts the reader does not need to experience and arrive at the parts they do. Second, it provides context. Showing is present-tense and immediate; telling can reach across years, can establish patterns, can name what the character does not yet understand about themselves. "She had been angry at her father for so long that the anger had become a climate, not a weather event" — that is telling, and no single scene can convey it as efficiently. Third, it creates contrast. When every moment is shown, nothing is emphasized — it all has the same weight. When most of the novel is told and certain moments are shown, the shown moments detonate. They are the islands of intensity in a sea of summary, and the sea is what makes the islands visible.
This is the part the "show don't tell" advice gets backwards. Telling is not the failure mode. Bad telling is the failure mode. Good telling is invisible — the reader does not notice they are being told because the telling is so clean and well-placed that it feels like narration. The reader only notices telling when it is clumsy: when it names emotions the reader could have inferred, when it explains what the reader already knows, when it lectures. The problem is not summary. The problem is lazy summary.
The Rhythm Between Them
Here is what I want you to take from this article, more than anything else: a novel is not a collection of shown scenes connected by told transitions. A novel is a rhythm of showing and telling, and the rhythm is the art.
Think of it musically. Showing is the downbeat — the accented, present, dramatized moment. Telling is the upbeat — the compression, the transition, the movement through time. A novel that is all downbeats is exhausting. A novel that is all upbeats has no weight. The art is in the alternation: show the moments that matter, tell the connective tissue, show again, tell again, and let the reader feel the shift in intensity as a shift in meaning.
The rhythm also operates within scenes. A scene is not pure showing — it contains moments of telling within it. The narrator summarizes a character's habit ("he always checked the locks twice") in the middle of a dramatized scene, and that summary creates texture without breaking the scene's forward motion. Conversely, a passage of telling can contain a shown moment embedded in it — a single image that grounds the summary in sensory reality. The cleanest novelists move between showing and telling within paragraphs, within sentences, and the movement is so smooth the reader never feels the gear change.
The AI Thoroughness Problem
Now I want to talk about the specific failure mode that AI-assisted drafting introduces into this rhythm, because it is pervasive and it is getting worse as more writers draft with models.
Models are thorough. When you ask a model to write a scene, it will write the scene completely — every moment dramatized, every transition rendered, every emotion embodied in physical gesture. This is not a flaw. It is what the model is trained to do. The training data rewards completeness. A model that summarizes is penalized for being "too brief." A model that dramatizes is rewarded for being "vivid." So the model shows. And shows. And shows.
The result is a draft where everything has the same intensity. The crucial confrontation and the drive to work are rendered with the same level of detail. The character's epiphany and their breakfast are both shown in full. The model does not know which moments deserve the downbeat and which are upbeats, because the model does not know the novel's emotional architecture. It knows the prompt. And the prompt says "write this scene," so it writes the scene — completely, vividly, thoroughly.
This is the AI thoroughness problem: the model flattens the rhythm by showing everything. And the flattened rhythm is not something you can fix in a line edit. It is a structural problem. You have a draft where nothing stands out because everything is shown, and the only way to fix it is to identify which moments should be told instead — and then cut the shown version and replace it with a summary. You are un-showing. You are taking vivid, complete prose and replacing it with compression. This feels wrong to most writers. You wrote the scene. The model wrote it well. It has detail, it has gesture, it has texture. Cutting it feels like destruction.
It is not destruction. It is composition. The scene you cut is not wasted — it informed your understanding of the moment, and that understanding now lives in the summary that replaces it. But the summary serves the novel better than the scene did, because the summary creates contrast. It makes the scenes you keep more powerful by giving them something to stand against.
How to Use AI for the Rhythm Audit
The model that flattened your rhythm can also help you see the flattening, if you ask the right questions.
The intensity map. Paste a chapter and ask: "Rate the narrative intensity of each paragraph on a scale of 1-5, where 5 is fully dramatized real-time scene and 1 is pure summary. Then identify whether there are long stretches at the same intensity level — more than five consecutive paragraphs at 4 or 5, or more than three at 1 or 2." The model will give you a map of where your rhythm goes flat. If you see seven consecutive 5s, you have a thoroughness problem. If you see the whole chapter at 4, you have a novel that is all downbeat.
The summary opportunity. Paste a scene that you suspect is over-shown and ask: "If this scene were compressed to a single summary paragraph, what would it lose? What would it gain in pace?" The model's answer is often illuminating because it will name what the scene contributes — and if the only thing it contributes is information the reader could get from a sentence of telling, you have found a scene to cut.
The dramatization test. Conversely, paste a passage of telling and ask: "Is there a moment in this summary that the reader should experience rather than be told about? Is there an image, a gesture, a line of dialogue that would detonate if it were shown instead of summarized?" Sometimes the model will identify a moment that is currently buried in summary and deserves to be pulled out and dramatized. The rhythm works both ways. You over-show and you under-show, and the audit helps you see both.
A Working Method
Here is how I manage the showing-telling rhythm in practice.
When I draft with a model, I expect the draft to be over-shown. I do not fight it during drafting. The model produces complete scenes, and I let it, because the complete scene often reveals things I did not know about the moment — a gesture, a line of dialogue, a detail that matters. I mine the draft for those discoveries. Then, in revision, I ask which moments should stay shown and which should be compressed to summary.
The test I use: does this moment change what the reader understands about the character or the story? If yes, it stays shown. If it conveys information the reader needs but does not change their understanding, it becomes telling. The confrontation where the character's secret is revealed stays shown. The drive to the confrontation becomes a sentence. The aftermath, if it changes understanding, gets shown again. If it is just denouement, it gets told.
I also watch for the emotional arc within a scene. A scene should have a turn — a moment where the emotional temperature changes. Everything before the turn is setup; everything after is consequence. If the setup is long and the turn is brief, I compress the setup. If the turn is the most powerful moment in the scene, I make sure the rhythm of the scene pushes toward it — shorter paragraphs, more telling, less detail, until the turn arrives and the prose opens up into showing.
And I read for monotony. If I cannot feel the gear changes between showing and telling — if the prose has one consistent texture from page one to page twenty — the rhythm is flat. A novel with a good rhythm feels like breathing. Inhalation and exhalation. Expansion and compression. The reader does not think about this consciously. They feel it as pace, as engagement, as the sense that the novel is alive.
For Next Time
On Friday we continue Novel II — Deep Craft with ["Novel II — Deep Craft: Pacing and Tension"](/harrys-desk/novel-ii-deep-craft-pacing-and-tension). Showing and telling determine the rhythm of the prose. Pacing and tension determine the rhythm of the reader's experience — the engine that keeps them turning pages. We will talk about the difference between pacing (how fast the story moves) and tension (how much the reader wants to know what happens next), the mistake of confusing the two, and how to use AI to diagnose where a draft loses narrative drive without mistaking fast prose for engaging prose.
Your homework until then: take two consecutive chapters from your draft. In the first, identify every passage of showing and every passage of telling. Do not judge them — just map them. In the second chapter, do the same. Now compare the maps. Is the rhythm consistent between chapters, or does one chapter show more and the other tell more? Is there a pattern — do you show in the first half and tell in the second, or vice versa? The map is not a verdict. It is a diagnosis. The question is not whether you are showing or telling. The question is whether the alternation between them has a pulse.
---
*Harry Mercury, Editor in Chief* *The SMF Works Project* *Week 14, Article 2*