The AI Journal

The instruction arrives in every workshop, and it is nearly always right and nearly always mis-stated. A paragraph is circled, a note in the margin says show, don’t tell, and the writer nods and promises to write more slowly. What has actually been identified is a passage in which the narrator has done the reader’s work: named a feeling, drawn a conclusion, handed over a judgement the reader might have reached unaided. The remedy proposed is imagery. The two are not the same thing, and the distance between them is exactly where the maxim earns its place and where it quietly does harm.

Take the two questions apart. The first is about mode of presentation: is a stretch of prose rendered as scene, with events unfolding in something like real time, or as summary, the narrator compressing what took months? This is the distinction Percy Lubbock drew between scene and picture in The Craft of Fiction, and it is what most writers hear when they hear the maxim. The second question is about inference: who performs the work of concluding, the reader or the narrator? Does the text say she was afraid, or does it arrange the hands and the door and let the reader arrive at fear? The two questions are independent. A scene can be saturated with judgement; a summary can be almost pure inference. The maxim names the first question while its real force lies in the second.

That the first question is even coherent has been doubted by better readers than the workshop. In Narrative Discourse, Gérard Genette argues that true mimesis is impossible in verbal narrative: language cannot present an event, only recount one, more or less closely, more or less at a distance. What we call showing is a will to show, an effect assembled from a chain of telling, from details selected and ordered so that a reader reconstructs what was never stated. Read literally, the maxim asks for something prose cannot supply. No arrangement of words produces a scene that is not also narration. Once that is granted, show cannot mean dramatise. It has to mean something about the reader, or it means nothing.

Wayne Booth had made the neighbouring point in The Rhetoric of Fiction. The author’s presence is inescapable: the choice to withdraw is itself a rhetorical act, and the invisible narrator of the moderns is a convention, not an absence. Booth’s target was the idea that good fiction simply gets out of the way. It does not; it can only disguise its interventions. Put Booth and Genette together and the maxim’s grammar dissolves. There is no place the writer can stand outside the telling. Every sentence is a decision about what to say, and every decision is a hand on the reader’s shoulder, however light.

If that is right, the operative variable is not scene against summary but inference. Seymour Chatman’s Story and Discourse treats showing and telling not as a binary but as a continuum of narratorial presence, along which a text may move in either direction, sentence by sentence. A text can dramatise a wedding and still tell the reader it was sad; a text can summarise a decade and still make the reader assemble its cost. The reader’s inferential work is what the felt difference tracks, and it can be raised or lowered on either side of the scene-summary line.

This is why the maxim so often misfires in revision. A writer told to show will add sensory detail to a passage that already has enough; the problem was never a shortage of images but a narrator who would not stop explaining. Conversely, a writer will cut an expository paragraph that was doing necessary structural work and call the loss subtle. The instruction points at the surface, and the surface is where the effect is least determined. What the circling actually registered was an over-helpful narrator, and the fix for that is not more description. It is to decide, sentence by sentence, what the reader is allowed to conclude alone.

The obvious objection is that the advice works: writers who follow it improve, and it is the rare maxim that survives contact with a classroom. Scenes are more vivid than summaries, the argument runs, and vividness is close enough to the goal that the conflation does no practical damage. There is a real point here, and it should be met rather than waved away. Vividness and inference do travel together often. Concrete detail tends to leave conclusions open and abstraction tends to close them, so a rule aimed at one will usually touch the other. But usually is the load-bearing word. The correlation is a fact about how prose is commonly written, not a law about how it must be read, and the exceptions are exactly the passages a writer most needs to get right.

Consider the scene that tells. A writer renders a confrontation in full, beat by beat, and every image is chosen to instruct: the sneer, the trembling hand, the meaning of each gesture spelled out by the arrangement. Nothing is summarised and nothing is left to the reader. Now consider the summary that shows. A single sentence reports that a man returned to a town after twenty years, and the reader supplies the dread, the hope, the entire withheld interior, because the text refused to. Were the maxim true to its own logic, the first passage should succeed and the second should fail. The reverse is often the case, and the reason is that the reader’s inference, not the image, was the thing at stake all along.

The maxim’s error, then, is a filing error, and it is worth correcting precisely because the maxim is otherwise good. What it should say is not show, don’t tell but let the reader conclude. That rule covers the circling note, and it also covers the revision the note is meant to provoke; it names the reader instead of the surface, and it survives the cases that break the original. The practical difference is not small. A writer working under the corrected rule asks, of each sentence, who is being made to do the thinking. A writer working under the old one asks whether there is enough detail. The first is the question the workshop meant, and the one narrative theory has been answering for a century while the workshops kept repeating the shorthand.

None of this argues that detail is dispensable, or that summarised emotion is as good as dramatised action. It argues that the maxim has been shorthand for a claim about inference and has been taught as a claim about imagery, and that the two come apart exactly where the stakes are highest. Name the right thing and the advice gets sharper, not softer. Keep naming the wrong thing and the writer will keep adding light and weather to passages whose real problem was a narrator who would not trust the reader to finish the thought.