The Science of Discovery Writing: Why Pantsing Works

Discovery writing is a documented cognitive process, not a lack of discipline. Writing researchers have studied it for more than thirty years, and David Galbraith's dual-process model treats unplanned drafting as one of two genuine engines for generating ideas. Which engine works for you appears to be a stable trait — not a stage you outgrow.

Discovery writing is drafting to find the story instead of executing a plan for one. If you write this way, you have probably absorbed the accusation embedded in a lot of craft advice: that discovery writing is what you do until you develop the discipline to outline. The research says otherwise. Psychologists have studied what they call discovery through writing for more than thirty years, and the leading account of it, David Galbraith's dual-process model, treats unplanned drafting as one of two genuine engines for generating ideas. Which engine works for a given writer appears to be a stable trait, not a stage of development. The science does not say pantsing is superior. It says something more useful: writing without an outline is a real cognitive process with a known shape, known strengths, and a known cost that comes due at revision.

What Is the Dual-Process Model of Writing?

The model, developed by Galbraith across three decades of experiments, proposes that new ideas arrive during writing through two different mechanisms.

The first he calls knowledge transforming. This is the one craft culture recognizes as thinking: you hold ideas in working memory, weigh them against your goals, arrange them, and then write down the result. Outlining is knowledge transforming in its purest form. Every beat sheet, every synopsis written in advance, every index card on a corkboard is a bet that this is where stories come from.

The second he calls knowledge constituting, and it has no equivalent in the outline-first picture of how writing works. In this mode, content is synthesized during the act of composing sentences, drawn from the implicit organization of everything you know, an organization you cannot inspect directly. The ideas do not exist before the sentence. The sentence is where they happen.

You have had the experience the model describes. You type a line of dialogue meant to move a character toward the door, and instead she says something that surprises you, and the scene ends somewhere that did not exist when you started the sentence. The outline-first picture has to call that a loss of control. Under the knowledge-constituting account, it is the system working. Your understanding of the story is implicit, distributed, and unavailable for inspection until text production pulls it into words. The draft is not a record of your thinking. The draft is the thinking.

The model's claim is not that one engine is better. It is that both exist, that they conflict, and that writers differ in which one produces discovery for them. That last part is where the research gets specific.

What Does the Research Actually Show About Discovery Writing?

Galbraith's foundational experiment, published in 1992, compared writers generating ideas under different conditions and found that new ideas emerged most when writing was not directed toward rhetorical goals. The psychology literature frames this as a test between the classical position, where discovery comes from planning toward an audience, and the romantic position, where it comes from spontaneous expression. The results, as Galbraith's own review of the program puts it, supported the romantic position. In the psychology of writing, romantic is a technical term.

The individual-differences work is more useful still. Using a personality measure called self-monitoring, which distinguishes people who regulate their behavior toward external expectations from people who act from internal disposition, Galbraith found the two groups discover through opposite routes. High self-monitors generated more new ideas when planning toward a goal. Low self-monitors generated more new ideas through spontaneous drafting. Plotting and pantsing, measured in a lab, sorting by trait.

The most rigorous test came in a 2018 study by Veerle Baaijen and Galbraith in Cognition and Instruction, which logged every keystroke as writers worked under outline-planning or minimal-planning conditions. The study was designed to test the assumption that discovery flows from strategic planning and produces better text. What it found was that writers' understanding of their subject developed through two independent routes: global revision of the text they had produced, and a particular kind of spontaneous sentence production. A companion study found that writers' beliefs about how writing works moderated whether outlining helped them at all.

Read plainly, the research record says: discovery through unplanned drafting is real, measurable, and for a stable subset of writers it is the more productive engine.

Doesn't the Research Say Outlining Beats Discovery Writing?

Some of it does. Ronald Kellogg's experiments in the late 1980s found that writers assigned to outline produced higher-quality texts than writers who drafted without preparation, a result replicated enough times to be considered reliable. The standard explanation is cognitive load: planning in advance frees working memory during drafting.

Three things sit alongside that record. The experiments tested undergraduates writing short expository essays, usually in around thirty minutes, a task about as far from a novel as a task can be while still involving prose. The later research found the benefit is moderated by disposition and belief, which means the group averages conceal writers being helped and writers being hindered. And the planning literature itself concedes that structured pre-planning may not serve writers who prefer to plan during writing.

So the honest summary is not that science vindicates pantsing. It is that science supports process-fit: neither engine is universally superior, which one works for you is partly a stable trait, and a writing culture that prescribes outlining to everyone is prescribing against the grain of a measurable fraction of writers. If you have spent years failing at outlines and concluding something is wrong with you, the more parsimonious explanation is that the instructions were written for the other kind of writer.

What Does the Science of Discovery Writing Mean for Novelists?

None of these studies involved fiction. No lab has run keystroke logging on ninety thousand words drafted over fourteen months. The extrapolation from a thirty-minute essay to a novel is exactly that, an extrapolation, and any discovery writer who cites this research should cite it at its actual size.

What transfers is the mechanism and the trait. What also transfers, less comfortably, is the model's account of what unplanned drafting produces. Knowledge-constituted text carries the writer's implicit organization. The structure is in there, because it came out of the structured contents of your head, but it is implicit: undocumented, unlabeled, invisible from inside the draft. The model is specific about how that changes. In the second phase, existing text gets reorganized and evaluated, and it is this reorganization that increases the writer's understanding of what they have said. Discovery finishes in revision. The first engine builds the thing; the second engine is how you find out what got built.

For a six-page essay, the second phase is a careful reread. For a novel, a reread will not surface an organization spread across three hundred pages, which is why the stall so many discovery writers hit is a comprehension problem rather than a craft one, whether it arrives at 40,000 words or after the draft is done. The working method the model points to is the one experienced discovery writers converge on anyway: build the outline from the completed draft, scene by scene, so the implicit structure becomes something you can read. It is also why discowriter derives that outline from the finished manuscript instead of asking for a plan up front; in the dual-process picture, the plan is the last thing a discovery writer produces, not the first.

The reframe this research offers is small and load-bearing. You were never writing without a plan. You were writing the plan in the only language the first engine speaks, which is scenes. The half of the work you owe is the same half everyone owes: making the structure explicit so you can judge it. You just do that half after the draft instead of before, with the advantage that your version describes a book that exists.

None of this makes the middle of a discovery draft feel less like open water. The research will not finish the book. But there is a difference between struggling inside a broken process and struggling inside a real one at its hard part, and thirty years of evidence says yours is the second kind.

---

Frequently Asked Questions

Is discovery writing a real writing method? Yes. Writing research has studied discovery through writing since the early 1990s, and the dual-process model identifies spontaneous drafting as one of two cognitive mechanisms that generate new ideas during writing. It is a documented process with experimental support, not an absence of method.

Is it bad to write a novel without an outline? No. Experimental research shows the benefits of outlining depend on the individual writer's disposition and beliefs, and that some writers reliably generate more ideas through unplanned drafting. What the research does suggest is that unplanned drafts carry implicit structure, so discovery writers owe their manuscripts a structural pass at revision, typically through a reverse outline.

Do pantsers write worse books than plotters? There is no evidence for that. Studies showing outlining improves quality tested short timed essays by students, not novels, and later research found those benefits are moderated by personality and beliefs about writing. No controlled study has compared plotted and pantsed novels.

What is the dual-process model of writing? It is David Galbraith's account of how ideas emerge during writing, developed across three decades of experiments. It proposes two mechanisms: knowledge transforming, where ideas are deliberately manipulated during planning and revision, and knowledge constituting, where new content is synthesized in the act of producing sentences. Different writers get more discovery from different mechanisms.

Should discovery writers learn to outline? Not before drafting; the evidence does not support prescribing pre-planning to writers whose ideas come through spontaneous production. After drafting is different. A reverse outline, built scene by scene from the finished manuscript, is how a discovery draft's implicit structure becomes visible enough to revise.

Why do I get my best ideas while I'm drafting? Because for some writers, drafting is where thinking happens. The knowledge-constituting process synthesizes content from the implicit organization of what you know, and that organization only becomes available as sentences. If you are a low self-monitor in the research's terms, spontaneous text production is your more generative engine.