Zero drafting

What zero drafting is

Zero drafting is a diagnostic procedure that happens to look like writing. You take a scene from a world you have already partly built, you write it at full speed with the established record open beside you, and you pay attention to every point where the scene asks the world a question the world cannot answer. Those points are the output. The scene itself is scaffolding, and it comes down when the work is done.

The reason it takes the form of a scene, rather than a checklist or a review pass, is that scenes are the resolution at which world logic actually fails. A contradiction is nearly impossible to detect at the altitude of an outline, because outlines are written in summary — and summary is precisely the register in which contradictions are invisible. "They argue about the lighthouse" is compatible with almost any set of facts. Two people in a room at a specific hour, with specific histories and a specific thing one of them is concealing, is compatible with very few.

So you force the resolution early, on purpose, and you treat the resulting breakage as the product. A zero draft that runs smoothly has usually failed. A zero draft that stalls four times in two pages has just paid for itself.

What zero drafting is not

The term has loose prior usage in writing circles, where a "zero draft" means roughly a messy first pass — the vomit draft, the discovery draft, the one you have given yourself permission to write badly. That usage grades the draft. It describes a first draft with the quality expectations lowered, and it is still on the path to the finished manuscript; you revise it into the thing you publish.

The definition on this page is not about quality. It is about purpose and disposal. A zero draft is not a worse first draft. It is not on the path to the manuscript at all — it runs alongside, produces findings rather than pages, and ends in the bin by design. Three specific distinctions are worth drawing out, because each one is a practice people confuse this with.

Why it exists — the logic problem

Plot holes are not a discipline failure. They are the predictable result of an asymmetry: a story world grows faster than any working memory can hold it, and prose returns no errors.

Every other field that manages complex interdependent state has some mechanism that says no. Code has a compiler and a test suite. Accounts have a trial balance. Engineering has tolerances. Fiction has nothing. A contradiction in a story world is syntactically perfect — a sentence establishing that Maren is nine and a sentence establishing that Maren is twenty are both grammatical, both readable, both survive every spellcheck, and both survive every read-through by anyone who is not specifically hunting for that one thing.

The situation is made worse by three compounding factors.

• The person best placed to catch it is the least able to. The writer knows the world too well to read the page. They read intention, not text, and the intention was always consistent.
• Revision does not fix it. Revision is largely a prose-quality operation — rhythm, cuts, clarity, sharpening. A contradiction between page 40 and page 310 is invisible at that altitude, because you never hold both pages in mind at the same moment. This is the direct reason plot holes survive rewrites, often many of them.
• The cost curve is brutal. A contradiction caught in prep costs an afternoon. The same contradiction caught in the room costs a story day. Caught after a script is locked, boarded, and out to an animation vendor, it costs a retake and a schedule.

Which leaves the actual problem in a sharp form: contradictions are cheapest to find at the moment of writing, and the moment of writing is exactly when you are least equipped to find them.

Zero drafting is a deliberate answer to that. It manufactures the moment of writing, early, under controlled conditions, with the record open — and then treats what breaks as the thing it came for.

The method, step by step

What follows is the complete practice. It requires no particular software. A printout of your notes and a blank document will run it, and running it that way once is the best possible test of whether it is worth building into your process.

1. 01

   Pick the scene most likely to break

   Not the opening. Choose the scene where the largest number of established elements have to be true at the same time — two characters with separate histories, a rule you invented early, a location you only ever sketched, and a date on the timeline, all loaded into the same room.

   If you cannot tell which scene that is, choose the one you have been avoiding. Avoidance is a reliable instrument. Writers put off the scene where they privately suspect the world does not add up, and they are almost always right.

2. 02

   Put the record where you can see it

   Physically visible, without navigating away: a second monitor, a printout, index cards across the desk, a notebook open to the right page. This step is not a convenience. The entire method depends on the cost of checking a fact being close to zero.

   The moment checking requires a search, you will stop checking and start guessing — and a scene drafted against your memory of canon is drafted against the exact faculty that caused the problem in the first place.

3. 03

   Write it at full scene resolution

   Real dialogue, real staging, real time of day, real distances. Speed over quality. No line editing, no going back to fix a sentence, no craft. You are not making prose. You are making demands on the world and watching which ones it cannot meet.

   Resolution is the whole point. An outline can say "Eli confronts Maren about the signal" and stay honest indefinitely. A scene has to decide where they are standing, what time it is, whether she has aged, what she calls him, and whether he already knows what she is. Summary is the register in which contradictions hide.

4. 04

   Log every question and keep moving

   This is the discipline the method lives or dies on. When you hit something the world has not decided, write it down in a separate place, drop a placeholder in the scene, and carry on. Do not stop to answer it. Do not go looking for it if it is not already in front of you.

   Keep two lists, not one:

   • Missing — things the world never decided. How old is she now. How far is the walk. What the thing is actually called.
   • Broken — things the world did decide, and this scene needs to be otherwise.

   The second list is the valuable one. Missing items are just work. Broken items are the contradictions you were looking for, caught at the cheapest possible moment.

5. 05

   Stop when the scene stops teaching you

   Usually two to four pages, or forty-five to ninety minutes. The signal to stop is internal and unmistakable: the moment you notice you are writing to finish the scene rather than to find out something, the diagnostic is over and you are just writing.

   Stop mid-sentence if that is where it happens. There is no version of this where finishing the scene earns you anything.

6. 06

   Resolve both lists, then delete the pages

   Take the Missing list and decide it — properly, written into the shared record where the next scene can see it. A decision that only exists in your head, or in the margin of a draft you are about to delete, did not happen.

   Take the Broken list and choose deliberately, item by item: either the scene is wrong, or the canon is wrong. Both are legitimate outcomes, and changing canon on purpose is a completely different act from contradicting it by accident. What is not legitimate is leaving it unresolved because it is uncomfortable.

   Then delete the pages. Actually delete them. This is not a flourish — it is load-bearing. The freedom to break your own world comes entirely from knowing the writing will not be graded. Keep the pages and you have quietly promoted the zero draft to a first draft, and from that moment you are protecting prose instead of stress-testing a world.

Where it fits in a production pipeline

Zero drafting is a prep-stage instrument. Its value is a function of how expensive a late-discovered contradiction is, which means it pays out very differently depending on what you are making.

Common failure modes

The method is simple enough that most failures are failures of discipline rather than understanding. These are the ones that recur.

• Keeping the pages. By a wide margin the most common. Someone writes a paragraph they like, and the zero draft is quietly promoted to a first draft. Everything downstream stops working, because the writer is now protecting prose and will unconsciously steer away from exactly the instabilities they came to find.
• Answering questions mid-scene. Stopping to decide a fact in the middle of a draft you are going to delete creates a second canon that lives only in the deleted document. You will remember deciding it. You will not remember where.
• Drafting from memory. Running the method without the record physically visible converts it into a confirmation exercise. It will feel productive, it will surface almost nothing, and it will actively reinforce the errors already in your head.
• Choosing the safe scene. A scene that collides with nothing validates cleanly and teaches you nothing. A clean zero draft is rarely a pass; it usually means the target was chosen badly.
• Treating the list as a backlog. Missing items need decisions written into the record, not tickets in a tracker. The list is not work to schedule. It is a set of questions the world is currently unable to answer, and it stays that way until someone writes the answers down where everyone can see them.
• Running it once. It is a repeatable check, not a phase you complete. The useful cadence is one per act, per break, or per major note.
• Running it too early. Zero drafting needs enough established canon to contradict. Before a record exists there is nothing to test against, and what you are doing is not zero drafting — it is just writing, which is fine, but it is a different activity with different rules.

How CanonBoard's Zero Draft surface supports the method

Everything above works on paper, and it is worth running that way at least once. But in practice the step that decides whether people keep doing it is the second one — the record visible without navigating away. That is the friction that quietly ends the habit. When checking a fact means leaving the document, opening another file, and searching it, the checking stops within about ten minutes, and once the checking stops the method is just writing.

Zero Draft is a drafting surface inside a CanonBoard project built for that specific problem. You write the scene in a focused rich-text editor, and your canon sits in a reference rail beside it in the same view — characters, locations, lore, world rules — one click from the sentence you are writing. It is free on every plan, and it involves no AI and no credits. Drafts can be saved and reopened, locked so they stay private on a public project, downloaded as PDF or Word, or moved into the project's Documents if you have decided a particular draft has stopped being a zero draft. Mostly, though, you delete them, which is the intended ending.

What makes the rail useful is what is behind it. The record it draws from is the project canvas: typed, connected cards for every element of the world, a relationship graph you can pull a single node out of to see what depends on it, and a timeline held in in-world time rather than episode order. That is the difference between a reference panel and a source of truth — the thing the scene is being tested against has itself been kept consistent.

Conflict Detection is a separate tool, on Core and above, and it is worth being precise about what it does. It reads your canvas, not your draft. Run it after you have resolved a zero draft's two lists back into the record, and it cross-references the whole world for contradictions you did not think to look for — age inconsistencies, broken world rules, a character in two places at once — and explains each one in plain English. It complements the manual pass rather than replacing it, because it can only check what has been written down, and the entire point of a zero draft is finding out what has not been.