A plot hole that survives six drafts was not missed six times through carelessness. It survived because nothing in the revision process was ever capable of catching it, and because the one person guaranteed to be looking at the page is the one person structurally unable to see it.
This is the problem the zero drafting method exists to answer. Before the method makes sense, the problem is worth stating precisely — because the usual diagnosis, that someone should have been paying more attention, is both wrong and the reason it keeps happening.
Fiction has no compiler
Every other discipline that manages large amounts of interdependent state has some mechanism that says no. Code has a compiler and a test suite. Accounting has a trial balance that will not reconcile. Engineering has tolerances and a simulation that fails. In each case the system itself refuses the contradiction, loudly, at the moment it is introduced.
Fiction has nothing of the kind. 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 idiomatic, and both survive every spellcheck, every grammar tool, and every read-through by anyone who is not specifically hunting for that exact pair. The error is invisible to every automatic check that exists because it is not an error in the writing. It is an error in the relationship between two pieces of writing that are three hundred pages apart.
That is the base condition. Everything else compounds it.
The writer is the worst available reader
The person best positioned to catch a continuity error — the one who has read the material most often and knows the world most completely — is the person least able to.
The reason is that a writer rereading their own work does not read the text. They read the intention, and the intention was always consistent. When you reach a sentence you wrote, you do not process it fresh; you recognise it and load the meaning you had in mind when you wrote it. The version of the world in your head is coherent, so the version you experience on the page is coherent too.
This is not a fixable attention problem. It is closer to a perceptual one, and it gets stronger with familiarity — which means it is strongest at exactly the point in the process where you are doing the most rereading.
Revision is checking something else
Here is the part that surprises people. Revision does not catch continuity errors because revision is not a continuity operation. It is largely a prose-quality operation: rhythm, cuts, clarity, sharpening a line, killing a redundancy, making a scene land harder.
Those tasks operate on a local window — the paragraph, the page, the scene. A contradiction between page 40 and page 310 is structurally invisible in that window, because you never hold both pages in mind at the same moment. You could do twenty passes at that altitude and the pair would never be co-present in your attention once.
So the error does not survive rewrites despite the rewrites. It survives because the rewrites were checking a different property, and no number of additional passes at the wrong altitude adds up to a pass at the right one.
The world outgrows working memory early
The number of facts in a story world grows roughly linearly as you write, but the number of pairs of facts that could contradict each other grows as the square. A world with 40 established facts has 780 possible pairs. At 200 facts it is nearly 20,000.
Human working memory does not move. So there is a crossover point, and it arrives far earlier than most people expect — comfortably inside a first season, or the second act of a feature — after which no one can hold the whole thing. Past that line, consistency stops being something you can maintain by being careful and becomes something that requires an external record.
This is also why the problem is worse for the most experienced people on a project. They are the ones carrying the most in their heads, which means they are the ones most likely to trust their recall over a lookup.
The cost curve is the reason it matters
If continuity errors stayed cheap, none of this would be urgent. They do not.
- **In prep**, a contradiction costs an afternoon of deciding something.
- **In the room**, it costs a story day and a rework of whatever was built on top of it.
- **After a lock**, it costs a rewrite, and every department downstream absorbs the change.
- **In animation, after boards have shipped**, it costs a retake, a schedule, and an invoice.
- **After release**, it is permanent, public, and becomes the thing a section of your audience knows your world for.
Which produces the sharp form of the problem: contradictions are cheapest to find at the moment of writing, and the moment of writing is when you are least equipped to find them.
What actually works
Two things, and they operate on different halves of the problem.
The first is an external record that is genuinely authoritative — one place where the world's facts live, that everyone checks rather than remembers. This is unglamorous and it is the whole game. A record that is out of date, or that competes with three other documents, is worse than none, because it produces confident wrong answers instead of honest uncertainty.
The second is a way to surface the questions the record has not answered, which a record cannot do for itself. That is the specific job of a zero draft: force a scene at full resolution against the record, log every question it cannot answer and every fact it contradicts, resolve both lists back into the record, and throw the scene away. It converts an invisible, accumulating liability into a short list of decisions you can actually make. That procedure is set out in the complete zero drafting method, and the practice it is most often confused with is covered in zero drafting versus outlining.
Neither of these is a matter of trying harder. That is the point. The problem is structural, so the answer has to be structural too.
Frequently asked questions
- Why do plot holes survive multiple rewrites?
- Because revision is largely a prose-quality operation working on a local window — a paragraph, a page, a scene. A contradiction between page 40 and page 310 is never co-present in that window, so additional passes at the same altitude cannot catch it no matter how many you do.
- Why can't the writer catch their own continuity errors?
- A writer rereading their own work reads the intention rather than the text — recognising a sentence and loading the meaning they had when they wrote it. Since the version of the world in their head is consistent, the page reads as consistent. The effect strengthens with familiarity.
- At what point does a story world outgrow memory?
- Earlier than most people expect. Facts grow linearly but the pairs that can contradict each other grow as the square — 40 facts produce 780 pairs, 200 facts produce nearly 20,000. The crossover typically arrives inside a first season or the second act of a feature.
CanonBoard scans your whole world and tells you where it disagrees with itself.
Start free