A document has one shape: top to bottom. It is a line. You scroll, you search, you skim. For a memo, that shape is perfect. For a world, it is the wrong container — and the wrongness compounds the more you put in.
A world is not a line. It is a graph. Characters relate to characters; rules constrain events; threads cross seasons. The relationships between facts carry as much meaning as the facts themselves — and a document has nowhere to put a relationship except a sentence that says it, which nothing can check.
This is not an argument about aesthetics or about which app feels nicer to write in. It is an argument about whether the shape of your container can hold the shape of your information. A story bible is a graph wearing a document's clothes, and the seams show exactly where worlds tend to break.
A line can't hold a graph
When you flatten a graph into a line, you lose the edges. "Maren is Toval's sister" and "Toval betrayed the council" sit in two different paragraphs, possibly two different docs, with nothing connecting them. The reader reconstructs the relationship in their head every time — and every reader reconstructs it slightly differently.
Worse, the line forces an order on things that have no natural order. Is Maren filed under M, under Characters, under the Rebellion arc, or in the chapter where she first appears? Whatever you choose, the other three readers go looking in the wrong place, find nothing, and write the scene from memory instead of from the bible. A doc has exactly one outline, and a world has many; the single outline wins and the others quietly stop being maintained.
On a canvas, the edge is a real object. The relationship between two cards is a thing you can see, click, and reason about. Maren connects to Toval with a labeled "sister" edge; Toval connects to the council with a "betrayed" edge. The structure of the world stops being something each collaborator rebuilds from prose and becomes something everyone is literally looking at the same copy of.
The search box is not a solution
The usual defense of the document is search: who cares about shape, you can just find anything. But search answers the question you already know to ask. It will hand you the page about magic when you type "magic." It will never tell you that the page about magic and the scene on page ninety disagree, because you did not know to search for the contradiction — if you knew it was there, it would already be fixed.
Continuity errors are precisely the facts you are not thinking about. A search box, a Notion database, a wiki — these are storage, and storage is very good at returning what you ask for. None of them performs the operation that catches a broken world: reading everything at once and looking for the pairs of statements that cannot both be true. That operation needs structure, not a better search field.
Cards that know what they are
A paragraph does not know it describes a character. A card can. When a piece of your world knows it is a character, a rule, a location, or a thread, the tooling around it can do real work — map relationships, place events on a timeline, and check rules against the scenes that should obey them.
That typing is what makes a consistency pass possible at all. You cannot scan a wall of prose for contradictions with any precision; the machine has no idea which sentence is a load-bearing fact and which is atmosphere. You can scan a set of typed, connected cards, because the structure tells the engine what each fact is and what it is allowed to touch — that a world rule constrains events, that a character has a status that changes over time, that a thread has a state.
The timeline is the clearest case. As prose, history is a paragraph, and a paragraph hides the error where a character witnesses something before they were born. As cards on a real timeline, the same events sort themselves into order, and the impossible sequence is visible at a glance. Same facts, but one container makes the error obvious and the other buries it.
The wrong container gets worse as you fill it
Here is the part that catches people. A document feels fine at first. Ten facts in a doc are easy; you can hold them in your head, search works, the lack of edges costs you nothing because there are barely any edges yet. So the doc never feels like a mistake at the moment you choose it — it feels like the obvious, frictionless choice.
The penalty arrives later, and it scales against you. Every fact you add multiplies the relationships the line cannot hold and the pairs of facts that can quietly disagree, while the doc's ability to surface any of that stays flat at zero. The container you picked because it was easy at ten facts is the one actively working against you at five hundred — and by then migrating out of it feels like more work than living with the breaks. The time to choose the right shape is before the world is big enough to make the choice painful.
The bible that argues back
The payoff of putting the world on a canvas is not prettier notes. It is that the structure becomes legible to a machine that can hold the whole thing at once and tell you where it disagrees with itself — surfacing the specific pair of cards in tension, with both sides quoted and the conflict named, before you write a season against the break. The document could never do that. The canvas can.
Same content, different container — and the container is the difference between a record you have to trust on faith and a world you can actually stress-test. CanonBoard is built on exactly this premise: your world lives as connected, typed cards on an open canvas, and the engine reads the graph the document was always pretending to be.
CanonBoard scans your whole world and tells you where it disagrees with itself.
Start free