Shaping a map from its subject
A map takes its shape from the dependencies between the ideas it covers, not from a template of counts.
Which internotes on a map reference which decides the map's shape, and this guide is about choosing those references from the subject. What a map is, and how one is made, is on Maps.
#The tree from the dependencies between ideas
A map's Tree preset draws it as a tree from the references between its internotes: an internote sits under the ones that reference it (internote). The tree is therefore a claim about the subject — this idea is reached through that one. Derived from the course outline and from which ideas need which, the claim is true. A shape chosen first and filled in afterwards makes the claim by accident, and the reader follows it anyway.
Two failures come from choosing the shape first. Topics get placed by theme rather than by what they need: an actuarial course put graduation of mortality rates under the life table, because both concern mortality, though graduation needs the crude rates estimated from exposed-to-risk in another branch. And a subject that is a sequence gets laid out as parallel branches under the root: a statistics course whose spine runs through sampling distributions buried them several levels down, where the line the course follows can no longer be seen. Where the ideas form a chain, the map shows the chain.
A shape copied from another map inherits that map's subject. Two computing courses mapped from one shared template came out nearly identical. Being able to say in one line why this shape fits this subject is the sign that it was derived.
#Counts as warning signs, not targets
A number of children per internote, or of scenes per internote, held as a target bends the tree to meet it. Internotes appear that exist to make up the number: a bucket such as Research design gathering topics that do not depend on it, or a minor hub promoted to the root to fill out its children. None of them is a step the subject has.
A count is still worth reading. One that looks wrong is a reason to go back to the dependencies and check them; the dependencies decide the shape, and the count follows from it.
#The root's opening
The root's first scene is where a reader enters the whole map, so its opening is chosen for the subject: a question the course answers, a data set it learns to read, a failure its methods prevent, or a worked problem. A dated real event suits some subjects; as a default it gives every map the same entrance, and the subject's own way in is lost.
The root's other scenes answer to the subject in the same way. A scene written so that every child of the root has one — a course's sorting algorithms attached to the root by a single sentence — introduces nothing the subject needed at that point.
#A forward link as a misplaced topic
An inline !internote reference in an internote's opening scene that points to an internote the tree places later, or in another branch, says the reader needs that idea now; the tree says they meet it afterwards. It is the most reliable sign of a misplaced topic. A regression course that taught leverage before multiple regression patched the gap with exactly such a reference.
The tree is what is wrong, not the reference. Moving the topic so its prerequisite comes first fixes the order for every reader who follows the map; a reference forward only tells them the order is wrong.