Sanderson's three laws of magic

Sanderson's laws are Brandon Sanderson's three guidelines for designing magic systems, published as essays on his own site and restated in his FAQ. Worth underlining up front, because Sanderson does: they are guidelines, not rules. They describe a strong default that a writer can depart from knowingly; they stop being useful the moment they are applied as law.

First Law: the reader's understanding sets the power's limit

An author's ability to solve conflict with magic, in a satisfying way, is proportional to how well the reader understands that magic. Only magic the reader understands well should resolve problems; magic that remains mysterious is better used to create problems.

The practical effect is a sorting decision you make per spell, per system: which parts do the reader see the workings of, and which stay unexplained? A well-explained ("hard") ability can carry a climax, because the reader can follow the move and its cost. A mysterious ability that suddenly resolves the climax reads as the author reaching into the box, and readers feel the cheat even when they cannot name it.

Second Law: limitations over powers

What makes a magic system interesting is not what it can do but what it cannot do: its flaws, its limits and, above all, its costs.

This is the law with the widest application outside fantasy, and it maps directly onto the systems that power progression fantasy and LitRPG: a power whose use is free is a plot decompressor. Every tension the story winds, unlimited power unwinds. Costs (fatigue, materials, time, moral debt) and limits (what the magic has never been able to do) are what let the same ability stay dramatically interesting for a hundred chapters.

A one-line test for your system: write down what it can do, then write down what it costs and what it refuses to do. If the second list is shorter, the system is not finished.

Third Law: expand before you add

Before adding something new, expand what already exists. A new power, creature or realm is cheap to introduce and expensive to believe: each addition ripples through the world's economy, culture, politics and history, and readers sense when those ripples are missing.

The discipline is to push one idea several layers deep rather than constantly widening. If the story has fire magic, ask what fire magic does to metallurgy, to winter cities, to funeral rites, to who is dangerous in the dark, before inventing a second magic. Worlds built this way feel discovered rather than stocked. The law is also, quietly, a continuity tool: fewer, deeper pieces means fewer pieces to contradict.

Applying the laws

1. For each ability, record: what it does, what it costs, what it cannot do. Three fields; the third is the one that prevents both deus ex machina and accidental rule-breaking.
2. Decide the reader's knowledge level per element. Deep workings for the load-bearing magic; mystery for the atmosphere.
3. Before adding anything new, spend one pass expanding something old. The new thing should be a consequence of the deepened old thing where possible.
4. Telegraph exceptions. If the story must break a stated rule, establish the exception's existence before it matters. An unexplained violation of a stated rule is not a twist; it is a continuity error with ambitions (see the seven continuity errors).

Why this matters at length

Magic rules are canon with teeth. A character limit stated in chapter two is a promise the story renews every chapter, and long serials pressure exactly these promises: late at night, at chapter eighty, the rule about teleportation is standing between the writer and an easy scene. The writers who survive that pressure are the ones whose rule is written down somewhere with its evidence, so breaking it becomes a decision instead of an accident. That is the same argument as the story bible, which is where your laws should live.

How Kanonara approaches this

World rules are records in Kanonara's story bible, and Story Doctor checks new chapters against them: if a chapter breaks a stated rule with no established exception, the alert names the rule and the evidence, and you decide whether it is a mistake, an intentional mystery, or a rule whose time is up. See the Story Doctor page.