How Strong Power Systems Are Designed: Five Decisions Behind Rules Readers Can Actually Track

The finished power system is what readers see. The useful design work happens earlier, when the creator decides what information will be stable, what will remain uncertain, and what kinds of choices the system should force.

Here are five decisions worth making before you add another rank, element, bloodline or spell list. They apply to many kinds of speculative systems, but they are craft tools, not universal rules. Mystery-driven stories may deliberately answer some of them only for the author.

Decision 1: choose the unit of action

What is the smallest meaningful thing a user does? “Uses magic” is too large. “Binds a spoken promise,” “moves heat between two touched objects,” “borrows a memory for ten minutes,” or “opens a route to a place previously marked” gives you something a scene can test.

Once the unit is clear, write three variations and three things it cannot do. This prevents the category name from doing all the creative work.

Mistake: defining ten schools before defining what any practitioner actually does on the page.

Better: define one repeatable verb, then see how training, culture and personality create variation around it.

Decision 2: decide where uncertainty lives

Readers do not need to know everything. They do need to know which uncertainty is intentional.

You can keep uncertainty in outcome, in cost, in interpretation, in hidden information, or in whether a character is telling the truth. What becomes frustrating is uncertainty everywhere at once: the audience does not know the rule, price, reliability, counter or evidence, so any outcome can be justified afterward.

Write a “known/unknown” box for every major mechanic. That box is especially helpful during revision because it exposes accidental revelations and accidental vagueness.

Decision 3: make the strongest ability create the hardest choice

The strongest ability should not merely have the biggest number. Ask what decision becomes painful when it is available.

Maybe the power can end a battle but publicly reveals a forbidden identity. Maybe it heals one person by consuming time needed to save others. Maybe it can compel truth but destroys diplomatic trust. Maybe it works perfectly only in a location the protagonist has sworn never to enter.

This follows the same craft logic behind Sanderson’s emphasis on limitations and weaknesses: constraints generate problem-solving.

Mistake: balancing a powerful ability only by saying it consumes “a lot of energy.”

Better: connect the price to a resource, relationship or consequence the current scene genuinely values.

Decision 4: choose how the world verifies claims

If power matters socially, people will develop ways to test, fake, certify, fear or regulate it. Decide what evidence exists.

Can ability leave a visible residue? Can an institution test capacity? Can witnesses distinguish one school from another? Are there forged credentials? Does the government recognize a classification that practitioners reject?

Verification creates plot beyond combat. It enables fraud, licensing, prestige, stigma, insurance, apprenticeship and court disputes. It also gives exposition a natural home: characters can learn through procedures, records and disagreement instead of lectures.

Decision 5: decide what growth cannot erase

Progression becomes dull when every advance removes the old constraints. Keep at least one class of problem that greater power does not solve.

A stronger telepath may still need consent to maintain trust. A faster cultivator may still face political obligations. A master healer may still be unable to reverse death. A deity may have immense force but remain bound by ritual jurisdiction.

This creates continuity across arcs. The character changes; the story’s deepest pressure remains recognizable.

Run a three-pass design review

After the five decisions, do three passes.

Pass one: scene test. Write a short conflict where using the strongest ability immediately would be a bad decision. If you cannot, the system may lack meaningful trade-offs.

Pass two: reader test. Give a reader the rules established in the first act and ask them to predict two plausible solutions to a later problem. You are not testing whether they guess the plot. You are testing whether the system supplies usable information.

Pass three: world test. Pick one mundane institution—school, bank, temple, shipping company, licensing office—and ask what changes if the power is real and sufficiently common. Follow only the consequences that survive questions of cost, frequency and reliability.

Keep a contradiction register

Long projects accumulate rule drift. Maintain a small table with rule, first appearance, later exceptions, current canonical wording, and scenes affected. When a draft introduces a clever new use, check whether it actually follows the rule or quietly rewrites it.

Sanderson’s third-law essay favors exploring existing material before adding more. A contradiction register helps because it turns “add another exception” into an explicit editorial decision.

The finished reader-facing effect

A well-designed system does not require readers to memorize a manual. It gives them just enough stable information to feel the pressure of a choice. They understand why a character hesitates, why a clever workaround is clever, and why an apparent shortcut would violate something already established.

That is the real target: rules readers can track when it matters, mystery where mystery serves the story, and consequences that survive beyond the moment an ability is used. \n## Add one decision about failure states\n\nBefore production, define what failure looks like. Does an ability simply not activate, partially activate, overshoot, choose a wrong target, consume the cost without producing the effect, or succeed while creating collateral consequences? Different failure states create different kinds of suspense.\n\nAvoid using random failure as a universal escape hatch. If a system “sometimes fails” whenever the plot needs uncertainty, readers cannot reason about it. Tie failure to observable conditions where possible: insufficient preparation, interference, degraded equipment, conflicting vows, broken concentration, inaccurate information, or a target outside the established domain.\n\nThen decide what characters can learn from failure. A visible residue can turn a failed attempt into evidence. A consumed resource can force the next scene into a different plan. A public misfire can change reputation. Failure should usually leave the story in a new state rather than resetting the board.\n\n## Build a small consistency harness for long projects\n\nFor a novel series, game or collaborative IP, keep a compact test set for every major mechanic. Include one ordinary use, one edge case, one prohibited use, one known counter and one costly but legal use. When a new draft changes the rule, run those five examples again.\n\nThis is the narrative equivalent of a regression test. It prevents a clever scene in book four from silently making an important book-one sacrifice unnecessary. The harness can be plain text; what matters is that it records the consequence of the rule, not merely its wording.\n\nFor team work, add an owner and revision date. If a writer proposes a change, record whether it is a clarification, a newly discovered in-world fact, or an actual retcon. Those categories produce different reader expectations and should not be mixed.\n A final editorial trick is to highlight every sentence in a draft that contains “always,” “never,” “only,” or “cannot.” Compare those absolute claims with the current canon sheet. Absolutes are useful because they create trust, but accidental absolutes are one of the fastest ways to create contradictions later in a series.