Power Systems That Hold Up Under Plot Pressure: A Framework for Rules, Costs, Limits, and Consequences

A fictional power system earns its keep when it makes scenes harder to write in a good way. It should narrow choices, expose character priorities, create second-order consequences, and let readers make informed predictions without turning the story into a spreadsheet tutorial. If the system only supplies new abilities whenever the plot stalls, it may look elaborate while doing very little structural work.

This framework treats “power system” broadly: magic, cultivation, psychic abilities, divine gifts, engineered enhancements, supernatural contracts, or any other repeatable capacity that changes what characters can do. The goal is not to make every story use hard rules. The goal is to decide which parts need to be legible so that the story can create fair tension.

Brandon Sanderson’s essays on magic systems are useful craft heuristics here, especially his emphasis on limitations, consequences, and expanding existing material before adding more. They are not universal laws of fiction, and many excellent stories deliberately preserve mystery. Use them as diagnostic prompts, not commandments.

First decide what the reader is allowed to predict

Before writing rules, choose the prediction level. Can readers predict exact outcomes, broad trade-offs, or only the kind of danger involved?

A detective-like battle story may benefit from precise constraints. A mythic horror story may need uncertainty. A political fantasy may care less about combat mechanics and more about who is legally permitted to use power, who pays for it, and how institutions react.

Write one sentence:

“After seeing the system three times, a careful reader should be able to predict ______, but should still be uncertain about ______.”

That sentence is more useful than a twenty-page encyclopedia because it defines the contract between information and suspense.

Build rules from verbs, not lore nouns

Creators often begin with categories: seven schools, twelve elements, nine ranks. Categories are easy to name but do not automatically generate scenes. Start with verbs instead.

What can a user sense, move, bind, transform, store, borrow, erase, exchange, summon, copy, or refuse? What must happen before the verb works? What changes after it works?

A compact rule card can look like this:

Field	Question
Action	What can the user actually do?
Target	What can and cannot be affected?
Trigger	What condition starts the effect?
Cost	What scarce resource, risk or obligation is consumed?
Limit	What blocks, caps or distorts the effect?
Evidence	What can another character observe?
Consequence	What remains changed after the scene ends?

If two abilities have the same answers except for color and name, they may not be two meaningful mechanics.

Costs should create decisions, not just arithmetic

“Uses mana” is a cost description, but it may not create drama if mana refills between chapters whenever convenient. A useful cost changes a decision in the current scene or a later one.

Consider five types:

Capacity cost: limited energy, charges, materials or time.
Opportunity cost: using one ability prevents another action or consumes a rare opening.
Exposure cost: the power reveals location, identity, allegiance or capability.
Relationship cost: another person must consent, cooperate, trust, owe or resent the user.
Institutional cost: law, guild rules, debt, licensing, surveillance or political consequences follow the act.

You do not need all five. One deeply integrated cost is usually stronger than six decorative ones.

A useful test is to ask: “If the hero could use this power for free in the climax, which earlier dilemma would disappear?” If the answer is “none,” the cost may not be doing structural work.

Limits are where tactics come from

A limit is different from a cost. A cost says “you can do it, but pay.” A limit says “you cannot do it under this condition,” or “the result stops here.”

Sanderson’s second law argues, in essence, that limitations can be more interesting than powers. The practical reason is simple: limits force characters to route around problems. A teleport ability becomes tactically interesting when line of sight, carried mass, destination memory, or interference matters. A healing ability changes plot structure when it cannot restore certain damage, requires time, transfers harm, or depends on consent.

List each important power’s three strongest “no” statements. If you cannot write them without inventing exceptions, the system may still be too elastic.

Put information into the world through evidence

Rules feel artificial when the narrator stops the story to lecture. Instead, decide how characters know the system.

Some rules may be measured by institutions. Others are folk knowledge, religious doctrine, battlefield experience, family secrets, or contested theory. These sources can disagree. A state academy may classify an ability differently from the people who have lived with it for generations.

That disagreement is useful because it separates the rule itself from what characters believe about the rule. It also creates room for error without making the author’s system arbitrary.

Make an evidence ladder:

directly repeatable observation;
reliable institutional or expert record;
experienced testimony;
common teaching or rumor;
ideological claim.

When a later reveal contradicts an earlier statement, readers can tell whether they learned a new rule or merely discovered that a source was wrong.

Design counters as interactions, not colored keys

A counter system becomes thin when fire always beats ice because the chart says so. Stronger counters arise from verbs, conditions and goals.

Suppose one character can create a perfect barrier but cannot move while maintaining it. The counter does not need to be “barrier-breaking magic.” It can be relocation, hostage pressure, environmental change, time, divided attention, or a goal that cannot be achieved while stationary.

For every major ability, write three kinds of counter:

direct: interferes with the mechanism;
situational: changes the environment or timing;
strategic: makes using the power the wrong choice.

This keeps conflict from collapsing into power-level comparison.

Track consequences outside combat

Power changes economies, architecture, crime, medicine, labor, religion, education, courtship, warfare and bureaucracy. But you do not need to redesign civilization for every trick. Follow the powers that are common, reliable and economically meaningful.

Ask four filters: How many people can use it? How often? At what cost? With what reliability? A rare miracle may reshape faith more than logistics. Cheap, reliable long-distance communication may reshape administration even if it never wins a duel.

Choose two institutions that cannot ignore the system and show their adaptations. That gives the world depth without writing an encyclopedia.

Create an exception budget

Every system eventually needs edge cases. The danger is not exceptions themselves; it is exceptions that arrive only to rescue the plot.

Keep an exception ledger with four fields: rule affected, exception, earlier evidence, price of using it. If an exception has no prior evidence and no meaningful price, delay it, foreshadow it, or remove it.

A reveal should ideally make earlier scenes more intelligible. Readers can say, “That explains why this happened,” rather than, “Apparently anything can happen now.”

Progression should deepen choices, not just numbers

Rank systems often default to more speed, more force, larger area. That can work for spectacle, but repeated numerical growth eventually flattens tension.

Alternative progression includes precision, efficiency, simultaneous control, information, reliability, social permission, reduced setup time, better combinations, or the ability to accept a different category of risk.

When a character advances, write one old problem that becomes easy and one new problem that becomes visible. Growth should move the story into a new decision space rather than simply invalidate the previous arc.

Stress-test the system with five scenes

Before finalizing the lore document, run the system through five scene types:

a duel where both sides know each other’s capabilities;
a rescue where raw destructive power is dangerous;
a negotiation where revealing an ability changes leverage;
an ordinary workday where the power has practical value;
a failure where the user follows the rules correctly and still loses.

If every scene resolves through the same strongest technique, the system needs more meaningful conditions. If the rules only matter in duels, the world integration may be too shallow.

A one-page power-system spec

You should be able to summarize the working core on one page: reader prediction contract; five most important verbs; three hard limits; two meaningful costs; visible evidence; three counter types; two institutions affected; progression axis; exception policy.

Keep the larger lore archive if you enjoy it, but let this one-page spec control drafting. When a new power appears, ask whether it deepens an existing dimension or merely adds another noun.

That follows the most useful spirit of Sanderson’s third-law essay: before expanding outward, see how much story already exists inside what you have built. \n## Make information a resource too\n\nOne of the easiest ways to keep a power system from turning into arithmetic is to let knowledge matter. Two characters with identical capacity can make very different decisions if one understands an interaction the other has never seen. This lets experience, research, espionage and teaching matter without inventing a higher numerical tier.\n\nTrack three kinds of knowledge separately: public rules that ordinary users are expected to know, specialized techniques that require training or access, and contested claims that may be incomplete or politically distorted. A secret should not merely be “there is a stronger spell.” It can be a forgotten limitation, a non-obvious combination, a false assumption in the official curriculum, or evidence that a supposedly universal rule has a narrower domain.\n\nInformation also creates fair reveals. If the decisive trick in the finale depends on a property established earlier, readers can recognize the chain even if they did not predict the exact use. That is more satisfying than a new mechanic arriving at the same moment it solves the problem.\n