Progression systems are the engine of most role-playing games, and the ways they break are consistent enough to be catalogued.
The core tension
Progression must make the player feel more capable while keeping the game challenging.
Which means enemy capability must scale alongside player capability, and how that scaling is handled determines whether progression feels meaningful.
If enemies scale exactly with the player, nothing changes and progression is illusory.
If they do not scale at all, the game becomes trivial in areas the player has outgrown.
Every system sits somewhere between, and the placement is the central design decision.
Level scaling
Enemies adjusting to player level, which keeps content relevant and removes the sense of growth.
The common criticism is that a bandit becoming stronger as the player does breaks the world's internal logic.
Partial scaling — with floors and ceilings per area — is the standard compromise, allowing areas to be outgrown while remaining playable.
Multiplicative stacking
The most common source of broken builds.
When two effects each multiply damage, combining them multiplies both, which grows far faster than the designer's mental model of adding bonuses.
Which is why systems with many percentage modifiers produce combinations doing orders of magnitude more damage than intended.
Additive stacking within categories, with multiplication only between categories, bounds this substantially and is the standard fix.
Dominant strategies
When one build outperforms across all situations, the decision space collapses.
Which is generally caused by one resource being universally valuable — a stat affecting everything, or an ability with no situational weakness.
Balance work involves finding these and introducing situations where they are less effective, rather than simply reducing numbers.
The grinding question
Where progression rate meets content volume.
If progression requires more play than the content supports, players repeat activities purely for advancement.
Which is a design failure in single-player games and a deliberate retention mechanism in games monetising time.
The distinction between engaging repetition and grinding is whether the activity remains interesting, and that is a design question rather than a duration question.
Respec and commitment
Whether choices can be undone.
Permanent choices carry weight and punish experimentation, particularly where the consequences are not visible in advance.
Free reallocation removes consequence and allows optimising for each situation, which flattens character identity.
Most systems have moved toward allowing reallocation at some cost, which preserves some weight while permitting recovery from a bad decision.
Number inflation
Where damage and health figures grow to unwieldy magnitudes over a game's lifetime.
Which happens when each expansion must exceed the last, and it eventually makes numbers meaningless.
Several long-running games have performed deliberate reductions, compressing all values proportionally, which changes nothing mechanically and restores legibility.
Testing
Balance cannot be established by playtesting alone, since testers do not explore the space as thoroughly as a player population.
Which is why systems are analysed mathematically and simulated, and why a game's community finds unintended combinations within days of release regardless.
Designing so that unintended combinations are interesting rather than game-ending is the more achievable goal, and several designers have said so openly.
Economy design
Currency and item flow, which is a separate balance problem from combat.
Sources add currency to the world, sinks remove it, and if sources exceed sinks over time, inflation follows.
Which is a real problem in games with persistent economies, where accumulated wealth makes prices meaningless for established players and prohibitive for new ones.
Sinks that scale with wealth — repair costs, consumables, cosmetic purchases — are the standard mechanism, and calibrating them is ongoing rather than one-off.
Reward pacing
How frequently meaningful upgrades arrive.
Too frequent and each means little. Too sparse and progress feels absent.
Which is why systems layer several reward tracks at different frequencies — frequent small gains, occasional significant ones, rare transformative ones.
The variable-interval structure of the rare rewards is what produces the compulsion players describe, and designers are aware of the mechanism.
Class identity
Balance pressure tends toward homogenisation, since giving every class an answer to every problem equalises outcomes.
Which removes the distinctiveness that made the classes interesting, and resisting it is a deliberate choice with a cost in balance complaints.
Player-versus-player interaction
Where balance problems multiply.
Abilities tuned against computer opponents behave differently against people, and a single ability set serving both creates constant tension.
Which is why many games apply separate tuning in competitive contexts, adjusting values only there.
That solves the balance problem and creates a learning problem, since abilities behave differently in different contexts.
Patch cadence
Frequent changes keep the game moving and prevent players from settling into mastery.
Infrequent changes allow depth to develop and let dominant strategies persist.
Competitive scenes generally prefer stability approaching tournaments, which conflicts with the wider population's preference for change.
Community datamining
Players extract exact formulas from game files within days of release, which removes any ambiguity the designers intended.
Which changes the experience considerably, since optimal play becomes calculable rather than discoverable.