Abstract
Problem: How should a designer approach creating a skill system for an RPG — choosing which skills to include, setting their ranges, distinguishing them from perks/traits, and balancing them across the game?
Approach: Tim Cain draws on his experience designing Fallout and The Outer Worlds to outline a top-down methodology: start from setting and story, derive gameplay paths, then build skills to support those paths.
Findings: Skill range (0–5, 0–10, 0–20, or 0–100) is one of the most consequential design decisions, as it dictates point distribution, item bonus viability, and stacking rules. Difficulty abstraction (easy/hard/near-impossible labels referencing a tunable table) makes late-stage balancing dramatically easier.
Key insight: Skills should never be designed in a vacuum — they exist to support the gameplay paths your setting and story demand. If you can't point to content that uses a skill, that skill shouldn't exist.
Skills Follow Setting and Story
Tim emphasizes that mechanics — and skills in particular — must be derived from the setting and story, never the other way around. He always works in the order: setting → story → mechanics.
In Fallout, he defined three core gameplay paths: combat, stealth, and dialogue. The Outer Worlds added a fourth: leadership. Skills were then created to support these paths. The skills weren't invented first and then shoehorned into the game — the desired gameplay experiences came first, and skills were built to enable them.
Only Build What You'll Use
If you're making a haunted house game with incorporeal ghosts, you probably don't need combat skills — you'd want perception, dialogue, and maybe psychic skills instead. Tim's rule is blunt: if you aren't going to populate your game with traps, don't add a traps skill. If there aren't many NPCs to talk to, skip dialogue skills.
He's seen games that include speech skills and then barely use them, which raises the obvious question: why did you bother?
Skills vs. Perks, Traits, and Backgrounds
Tim draws a clear distinction:
- Skills are things that improve over play — you use them, you invest in them, they grow gradually
- Perks, traits, and backgrounds are either preconditions (things you were born with) or very coarse improvements with only a few levels (3–5)
He describes perks as "really chunky skills" — if something only has three meaningful levels, it's better modeled as a perk than a skill.
Skill Ranges: The Critical Decision
Tim considers the skill range one of the most important decisions in RPG design. He walks through four options:
0 to 5
A very narrow range where each point represents a 20% jump. At this scale, item bonuses become absurd — "silken gloves of pickpocketing" with a +2 bonus would cover almost half the entire skill range, making point investment pointless. Tim's advice: if you're considering this range, just use perks instead.
0 to 10
Slightly better, but bonuses are still problematic. A +2 item bonus equals one or two full levels of investment. Tim would limit increases to leveling only (perhaps two points per level, spent in different skills) and likely skip item bonuses entirely at this range.
0 to 20
Tim's comfort zone begins here. At this range, you can introduce occasional bonuses from items, spell buffs, or drugs, but he'd cap them at +1 or +2 and make them non-stacking — only the highest bonus from any source applies.
0 to 100
The most flexible option. You can distribute many points per level, allow bonuses from multiple sources (armor, spells, drugs), and implement stacking rules. Tim suggests letting the highest bonus from each category stack — so the best armor bonus + the best spell bonus + the best drug bonus all combine. This creates interesting itemization without breaking the system.
Tim's personal preference across his career has leaned toward 0–10, 0–20, or 0–100.
Balancing Skills Across the Game
For balancing skill checks throughout the game, Tim describes two techniques:
Map-Based Difficulty Scaling
Each map gets an estimated difficulty based on creature level, expected player level, or story act. Skill checks on that map are calibrated accordingly. Dialogue checks, for example, test against progressively higher speech ranks as the player advances deeper into the game.
Abstract Difficulty Labels
Rather than hardcoding specific skill values into every check, Tim recommends defining abstract difficulty tiers: easy, hard, very hard, and near impossible. These labels reference a central table that maps to actual values.
For a 0–20 range, the table might look like:
- Easy: 3
- Hard: 12
- Very Hard: 16
- Near Impossible: 19
This applies to everything — locks, hackable computers, pickpocket difficulty (often based on NPC level). The power of this approach is that you can adjust difficulty globally by tweaking the table, without touching individual content. If all "hard" checks feel too easy, bump them from 12 to 13 in one place and every hard check in the game updates.
References
- Tim Cain. YouTube video. https://www.youtube.com/watch?v=SL_aTjKsxok