Abstract
Problem: How do you distribute experience points in an RPG so that different player builds (combat, stealth, dialogue) all progress at roughly the same rate?
Approach: Tim Cain draws on his experience designing Fallout, Arcanum, and other RPGs to present a structured, engineering-style approach to XP distribution based on story acts, quest-only XP, and external tuning tables.
Findings: Giving XP exclusively through quests (not kills, skill use, or achievements) solves the build-parity problem by decoupling rewards from playstyle. An 80/20 split between main quests and side quests, combined with externalized XP tables, provides robust control over progression even when content gets cut.
Key insight: XP tells players what you want them to do — if you reward kills, stealth, and dialogue separately, you're telling them to do all of it. Quest-only XP aligns player goals with designer goals by rewarding completion, not method.
1. Plan XP Per Act, Not Per Hour
Tim argues against thinking about XP rate in terms of time played, because players of nonlinear RPGs play at wildly different speeds. Instead, plan XP per story act.
If your game has 10–15 acts, decide what level the player should be at the end of each one. Work backward from your level cap (or expected end-game level) to determine how much XP each act should grant. Fallout had 3 acts; Arcanum had 27. Most games fall somewhere in between.
1.1. Prerequisite: Have Your Story First
Tim emphasizes that by the time you're designing XP systems, your story should already be written. His recommended order: setting → story → system mechanics. XP design is a system mechanics problem.
2. Quest-Only XP
Tim's core recommendation: only give XP from quests. No XP for kills, skill use, stealth success, or other activities.
The reasoning is elegant. When a quest says "retrieve an item from a cave full of bandits," the player can:
- Kill all the bandits and take the item
- Sneak past the bandits
- Intimidate the bandits into handing it over
- Buy or trade for the item
The quest rewards completion regardless of method. This naturally solves the build-parity problem — a stealth character and a combat character earn the same XP for the same quest.
2.1. Why Not XP for Kills, Skills, or Achievements?
XP is a signal to players: "this is what I want you to do." If you give XP for kills, skill use, speech checks, stealing, and quest completion, you're telling players the optimal path is to exhaust every dialogue tree, complete every quest, steal everything, then kill the NPC — because that maximizes XP extraction.
Tim asks bluntly: is that really how you want people to play your game?
2.2. The Achievement Exception
Tim vacillates on achievement XP. Aspirational achievements like "complete 50 side quests in this zone" can work as "icing on the XP cake." But achievements like "fall 100 feet and survive" or "kill 50 ants" reintroduce the same problem — they tell players to chase arbitrary goals instead of playing naturally.
3. The 80/20 Split
Divide quests into main quests (required to finish the game) and side quests (optional). Then:
- 80% of each act's XP comes from main quests
- 20% comes from side quests
This means a player who only does main quests finishes the game at roughly 80% of the level cap — viable but challenging. Side quests provide a buffer for players who want to be stronger, without being mandatory.
3.1. Balance Side Quests Across Builds
Each act should have roughly equal numbers of side quests for each player build type: some combat-focused, some stealth-focused, some dialogue-focused. This ensures every build has access to that extra 20% if they want it.
4. External XP Tables
A critical implementation detail: never hardcode XP values in quests. Instead, quests should reference a level (e.g., "grant level-5 XP"), and an external table maps that level to an actual XP number.
This matters because content always gets cut — areas, NPCs, entire quest chains. When act 2 loses a bunch of quests late in development, you can bump up the XP values for level 3–5 quests in the table without touching anything else. Without external tables, you'd be editing individual quests across the entire game.
4.1. QA as a Tuning Tool
Tim recommends giving QA a specific goal: verify that the player is an adequate level at the start of each act to handle that act's content. This makes progression tuning a testable, measurable process rather than guesswork.
5. Why One XP Source Matters
The final argument for quest-only XP is purely practical: if you have one source of XP, you can control it. The moment you add kill XP, skill XP, and achievement XP, tuning becomes a nightmare. Multiple XP sources interact in unpredictable ways, and no amount of planning will prevent some builds from leveling faster than others.
Tim describes this as the closest he's found to an "engineering approach" to XP balance, as opposed to seat-of-the-pants playtesting and ad hoc adjustments.
6. References
- Tim Cain. YouTube video. https://www.youtube.com/watch?v=NyVZlvaAYX4