Canonical Endings

1. The Problem: Nonlinear Games and Sequels

For linear games with a single ending, the canonical ending is obvious — everyone experiences the same thing. But for nonlinear, choice-driven RPGs like Fallout, establishing what "actually happened" becomes a real challenge.

Tim recalls that during Fallout's development, the person writing the official strategy guide kept asking "what's the player supposed to do?" and the team had no answer — it depended entirely on the player's character and choices. If it was hard to write a walkthrough, imagine trying to establish a definitive ending for a sequel.

The core tension: if you're making a sequel, you need a jumping-off point. You need to know what the world looks like when the new game begins.

2. Method 1: Find the Commonalities

Used for: Fallout 2

Look at all possible endings and identify what they have in common. Only assume those shared elements are canon — ignore everything else.

For Fallout 2, the team identified three universal truths across all Fallout 1 endings:

• The player survived
• The Master and his army were destroyed
• The player was kicked out of the Vault

Everything else — what happened to individual townships, who lived and who died — was left ambiguous. They avoided referencing specific player choices and only referenced locations the player would have necessarily passed through.

One small exception: they assumed Tandi had survived, since they made her the head of the NCR in the sequel. But beyond that, they kept references minimal.

3. Method 2: Just Pick One

Used for: Arcanum 2 (planned), Temple of Elemental Evil sequel (planned)

Simply declare one ending as the canonical one and build the sequel from there.

For the planned Arcanum 2, the team decided to assume the player had defeated Kyan and learned about Franklin Payne the great Explorer — and built the sequel premise around that specific ending.

For Temple of Elemental Evil, Tim had grandiose plans to continue into the classic D&D module chain: Against the Giants (G1-3), Descent Into the Depths (D series), and Queen of the Demonweb Pits (Q series) — now collected as the GDQ mega-module. The sequel would have assumed the "good" ending: the player descended to the bottom of the temple, encountered Zuggtmoy, fought and banished her — not joining her or becoming the ultimate evil.

4. Method 3: Let the Player Choose

Used by: The Witcher 3

At the start of the sequel, ask the player what happened in the previous game. This can be done through dialogue questions or menu selections, letting the player define their own canonical ending.

Tim finds this approach "really cool" because it means each playthrough of the sequel can start from different conditions, adding replay value. The player essentially gets to see their end slides play out — but not until they buy the next game.

Tim never used this method himself but appreciates the elegance.

5. Method 4: Import the Save Game

Used by: Mass Effect series

Have the player load a save file from the previous game. The sequel reads all the tracking variables — the same data used for end slides, achievements, and NPC reactions — and constructs the starting state of the new game accordingly.

This is the most granular approach, reflecting exactly what the player did rather than a simplified summary. Tim calls it "a pretty cool way of doing it" but never implemented it himself.

6. Summary

Tim identifies four approaches to canonical endings, ordered roughly by complexity:

1. Commonalities only — assume only what every ending shares
2. Developer picks one — declare a single ending as canon
3. Player picks one — let the player select their ending at sequel start
4. Save import — read the player's actual save data from the previous game

Tim admits he only made one sequel (Fallout 2) in his career and chose the first method. He can't think of a fifth approach and considers these four to cover the full design space.

7. References

• Tim Cain. YouTube video. https://www.youtube.com/watch?v=V6rDgA5NXQc