Week 11: Write the bad ending first

Why ambiguous NPC dialogue functions as navigation, and why a tragic ending can outlast a happy one

10 May 2027

A figure who never states their goal plainly is still giving directions, just not the kind a quest log prints — and a genre built on withheld comfort has little reason to hand back an uncomplicated happy ending at the finish line either. Write the bad ending first, this week’s workshop insists, because a “true” ending that only ever compromises between good and bad hasn’t earned either. The full argument for both is in Weak guidance and the value of a “bad” ending.

The argument#

Sekiro sharpens Dark Souls’ ambiguity rather than softening it. None of its four endings — Immortal Severance, Purification, Return, Shura — is uncomplicated good, and Shura in particular reads as close to a worst-case outcome in tone (the protagonist consumed by the same battle-hunger the game spent hours warning against) without the game ever labelling it that way in-fiction. A player has to arrive at “this was the bad one” through inference, the same read this week’s workshop asks students to write toward rather than announce. Dark Souls’ kindled-versus-dark split works the same way from a different angle: kindling a dying flame for one more age of an already exhausted world, or letting it go out and something genuinely unknown begin — neither framed by the game as the reward path.

The failure case is a “true” ending unlocked by a completion checklist rather than earned by understanding: a structure where seeing everything the game has to offer is what the true ending requires, which turns tonal payoff into a scavenger-hunt reward instead of a shift the player’s own grasp of the fiction produces. A true ending earned by information — you finally understand what the flame actually costs, who actually benefits from kindling it — is a different, better kind of “true” than one earned by percentage completed, and it’s the version this course’s own true endings are graded against.

Outline#

• ambiguous dialogue as navigation: pointing at significance without instructing
• Sekiro’s four endings, none uncomplicated, none labelled — inference standing in for a difficulty rating
• Dark Souls’ kindled-versus-dark ending as a test case neither side of which is simply “good”
• the failure case: a “true” ending unlocked by a completion checklist rather than earned understanding
• why an ending’s restraint has to match eleven weeks of the game’s own restraint, or it reads as a late betrayal of the design
• this week’s workshop: writing three 50-word endings — good, bad, true — in item-description style

Into the session#

Bring the boss and world you’ve been building since Week 2 — or whatever original setting you’re using for the Final Project. In the session you’ll write three 50-word endings for that world in the same unreliable-testimony voice as your Week 2 item descriptions, none announcing its own category. If a reader can sort your three into good, bad and true without hesitation, the ambiguity hasn’t been earned yet.