Design Pillars
"To question the habitual. But that's just it, we're habituated to it. We don't question it, it doesn't question us, it doesn't seem to pose a problem." — Perec, more or less, from memory
Four pillars. They're pinned for a reason: when a feature and a pillar disagree, the feature loses. I've killed a lot of good, clever, expensive ideas by holding them up against this list and watching them go grey. That's what the list is for. It's not inspiration, it's a bouncer.
the whole feeling in one JPG — moss, oxblood, bone, wet stone, and a light that isn't going to save anyone
1. Agency moves from the world to the witness
The player cannot change what happens to the Bissett Shore. The player changes what Wren notices, records, and carries.
Consequence for design: there is no "world state" fork. No choice you make turns the tide, saves the village, or relights a dead lamp. Every meaningful choice branches the almanac, not the coast. If I catch myself building a decision that alters the outcome, I've built the wrong game. The verbs are look, kneel, record, turn the page — never fix, defend, win.
2. The sea rises regardless
The outcome is a constant, not a variable. See The sea rises regardless.
Consequence for design: I do not build a "good ending" toggle. I spent months trying to earn one and it was hollow every time — that whole scar is written up in The Ending Problem. The rising sea is set in the engine like gravity. You don't get points for defying gravity; you get a life inside it. Erosion runs on a clock the player never touches.
3. The almanac is the save file
Game state = what Wren has noticed. Not a pile of flags. The book she keeps IS the save file. See The Almanac Mechanic.
Consequence for design: no hidden variables. If it isn't written in the almanac, it didn't happen — not to the game, not to Wren, not to me. This keeps me honest: the player can always open the book and see the entire truth of their playthrough, because the book is literally all there is. Save/load is just serializing the pages.
4. Attention as love
Looking closely at a thing is how you keep it. That's the moral engine under all of this. See Attention as love.
Consequence for design: the game rewards noticing, not solving. Pacing, UI, and feedback all privilege the act of attention — a pool you kneel at yields more than a pool you walk past, and the "more" is never a resource, it's a sentence. If a system makes the player optimize instead of look, it fights this pillar and it goes.
None of these are original. They're just the truest four I've found, and I keep them here so future-me — who will absolutely try to add a crafting system at 2am in August — has to walk past all four to do it.
A game about a disappearing place should not, itself, let you save the place. It should teach you how to lose it with your eyes open.