The temperature the plants feel: wiring a century of cold into the ground, not the sky
The Long Watch quietly tracks two temperatures, and the difference is easy to miss. One is the air — the value that paints frost on the screen and reads out on a debug panel. The other is the ground: the warmth of the soil, and the only one the living world actually consults. Setting out to slowly cool a world across a century, we walked straight into that split.
The job was a long-arc, late-game challenge: over a very long span, a maturing world should slowly cool — or warm — and that drift should genuinely change how the place lives, not merely how it looks. What a cooling century feels like to tend is told on its own. This is the engineering underneath it: where the cold actually goes, and how we proved it bites.
Two thermometers, one of them inert
A world holds a temperature for the air and a separate one for the ground, and until this work we hadn’t fully reckoned with how differently they earn their keep. The air temperature is the one a player perceives: it drives the frosty shimmer on the screen and surfaces on a debug readout. It is, in the most literal sense, for looking at — nothing that lives in the world consumes it.
The ground temperature is the working one. Three ecological processes read it, and only it: how fast a plant grows, how fast a fallen body decomposes, and how fast dead plant litter breaks back down into the soil. Growth and both kinds of decay all take their pace from the warmth of the earth at the exact spot they happen. The plants, in other words, feel a temperature the player never sees — the fuller account of why everything alive reads the ground and never the sky is told separately.
Where the drift had to land
That split decided the whole design. A drift wired into the air temperature — the thing on screen — would have been ecologically inert: the frost would deepen, the debug number would slide, and not one plant, root, or rotting leaf would notice. Weather as decoration. So the cold had to be delivered into the ground temperature instead, and delivered the same way everywhere that temperature is read — added identically to the growth step, the decomposition step, and the litter-breakdown step alike, so the three can never fall out of agreement about how cold the world has become.
The world’s temperature is a signed scale rather than degrees. It runs from the coldest biomes at about −0.55, through a neutral meadow at 0.0, up to the warmest desert near +0.55 — the same scale behind how the land reads that warmth to decide what country it is. The drift is simply a small negative delta shifted along that same scale — the units the ground already spoke — so a cooling century nudges every reading a little further toward the cold end.
A cooling you can only see changes nothing. To bite, the cold has to land in the ground the plants actually read.
A drift with no dice in it
The drift itself is deliberately dull, and that is the point. It carries no randomness at all: it is a pure function of how much in-game time has passed. It starts at zero, ramps gently upward over roughly a century of in-game years, and then holds at a bounded maximum rather than running away — a thousand-year world sits exactly where a hundred-year one arrived.
Because it’s a plain function of elapsed time, the same world always drifts the same way, on any machine — the reproducibility the whole simulation is built to guarantee (a discipline we’ve written about at length). And it has a property we leaned on hard: set the drift amount to zero and the function returns zero, and the world behaves exactly as it always did — not approximately, but identically.
Shipped at zero
Which is how it shipped. The entire mechanism is wired in, but its strength is set to zero, so a world you play today drifts by nothing you could measure. That’s on purpose, and it’s a call we’ve grown comfortable making: build the structure, prove it, and leave the felt intensity — how hard a cooling century should actually hit — as a separate, deliberate tuning decision for later. Structure first; effect when we can judge it by feel.
An instrument, not a guess
Leaving the intensity for later only works if the later decision has something honest to lean on. So the same work built the gauge. The first piece is a measurement instrument that sweeps a series of cooling amounts through the real plant ecology — not a stand-in, the actual growth path — and prints the resulting growth as a dose-response curve. Across four increasingly cold settings the growth stepped steadily down: a smooth slope, not a cliff, which is exactly what you want from a knob you intend to turn by hand. At the deepest cooling we tried, growth landed about fifteen percent below an uncooled world.
The second piece is a demonstration built to rule out coincidence. It grows the same seeded world twice — everything identical but the drift, one run cooled and one left alone — and compares them. On that seed the cooled world’s meadow grass came in measurably lower than its untouched twin: roughly 130 against 156 in the test’s own growth units. Same seed, same everything else, less forest. The cold is doing the work.
Two honest grains
Two caveats keep the story truthful. The first is that cold only slows a plant when that plant’s patch of ground is already at or below the temperature it likes best (the comfortable temperature band each plant grows within). Above its optimum a plant is warmer than ideal, and a small cool-down does it no harm. In the demonstration this was exactly the setup: the untouched ground sat at around 0.06, a hair under the meadow grass’s preferred 0.1, which is precisely why a further chill could bite. Cool a patch that’s already too warm and you’d measure nothing.
The second is that creatures are never chilled by the ground at all. An animal’s cold-survival reads the season, not the soil — so a slowly cooling climate never lays a cold hand on any one creature directly. It reaches them from underneath: colder ground grows less forage, and animals then go hungry. The direct danger of cold belongs to a hard winter, and the slow starving of a herd on land that can’t feed it is its own story; the climate drift only ever reaches an animal through the plants it eats.
A cold we can trust
None of this changes a world you can play right now — the drift is off. What it leaves behind is a mechanism we understand and a way to measure it, so that when we do decide a hundred-year world has earned a colder, slower forest, we’ll be turning a knob whose curve we’ve already read rather than guessing at a feeling. The cold, when it comes, will land where the plants feel it. We just wanted to be sure, first, that we knew which temperature that was.
