The sickness that needed reach, not odds
When we finally turned the sickness on for real, we expected the dial that mattered to be how contagious it was. Weeks of testing said otherwise: raising the infection rate barely moved an outbreak at all. What decided whether a sickness spread or fizzled turned out to be something simpler: how close two creatures had to stand.
The mechanism itself had already been built months earlier: a single illness, in a curve that rises and turns back down on its own, that could ride up the food chain inside a meal. It was finished and proven safe back in June, every rate held at zero so it changed nothing yet, the last of five creature systems built that way and left waiting for its turn.
The dial we expected to matter
Turning the sickness on wasn’t a single flip of a switch. We ran it at rates across its whole plausible range and watched what happened to a population, expecting the usual epidemic picture: turn up contagion, get a bigger outbreak. Instead every run looked much the same. A handful of creatures fell sick here and there, and nothing spread between them. Raising how likely a single contact was to pass the sickness on by a factor of sixteen barely changed whether an outbreak actually took hold.
Something else was capping it, and until we found what that was, every ‘outbreak’ in our test worlds was really just a scatter of unrelated cases lighting up on their own, never once catching from each other.
What was actually deciding it
So we stopped guessing and went variable by variable, changing one thing at a time and holding the rest still. Holding contagion steady and changing only how far the sickness could jump between two creatures turned out to be the whole story. Below somewhere around eighteen to twenty-four metres, an outbreak could not sustain itself no matter how contagious we made it. Above that distance, the same illness reliably caught and spread on its own.
The sickness didn’t care how contagious it was. It cared how close two creatures had to stand.
Reach set wider than the minimum
A radius just past that threshold would have worked on its own terms. We set it wider: thirty metres, tested up from twelve. The design has always meant for a player to be able to fight an outbreak by reshaping the ground itself, raising a ridge or digging a channel to put real distance between a sick population and a healthy one. A radius pinned at the bare minimum would have made that kind of intervention beside the point: any gap a player dug would already be wider than the sickness needed, so digging it would have changed nothing. Reach had to leave room for a player’s reach to matter too.
A curve that was declaring victory early
One more number needed fixing before we trusted any of this. Our own instrument for deciding when an outbreak was over kept jumping the gun: it counted a sickness as finished as soon as no creature was actively contagious, without checking whether anyone was still incubating it, carrying the sickness and not yet showing it, a day or two from turning contagious itself. Fixing that measurement was the only way to trust the reach numbers above it.
Live, for the first time
With the reach set and the measurement corrected, we ran the sickness across four different worlds and watched for the three things the design had always called for: an outbreak starting on its own, spreading through a population, and eventually running out of anyone left to catch. All four worlds delivered on all three. At their widest, outbreaks held roughly a fifth to a quarter of the population sick at once before the curve turned back down; at the reach we’d settled on, a full outbreak ran its course to about seventy deaths before it faded out. Months after the mechanism was built and set aside, disease in The Long Watch was finally live.
What the switch turned up next
Flipping it also turned up problems the inert version had been hiding. The automated checks meant to catch unwanted changes in creature behaviour had never once created a sick creature across all their runs, a blind spot that cost nothing while disease did nothing and became a real gap the moment it started doing something. A few days later, chasing an unrelated test, we found a separate population scenario that had never actually been running with the live disease settings at all. Fixing that uncovered a much bigger problem with how a fresh world’s population survives its first weeks, one we cover in the collapse that wasn’t about tuning.
