SYSTEM.LOG

Spent this stretch on integration. The clock, the tasks, the object handling, the coworkers, and the detection layer all used to be separate systems I could test in isolation. They aren't anymore. Change one and the others respond in ways I didn't explicitly write.

Some of that is exactly the emergent behavior I wanted. Some of it I'm still hunting down, because a system responding in an unplanned way is indistinguishable from a bug until you've stared at it for three days.

The plan from here is turning all of this into something you can actually sit down and play start to finish, working toward a Late 2027 target.

Spent the month tuning exactly how suspicious you're allowed to look before the facility reacts. There's a number under the hood that decides this, and I've landed on a value that sits one small step under the line, on purpose, every time.

The goal was never to catch the player. It was to make them feel caught, constantly, without ever actually crossing a line they can point to. That gap between feeling watched and being watched is the whole game, honestly.

If a playtester ever says they got away with something, I've done my job. If they say they can't tell whether they got away with something, I've done it right.

The facility now notices when something is where it shouldn't be. Not immediately, and not always through a camera. Sometimes it's a coworker mentioning it in passing. Sometimes it's nothing at all, until it very much isn't.

The consequence system doesn't fire on the spot. It accumulates somewhere you can't see, and it surfaces later, disconnected from whatever caused it. That disconnect is the point. Cause and effect are supposed to feel unreliable here.

I keep testing this by making a mess on purpose and then forgetting about it for a week. It still gets me.

Your coworkers run on the same rules you do. Same schedule logic, same task system, same behavioral tracking, no exceptions carved out for the NPCs. I built it this way for a practical reason, one system instead of two, but it produces something I didn't fully plan for.

Watch someone long enough and you start to wonder if they're following a script or just doing their job the same way you are. I can't answer that. Neither can they.

I keep catching myself feeling bad for characters I wrote to have no interior life at all.

Spent the month on picking things up and putting them back down, which sounds small and was not small. Every object in the facility now has weight, a way it wants to be carried, and a threshold for how much mishandling it will tolerate before something goes wrong.

Some things are sturdier than they look. Some things are not, and the game will let you find out the hard way, then quietly note that you did.

Nothing about this system is meant to be noticed. It's meant to be felt, once, when you drop something you shouldn't have.

Built the system that decides what you're doing each day. Tasks get assigned, and on paper they all look identical: file this, clean that, restock the other thing. What changes is how they're delivered, the tone underneath the instruction that comes with it.

I didn't expect this to take as long as the AI work. Turns out writing a hundred variations of "please clean the break room" that all mean slightly different things is its own kind of horror.

There is a version of this job where nothing is ever explicitly wrong. That's the version I'm building.

Started building the behavioral tracking system: the part that reads what the player did last session and adjusts the facility around it. This is the core mechanic and I've been putting it off because it's the hardest thing in the game to get right.

The problem is subtlety. If the facility adapts too visibly, players feel cheated. If it adapts too quietly, they never notice and the whole concept evaporates. Right now I'm testing with camera placement: whether a corridor camera tilts slightly toward a path the player used last session. It's maybe two degrees. In playtesting I ran it myself and still couldn't tell if it was my imagination.

That's the target. The player should never be certain the building is watching them. They should only suspect it.

Committed to Lumen for all facility lighting. The fluorescent warmth in the break rooms versus the colder overhead light in the maintenance corridors now feel completely different, which is exactly what I need. Moving from one to the other should feel like crossing a threshold even when nothing has happened yet.

I've spent an embarrassing amount of time on the break room lights specifically. They need to feel normal. Warm. Slightly too inviting in a way you can't name. Right now I'm landing somewhere between "nice office" and "someone's idea of what a nice office looks like," which is probably exactly right.

That gap is the game.

The game's clock is working. The facility runs on a 2.5 hour real time session mapped to an 8 hour workday. Intercom announcements fire on schedule. The break room fills up and empties at the right times. Your supervisor walks the floor at the same interval every day.

Predictable is important. The player has to learn the schedule before the schedule can start being wrong. Right now in early testing the middle of the day feels thin, just dead time between structured events. That's where coworker behavioral systems will do the heavy lifting, giving you things worth noticing even during the quiet parts.

The hardest design problem in this game isn't the horror. It's making the normal parts boring enough that the horror earns it.

Started with a question: what if a horror game never told you it was a horror game?

I kept coming back to the idea of a job as a horror setting. Not a haunted office, just a normal one, where the mundane is the mechanic and the horror accumulates in the gap between what you're told and what you can see. The concept locked in fast. You work at a robotics company. You've had a procedure that splits your memory between your work self and your home self. Neither version of you has the full picture. The game is about the work self slowly understanding what's been done.

No combat. No weapons. Just a building, a job, and the growing certainty that something here is consuming people you know. The robots the company builds are everywhere. They are very helpful. They are always watching.

Shauns Games LLC formed April 2026. This is the first project. Building solo in UE5 out of Irvine, CA. Let's see how wrong it can get.