RimWorld doesn't give you a story. It gives you things worth turning into one.

I lost quite a lot of time to RimWorld this week. Not in the catastrophic sense — more in the dangerous I’ll just fix this one thing sense.

The freezer wasn’t working. Then somebody was injured. Then I needed more wood. Then I realised my outer walls were still made of wood. Then a prisoner went berserk. Then another colonist left. Then somebody got eaten by a spider.

At some point I was planning a defensive funnel while also worrying about whether I had enough medicine and whether I should start growing healroot.

This is how RimWorld gets you.

It is often described as a colony simulator, but the game describes itself slightly differently: a story generator.

That distinction is interesting.

The game doesn’t write the story

There isn’t really a conventional plot. You start with some people, some resources and a patch of hostile wilderness. Then systems start interacting — weather, food, power, relationships, health, animals, raiders, religion, mood, construction, trade, and your own increasingly questionable decisions.

Most individual events aren’t particularly interesting. A battery running out isn’t a story. A colonist becoming unhappy isn’t a story. Running low on wood isn’t a story.

But once those things start colliding, something changes. You start constructing meaning from them.

The freezer failed just as food became important. The person you were relying on got injured. The prisoner you thought you might recruit died. The colonist who had already caused problems suddenly became essential because another system changed.

RimWorld supplies the evidence. Your brain supplies the narrative.

You start editing the colony

This is probably the bit I find most interesting. Playing RimWorld feels strangely close to editing.

You’re constantly deciding: what matters? What can wait? Which signal is actually important? Which problem is connected to another problem? What should happen next?

The game produces far more information than you can sensibly care about. So you filter it. You create priorities. You remember certain characters. You ignore hundreds of routine meals but remember the dinner that gave everyone food poisoning just before something attacked the colony.

That is narrative construction. Not because somebody scripted the sequence, but because events gain meaning through their relationship to other events.

Failure makes the story better

I initially played RimWorld as if the objective was to run the colony correctly. I’m beginning to think that’s the wrong way to play it.

Some of the best moments have come from mistakes. I didn’t understand prisons properly and lost people I might have recruited. I underestimated fire. I built too much from wood. I didn’t understand why my freezer had stopped working. I reloaded fights because I was still learning what the game expected from me.

Eventually you realise there isn’t a perfect colony hiding underneath all your mistakes. The mistakes are the colony. Remove them and you probably remove much of the story.

That’s quite different from most productivity systems, where friction is something we’re constantly trying to eliminate. RimWorld needs friction. Without scarcity, bad decisions, conflicting priorities and occasional disaster, very little would be worth remembering.

Then the characters start writing themselves

My second run made this much clearer.

I started a settlement called Amphawa with a small group I thought I understood. Then the game kept adding context I hadn’t asked for.

Zac, the young writer who spent most of his time researching, turned out to be married to Glasses, who was much older than him. He became the colony’s religious figure. Then he ended the marriage. Glasses had a breakdown. Later, Lamp — the doctor I had rescued from another map — flirted with her by complimenting her muscles and became her lover.

None of that was a plotted storyline. It was relationship logic, traits, moods and random interactions. But once it happened, I couldn’t stop reading it as a story.

Karen was even more obvious. She arrived as a courtesan, became useful mainly because she could negotiate and hunt, survived winter and the general chaos of the colony, then got unexpectedly attacked and eaten by a cougar.

It felt like an episode of The Walking Dead. Not because the game had decided Karen’s arc should end there, but because she had been around long enough to become a character. The hunter got hunted. After that, predators near the settlement stopped feeling like background wildlife and started feeling like threats.

Then there was Ella: an excellent miner and soldier who gradually became part of the colony’s infrastructure. Only much later did I properly confront the fact that, as a Hussar, she was dependent on Go-juice. Suddenly a character trait I’d treated as flavour became an operational crisis. She ended up in a coma and the colony needed an external trading mission simply to keep her alive.

Again: systems first, story second.

Eventually the story becomes administration

This is the part I hadn’t appreciated when I first wrote this.

For a long time, every problem felt interesting because I was learning the game. How do I build a freezer? Why is the power off? How do I make a prison? Should I build with stone? How do caravans work? What happens in winter? How do I stop animals eating the entire pen?

Then, gradually, I knew the answers.

And the tasks changed.

Repair the fence. Expand the animal pen. Make more hay. Smelt the junk weapons. Add another generator. Make enough parkas before winter. Train a backup cook. Organise the warehouse shelves. Keep enough medicine. Maintain a Go-juice reserve for Ella.

The story generator had quietly become an operations generator.

That’s not a criticism. In some ways it’s proof that the simulation works. Once the unknown becomes known, surprise turns into maintenance. The same systems that previously produced drama become infrastructure you know how to manage.

But it changes the feeling of play. The I’ll just fix one more thing loop starts to feel less like discovery and more like an admin queue.

That was when I first thought about stopping.

But there was one open loop I actually cared about closing: Ella.

So I kept going just long enough to do that. I organised the caravan. I raided a steel work site on the way. I found a trader with neutroamine. I got the ingredients back into the colony, made the Go-juice and saved her.

That felt different from clearing an admin backlog. It felt like finishing a story beat.

Then I saved the game.

Not because Amphawa was finished. It absolutely wasn’t. There were still animals to manage, research to complete, resources to find and a hundred other things RimWorld would happily turn into another evening.

But I had reached a point where the only open loop I cared about was closed.

I saved Ella. Then I saved the game.

That felt like the right ending precisely because RimWorld wasn’t going to provide one for me.

Maybe that’s why it’s so difficult to stop

There is almost never a natural ending to a RimWorld session. Every solved problem exposes another unfinished one.

I’ll build the wall. Then I’ll sort the bedrooms. Then I’ll research batteries. Then I’ll deal with the mountain behind the settlement. Then I’ll just see whether this colonist joins. Then I’ll survive the next raid.

Each task is small enough to feel finishable. But the narrative never closes. There is always an open loop.

That’s an extraordinarily powerful design principle. It also explains why, after spending far too much time playing, I eventually had to consciously create my own stopping point. Save the game. Walk away. Come back later.

I’ve noticed I increasingly do something similar with my own projects. I put ideas into GitHub — not because every thought needs to become software, but because once the idea has a place to live I don’t have to keep carrying it around in my head.

It has a save point. I can leave the world.

The interesting bit isn’t the simulation

RimWorld is obviously a game. But I think there’s something useful hiding inside it about how humans understand complicated systems.

We rarely make sense of complexity by remembering everything. We select. Connect. Prioritise. Sequence. Name things. Notice patterns. And eventually construct a story that explains what happened.

That’s remarkably close to what I’ve been thinking about with Editorial Intelligence. Organisations generate enormous quantities of information: meetings, research, customer conversations, analytics, documents, messages. Having more of it doesn’t automatically create understanding.

Someone still has to decide what matters. Someone still has to connect the signals. Someone still has to turn the system into a story people can understand.

The second run added another layer to that thought. Once a system becomes familiar, the challenge changes from interpretation to maintenance. You still need judgement, but now it is about deciding which processes deserve attention, which dependencies matter, which risks are worth carrying and when to stop adding more complexity.

And maybe that is part of narrative judgement too: not every open loop deserves to be closed. The useful question is which one matters enough to keep going for.

RimWorld makes both sides unusually visible: narrative emerges from complexity, then operations emerge from understanding it.

Although, unlike most editorial projects, there’s always a chance somebody will get eaten by a giant spider. Or a cougar.