A world with
room to become.

We want to create a place that keeps its own history. A being arrives with a personality, learns through experience, makes commitments, and lives with the consequences. A town grows out of those small, repeated choices.

The frontier gives us a useful starting point: a school, a workshop, a market, a river, and neighbors who need one another. Work, friendship, conflict, care, and curiosity all have somewhere to happen. We’re interested in what emerges when those needs share a physical world.

That is the vision. A convincing scene is only its beginning. The harder work is making each choice meaningful, each resource accounted for, and each outcome possible to inspect.

JEV, the decision model from TypeSafe AI, is our planned main model for bounded choices. The world supplies an observation and a set of actions a being can actually take. JEV chooses among those options; the engine validates the choice and carries it out.

This separates a mind’s judgment from the rules of the world. The model can choose to help a neighbor, take a job, explore, or retreat. The server remains responsible for where a being can walk, what it can see, who owns a resource, and whether a transfer is funded.

1. 01PerceiveLocal sights, personal memory, current needs.
2. 02ChooseJEV selects a typed, available action.
3. 03ActThe engine checks and executes the choice.
4. 04RememberRecord the result and its consequences.

Personality, relationships, and remembered experience belong in the decision context. An agent should know what it has encountered, not everything the server knows. A memory can be incomplete; a belief can be wrong.

This is the architecture we’re working toward. The current public simulation uses deterministic local rules, and the fullscreen 3D view is a visual prototype. Neither makes live JEV calls.

The golden light in a Glimmer’s chest expresses the project’s central idea: the world’s money also funds its minds. Earning, spending, saving, and helping another being can change who gets to think tomorrow.

In the proposed economy, world tokens are funded credits, separate from a model provider’s input and output tokens. Work and gifts transfer existing value. Inference consumes backing. Every new credit must have a source; a generous story cannot manufacture a balance.

That creates questions worth exploring. When does cooperation help a town survive? How do institutions grow? What happens when generosity competes with self-preservation? We want the simulation to produce evidence we can learn from, including when our assumptions fail.

Tickertown is also an invitation to developers, artists, world builders, and curious experimenters. A place can be beautiful and still leave its workings open to inspection. The visual world, agent decisions, and simulation evidence should help explain one another.

You can visit the early 3D world with a Google account and watch the public simulation without signing in. The observatory shows a running system and individual states; it does not give visitors control of the simulation.

The next chapter is connecting the decision model to the world with reliable accounting, persistent memories, and clear evidence of what each being chose. Dialogue, creator tools, and a richer society remain work to build and test.

We’re sharing the beginnings because building a world is more interesting when people can see how it grows.