# Player guide: shape a world, read its history LifePot is a bounded artificial-life game. Describe an environment, review the proposed ecology, then watch a deterministic simulation. The model may propose typed choices; it cannot write rules or control organisms. Extinction is a valid outcome, and this is not a biological forecast. ## Start from the questions The setup asks three short questions. Each answer is trimmed, must contain 1–140 characters, and is interpreted into a fixed vocabulary—not compiled as a rule. - **What exists in this world?** Resource abundance and distribution; phrases such as “rich,” “scarce,” “clustered,” “oasis,” “seasonal,” and “cycle” map to configured environment choices. Founder balance can also be inferred from “prey heavy” or “abundant prey.” - **What threatens life here?** Selects a hazard (`drought`, `toxin`, `heat`, `crowding`), volatility (`stable`, `pulsing`, `chaotic`), and related pressure. “toxic,” “heat,” “competition,” “pulse,” and “unpredictable” are examples recognized by deterministic setup. - **What should evolution favor?** Weights five priorities: `survive`, `replicate`, `cooperate`, `explore`, `adapt`. Words such as “endure,” “offspring,” “share,” “spread,” and “diversify” influence the mapping. Weights are normalized to sum to one. The mapping is deliberately literal and limited. For example, “toxic” selects the toxin hazard, but does not create a new chemical simulation. Unmatched prose falls back to the documented defaults; it is not an instruction to the engine. ## Try the recorded preset The in-game **Explore deterministic preset** button fills these exact answers: ```text World: Rich mineral pools with abundant prey Threat: Predator packs hunt through pulsing droughts Reward: Diversify while prey and predators coexist ``` Review the resulting setup before starting. The preset is an example, not a guaranteed winning strategy. With no server API key, the game still runs using labeled deterministic fallback choices; a live interpretation may vary among allowed values. ## Review before starting The review step exposes the generated configuration and its source/provenance. Check resource conditions, fitness weights, founder balance and strategies, and the species interaction graph. Setup establishes initial conditions; it does not guarantee population survival. **Restart** in the running world's controls reruns the current answers, config, and seed; it is not a new randomized setup. At completion, **Change objective** returns to the third setup question; reload the page to begin again with all three answers. To share/replay an accepted completed run, use **Copy challenge link**. Replay starts from the recorded choices and seed instead of asking the model to interpret them again. ## Observe and inspect The world is a 50×50 toroidal grid: movement across an edge wraps to the opposite edge. The canvas depicts organisms, species colors, and resource patches. Use **Pause** before close inspection: a paused run keeps the state stable while you select an organism. On a narrow screen, the interface stacks the world, controls, inspector, event feed, statistics, and observatory vertically; scroll to reach them. Select a living organism to open the Creature Inspector. It describes the selected individual's current species, guild, strategy, energy/age/traits, and lineage context where available. Selection is an observation, not an intervention: it does not change the simulation. A selected organism may die as the world advances, so inspect its status again after resuming. The World Observatory summarizes population, prey/predator counts, births, deaths, kills, resources, species richness, lineages, and the recent ecological history. These are aggregate simulation counters, not estimates of real-world populations. The decision ledger records accepted evolution decisions and their evidence/source. A fallback entry is not Jev evidence. ## Read outcomes without overclaiming A simulation can finish as `extinct`, `surviving`, or `thriving`. Extinction means the simulated population reached zero; `thriving` is an engine-defined end state, not a biological judgment. Runs stop at generation 180 at the latest. The world may end earlier if it becomes extinct. Resources regenerate and can be depleted; organisms feed, move, reproduce, mutate, age, and die according to engine rules. Fitness weights influence bounded mechanics; they do not promise that the named priority wins. Mutation can be neutral or harmful. The engine does not rescue a population to make a better story. ## Replay and its limits A replay carries the engine version, canonical answers, validated config, unsigned 32-bit seed, setup hash, ordered decision ledger, and contract hash. Replay validation re-runs the deterministic engine against that ledger. It does not call the interpretation model. Modified, incomplete, reordered, unreachable, unsupported, or tampered replay data is rejected rather than “best effort” played. Replay compatibility is versioned: a replay recorded for another engine version may be unsupported. A valid replay establishes reproducibility under its recorded engine contract; it does not prove the model's ecological explanation is true. ## If something looks wrong - **Choices differ from what I wrote:** setup maps prose to finite values; inspect the normalized review rather than assuming every phrase has a direct mechanic. - **The world ends quickly:** extinction is an allowed result. Try different resources, hazard, founder balance, or priorities, then start a new run. - **No live model interpretation:** the game can use deterministic fallback. Check the displayed provenance; never label fallback as a model response. - **Replay rejected:** check for truncation/tampering and supported engine version; replay validation intentionally rejects unsupported payloads. - **Controls or inspector are off-screen:** pause, scroll, and use the vertically stacked mobile layout. ## Further reading - [Developer reference and engine internals](/learn/developer-reference) - [Architecture](https://github.com/tugrulguner/lifepot/blob/main/docs/architecture.md) - [Simulation contract](https://github.com/tugrulguner/lifepot/blob/main/docs/simulation-contract.md) - [Deployment and protection](https://github.com/tugrulguner/lifepot/blob/main/docs/deployment.md)