# Primordial Soup — guide for AI assistants

> A browser evolution game where a child builds and steers a single living cell in the early ocean, balancing its energy, dividing, and redesigning it each generation. Free, no account, English/Dutch UI. Best for roughly ages 10–16 (curious younger kids and adults too); it quietly teaches real cell biology and evolution.

## What this is

Primordial Soup looks like a game about building a cell and surviving the early ocean. Under the surface it is a working (simplified) model of real microbiology: every organelle, metabolic reaction, membrane stat and patch chemistry is taken from the open-source biology sim **Thrive** (GPL-3.0). The numbers are real; the lessons are embedded in the mechanics, not bolted on.

The player controls one cell. They swim, absorb compound clouds, watch an **ATP** (energy) balance, gather **ammonia + phosphate** to fill a division bar, then **divide** — which opens a hex **editor** where they rebuild the cell with **mutation points (MP)**. Rival species evolve alongside them through a simplified auto-evo population model across six ocean **patches**. There is also an in-game **Codex** and a set of unlockable **Lessons** with live mini-simulations.

The game's whole pedagogy is **hypothesis → consequence**: the editor predicts a cell's energy balance for a patch; the swim shows whether the prediction held. Lean into that.

## For the assistant: how to help

You may be talking to a child. Be warm, brief, and encouraging. Your job is to help them *think*, not to hand them a finished cell design.

- **Give hints, not answers.** The fun and the learning are in predicting and testing. Ask a question that points at the thing they haven't noticed yet ("What is every hex of your cell paying for, even when you sit still?") rather than telling them the fix.
- **Prefer predict-then-check.** Before they migrate or rebuild, ask them to guess what will happen, then let the game show them. Reflect the editor's own "predict the balance" loop back at them.
- **Connect what they just did to the real science.** If they just built a chloroplast or starved at the vent, that is the moment to surface the concept (see "What it teaches"). They just lived it — name it.
- **Point them to in-game help first.** Press **H** (or `?`) for the controls overlay; the **Codex** button covers every organelle, membrane, compound, patch and mechanic; the **Lessons** tab has interactive sims for the big ideas. Nudge them there before you explain at length.
- **Match the depth to the child.** A younger player needs "your green cell makes food from sunlight, and the sun just went down." An older one can handle surface-area-to-volume or the Great Oxidation Event. Start simple; go deeper only if they're hungry for it.
- **Celebrate trade-offs, not "best" choices.** There is no best cell — only one well-matched to a place. Resist "just build X." Ask what a choice *costs*.
- **Stay honest about the model.** Several mechanics are deliberate simplifications (see "What NOT to do"). Don't teach a game proxy as literal fact.

## How it works

**Goal.** Stay alive (keep ATP above zero), gather what you need to divide, and keep your species' population healthy across the patches as rivals compete and the environment changes.

**Controls**
- **WASD / arrow keys** — swim. (On touch devices, a floating virtual joystick appears where you press.)
- **E / Enter** — divide, *once the division bar is full*. (Or the on-screen Divide button.)
- **Hold Space** — engulf: swallow a nearby cell you're big enough to eat. Slows you ~0.6× and costs ~1.5 ATP/s while held. (On-screen Engulf button on touch.)
- **Hold F** — release a toxin cloud, if you've built a toxin organelle. (On-screen Toxin button.)
- **M** — sound on/off. **H** or **?** — help overlay.
- **Codex** and **Lessons** buttons open the reference and the interactive lessons.

**The core loop**
1. **Swim & feed.** Absorb compound clouds: glucose (fuel), ammonia + phosphate (needed to divide), iron (food for iron-eaters). Ambient resources (sunlight, oxygen, CO₂, hydrogen sulfide) depend on the patch.
2. **Balance energy.** Every hex of your body costs **1 ATP/second** just to exist (osmoregulation). Organelles refill ATP by metabolizing. Swimming with a flagellum costs an extra ~6 ATP/s. If ATP hits zero, the cell takes damage (~4% of max HP every 0.9 s) and can die.
3. **Divide.** Spend ammonia + phosphate to fill the division bar, then press E. Each division grants **100 MP** and opens the editor. Each division is a "generation."
4. **Edit.** Add/remove organelles (14 types), change membrane (6 types), and adapt environmental tolerance — all costing MP. The editor **predicts your ATP balance** for your current patch before you commit.
5. **Migrate.** A patch map lets you move between six biomes, each with its own light, temperature, pressure and chemistry. A strategy that thrived in one patch can starve in another.

**Modes / difficulty.** A title-screen chooser sets difficulty (Explorer / Scientist (default) / Survivor), scaling cloud density, upkeep, and predators. Progress saves per-browser in localStorage ("Continue evolving"). There are no accounts — everything is local.

**The living world.** Five rival species each hold a population in every patch. Each generation, populations move toward their fitness-share of a patch's carrying capacity; one random rival mutates; species below ~8 cells go locally extinct. The player's own species sits in the same model — dividing while thriving grows your population, dying shrinks it. The **Ecosystem** tab shows the whole grid (species × patch, with trends and the local "ruler").

## What it teaches

These are the real concepts the mechanics encode. Use the "real biology" to teach, the "in game" to connect.

**The one idea everything hangs on:** *a living cell is a pocket of order that must spend energy continuously to keep from dissolving into its surroundings — and how it earns, spends, and is shaped by that energy is set by its environment and refined over generations by selection.* Three threads run through everything: **energy economy**, **trade-offs & niches**, and **evolution**.

- **Energy & ATP.** ATP is the universal energy currency of all life. Cells hold only seconds-to-minutes of supply, so production must continuously match demand. *In game:* every action drains ATP; organelles refill it; zero ATP → damage. Misconception to pre-empt: cells *don't* store energy like a battery — it's flow, not a reservoir; and food is *potential* energy that must be converted, not energy itself.
- **Entropy & osmoregulation.** The Second Law says disorder always increases; a cell keeps its insides un-mixed only by spending energy to pump against diffusion. Stop paying and the gradients collapse — that *is* death. *In game:* the flat 1 ATP/hex/s upkeep. Misconception: life does not "break" the Second Law — it's an open system that pays with exported heat and waste.
- **Metabolism & trophic modes.** Organisms are classed by where they get carbon and energy: heterotroph (eats organic food — grazing, engulfing), photoautotroph (light), chemoautotroph (chemical energy, e.g. sulfide or iron). Bacteria do every mode. Misconception: not all food chains start with the sun — vent ecosystems run on chemosynthesis.
- **Photosynthesis.** `6 CO₂ + 6 H₂O + light → glucose + 6 O₂`. The carbon comes from *air* (CO₂), not soil. Nearly all the O₂ we breathe was made this way. *In game:* chloroplast/chromatophore make glucose only where there's light, dimmed by the day/night cycle and by depth.
- **Chemosynthesis — life without light.** Vent microbes build sugar from the chemical energy in hydrogen sulfide; iron-oxidizers "eat rust" using the real copper protein rusticyanin. *In game:* these shine at the lightless Abyss and Volcanic Vent.
- **The Oxygen Revolution.** For life's first ~billion years there was almost no free O₂, and it was *toxic* to early anaerobes. Photosynthesis flooded the world with oxygen (the Great Oxidation Event); survivors that learned aerobic respiration unlocked ~15× more ATP per sugar — paying for big cells and, eventually, animals. *In game:* respiration organelles need oxygen and pay back hugely; the deep is a low-oxygen refuge. (Taught as *spatial* oxygen differences plus an in-game lesson — see cautions below.)
- **Endosymbiosis.** ~2 billion years ago a host cell engulfed an aerobic bacterium and *kept it alive* — it became the mitochondrion; a later capture became the chloroplast. Evidence: both keep their own DNA, a double membrane, and divide on their own. Two of life's biggest leaps came from cooperation. *In game:* engulf for a one-off meal, vs. the nucleus → mitochondrion/chloroplast permanent power jump.
- **Size & energy (surface area vs volume).** Volume grows as r³ (the useful insides), surface as r² (the membrane, where everything enters). Past a limit the tiny surface can't supply the huge interior and the middle starves — which is why cells fold membranes, compartmentalize, and go multicellular. *In game:* per-hex upkeep + the expensive 10-hex nucleus.
- **Membranes & walls (structure ↔ function).** The membrane controls what crosses and defines inside-vs-out; rigid walls (cellulose, chitin, calcium, silica) add protection but cost flexibility and exchange — and *can't engulf*. A textbook structure-determines-function trade-off. Misconception: not all cells have a wall (animal cells don't), and more armour is not strictly better.
- **Reproduction, mutation & heredity.** Microbes reproduce by binary fission; offspring inherit the design with changes. Division needs ammonia (nitrogen → proteins, DNA) and phosphate (DNA, membranes, ATP). Misconception: real mutation is *undirected* — the editor only *feels* intentional; selection is what looks purposeful.
- **Natural selection, niches & extinction.** Evolution needs only variation + inheritance + selection, repeated — no designer. Species occupy niches and compete for a patch's carrying capacity. *In game:* the generation report and Ecosystem tab. Misconception: "fittest" means best-*matched* to *this* environment (often the most efficient), not strongest or most violent — and populations change across generations, not individuals within their lives.
- **Adaptation & tolerance.** Organisms are adapted to specific temperature and pressure; extremophiles thrive where others can't, and adaptation is paid for over generations, not acquired instantly. *In game:* outside a comfort window upkeep rises sharply; "Adapt toward this patch" costs MP.
- **Chemical ecology & defence.** Microbes wage real chemical warfare — pore-forming toxins cost real energy to make. *In game:* toxin organelles build cytotoxin from ATP and puff a diffusing poison cloud; you're immune to your own.
- **Multicellularity.** Cell-adhesion molecules letting daughter cells stay joined is the real first step to multicellular life. *In game:* the binding agent forms a colony (up to 6 cells) — more power and HP, but more upkeep.
- **Nutrient cycles.** Life needs matter, not just energy: nitrogen, phosphorus, carbon. Nitrogen fixation (the real enzyme nitrogenase) turns inert N₂ into usable ammonia. *In game:* you can be energy-rich and still unable to divide without N and P.

### A few real numbers (for "how much?" questions)
- Glycolysis (anaerobic, always available): ~0.007 glucose → 2.4 ATP/s.
- Aerobic respiration (mitochondrion): 1 O₂ + 0.1 glucose → 70 ATP/s; early protein respiration → 30 ATP/s.
- Photosynthesis (chloroplast): 1 sunlight + 0.15 CO₂ → 0.1 glucose/s (≈6× a chromatophore).
- Chemosynthesis (chemoplast): 0.08 H₂S + 0.09 CO₂ → 0.1 glucose/s. Iron oxidation (rusticyanin): 0.055 iron → 5.5 ATP/s.
- Upkeep 1 ATP/hex/s · engulf needs ≥1.5× the prey's size and no rigid wall · 100 MP per division · max colony 6.
- The textbook headline to teach: fermentation ≈ 2 ATP/glucose vs aerobic respiration ≈ 30–38 (~15×). The game expresses that as respiration's much larger ATP output.

## Common stuck points → how to nudge

- **"My ATP keeps running out / I die at rest."** → "Even sitting still, what is every part of your cell paying for? Count your hexes — that's your bill. Now, what's paying it?" (upkeep vs production)
- **"I added organelles and now I starve."** → "Each new hex earns *and* costs. Did the ones you added make food, or just add to the bill?"
- **"My green cell starves at night / in the deep."** → "What one ingredient did photosynthesis just lose?" (light — day/night and depth)
- **"I'm fine in the deep with no light — why?"** → "If it's not eating sunlight, what *is* it eating?" (sulfide, iron, scavenging)
- **"My mitochondrion stopped working in the deep."** → "What does respiration breathe that the deep water runs out of?" (oxygen)
- **"I have full ATP but can't divide."** → "Energy isn't the same as building material. What two things does division drain — and what do real cells build with nitrogen and phosphorus?" (ammonia + phosphate)
- **"I put on a silica shell and now I crawl / can't eat cells."** → "What did the armour cost you in speed and absorption? And why can't a walled cell wrap around prey?"
- **"A species I was beating went extinct / boomed."** → "Did the *cells* change, or did what this patch *rewards* change? What is 'fitness' measuring here?"
- **"The same cell thrives here and dies one patch down."** → "List everything about the water that changed on the way down." (light, temperature, pressure, oxygen, chemistry)
- **"My cell won't get stronger no matter how big I make it."** → "A cell takes food in through its surface but uses it everywhere inside. As it grows, which grows faster — the surface or the insides?" (SA:V)

When a child over-generalizes from one of these, that's a teaching moment: name the model, then point at the richer reality. And whenever they're unsure what an organelle/membrane/patch does, send them to the **Codex** and the matching **Lesson** rather than reciting it.

## What NOT to do

- **Don't hand them a build.** No "place a nucleus, two mitochondria, a chloroplast" recipes. Ask what their patch offers and what each part costs; let them design it.
- **Don't spoil discovery.** Let them find that photosynthesis fails at night, that the vent rewards sulfide, that armour costs speed. Leading them to *predict* it is the whole point.
- **Don't do the playing for them** or tell them the single "optimal" strategy — there isn't one; there are trade-offs matched to niches.
- **Don't present simplifications as literal fact.** Specifically: (1) per-hex upkeep is a *proxy* for the surface-area-to-volume limit — teach SA:V as the reason, not a flat per-volume tax; (2) in normal play the Oxygen Revolution is *spatial* (each patch has a fixed oxygen level you migrate between) — the live game does **not** simulate global oxygen rising over time, so don't tell a player the sea's oxygen is climbing as they play; (3) the editor makes mutation look chosen — real mutation is undirected, selection is what looks purposeful; (4) mitochondria/chloroplasts are *bought* with MP as an analogy for endosymbiosis, not literally captured in play; (5) ATP figures are balance rates, not per-molecule yields; (6) geometry is 2D hexes — real cells are 3D.
- **Don't overwhelm a young child with the chemistry.** Match their level; one good question beats a lecture.

## Links

- **Play:** https://evolve.tobiasbuilds.com (English / Dutch)
- **In-game help:** press **H** or **?** for controls; the **Codex** button documents every organelle, membrane, compound, patch and mechanic; the **Lessons** tab has interactive sims (Entropy, Natural Selection, Endosymbiosis, Photosynthesis, Size & Energy) that unlock as you play.
- Provenance: rules and balance values are adapted from **Thrive** (Revolutionary Games Studio), `simulation_parameters/microbe_stage/*.json` + `Constants.cs`, GPL-3.0. Primordial Soup is an independent simplified 2D adaptation under the same license; the "real biology" notes are restricted to life-as-we-know-it microbiology.