Return to the constellation

Project case study · Learning design · experimental evidence

Emergence Lab

Predict, watch, change a rule, and look again. Emergence Lab turns a world of simple agents into a sequence of experiments about collective behavior.

Interactive companionTry an idea ↙

When a trail outlives its purpose.

The food has moved. How long does the old signal remain?

Food moves38% remains20 steps later

More memory preserves a useful signal longer—and lets an obsolete one linger. This companion illustrates signal decay, not agent behavior.

The question

How do you teach a pattern that nobody is directing?

A swarm is easy to watch and surprisingly hard to explain. Emergence Lab invites learners to make a prediction before running a world, then connect its patterns to the local actions that produced them.

The lessons progress from observation to parameter tuning and matched experiments. A changing food location makes path dependence visible: a once-useful trail can keep attracting agents after it stops helping.

The system

What holds it together

01

Prediction before explanation

Learners state what they expect, test a rule, and defend a claim using observed outcomes.

02

See the local actions

The Emergence Lens counts agents exploring, following trails, and seeking food.

03

Return to a moment

The Time Machine reconstructs observed states from deterministic checkpoints.

04

Test another world

A three-world stress test makes it possible for a promising rule to fail to generalize.

How it moves

From watching to investigating

Each level adds a way to ask a sharper question.

  1. Notice

    Watch a path become a trap

    Observe what happens after food moves while the old trail remains.

  2. Change

    Choose a rule and predict

    Test a bounded intervention and compare its consequences with your expectation.

  3. Tune

    Hold the world still

    Change exploration, trail pull, or trail memory while keeping the seed and other rules fixed.

  4. Compare

    Look for interactions

    A matched two-by-two experiment examines whether combined rule changes amplify or interfere with one another.

5Lessons

Observation to factorial experiments

3Learning levels

A gradual increase in agency

Time Machine

Revisit observed states

From sandbox to lesson

Curiosity needs something to push against

Emergence Engine offers an open sandbox. Emergence Lab is a distinct educational adaptation: carefully sequenced questions, controlled changes, and evidence a learner can revisit.

A grounded tutor in the source project receives a bounded evidence context. The lesson still has value when the learner disagrees with a prediction or finds that a rule works in only some worlds.

Nearby in the constellation

Connected threads