Preview environment - you also see content that isn't published yet.

Hardcode

Stage 1 - four if-then rules. No machine learning, and that is exactly the point.

The simplest brain in the world: four if-then rules, expressed over the state index.

function hardAction(s) {
  if (s & 8) return 3; // carrying sugar → ZUM_BAU
  if (s & 4) return 1; // sees sugar    → ZUM_ZUCKER
  if (s & 2) return 2; // smells trail  → FOLGE_SPUR
  return 0;            // else          → ERKUNDE
}

Agent lesson: This is not machine learning - and that is the point. Hit reset and the colony behaves identically. No model, no training. Yet it is a full agent: it has a policy (state → action). Remember this table. The next stage will try to find it on its own.

→ In the lab: pick the “Hardcode” brain and hit reset