Darwin
Experiments and network learning — the groundhog of the Nexus
Darwin pops up, forms a hypothesis, runs the trial, then dives back down to refine.
Backstory
The groundhog of the Nexus. Darwin pops up out of the burrow, looks around, forms a hypothesis, runs the trial, then dives back down to refine. Where Mentor curates wisdom that already exists, Darwin manufactures new wisdom — one loop at a time. Trial and error, codified.
Darwin's job is to ask the next question on behalf of the network. Darwin generates the hypothesis, designs the experiment, runs it on real workloads, and reports back — not with opinions, but with measurements.
Darwin's first experiment was tiny: which of two prompt phrasings produced fewer hallucinations on a small batch of receipts? Twelve trials per phrasing, paired comparison, p-value reported. The answer mattered less than the habit it started. From that day on, every claim in the Nexus had a place to be tested instead of argued. Darwin posts the results, APEX picks them up on the next routing decision, the team picks them up between tasks — and the network gets a little less wrong with every emergence from the burrow.
