Recursive Self-Improvement (RSI)
Recursive self-improvement (RSI) is the process by which a system improves its own ability to improve: each round of change makes the next round better, faster or safer. In AI research the term usually describes a hypothetical system that rewrites itself without limit.
Squillo uses the term narrowly and mechanically. N Lang makes self-evolving code a first-class language feature: a type opts in through the Adaptive trait, the ledger records every call as training data, the runtime keeps a population of implementation variants, a fitness function scores them, the best-performing variant is selected and a regression rolls back. Snapps built this way can write, test and improve other Snapps, and every improvement is type checked and versioned before it lands.
What Squillo does not claim is that the system rewrites itself without limit. Every evolution is scoped to a type that opted in, scored against real calls, recorded on the ledger and reversible. See self-evolving code on the N Lang page.
