Core Principles

LDD rests on three principles. They describe the terrain: what the unit of work is, what constrains it, and why the classic incremental approach breaks with agents. Practices follow from these principles.

1. The Learning Loop is the Atom. The smallest unit of work is a complete learning cycle. Progress happens when you learn from users, not when you ship features.
2. Users Set the Absorption Ceiling. The pace of learning is capped by how much change humans can absorb, not by how fast an AI can ship.
3. Agents Drift When Context Is Rebuilt. Incremental builds force agents to reconstruct context on every slice, and each reconstruction introduces subtle drift.