This series is intentionally exploratory and will likely feel messy at times. Because current large language models have very little internal machinery that clearly corresponds to what we normally call "understanding," the investigation has limited established language and concepts to draw upon. As a result, some chapters may feel provisional, repetitive, or incomplete as we probe the boundary between surface statistical coherence and deeper mechanical robustness. The messiness is not a flaw — it reflects the genuine difficulty of the subject.