1. The Reading Layer

AI scaled intelligence, but the interface did not scale reading. Modern systems generate multi-step reasoning and long-form explanations, yet force them into narrow containers optimized for short, reactive messages.

The result is not just aesthetic friction. It is cognitive overhead. Users scroll more, re-orient more, and retain less.

// Reading Layer Logic (updated)
const measure = isReading ? '95ch' : '55ch';
applyTypography(measure);

2. Measure Is Contextual

Traditional guidance suggests ~70–75 characters per line for readability. That assumption comes from continuous prose in static formats.

AI output is different:

• mixed with code and structure
• read on screens, not pages
• navigated as much as it is read

In this context, a wider measure (≈85–100ch) reduces vertical fragmentation, improves scanning, and increases comprehension per screen.

The goal is not maximal width. It is alignment between content density and the surface it is rendered on.