Rebuilding digitaltraction.co as a structured content system

Why structure the content at all

The previous site was a visual-builder project: fine for a brochure, wrong for a research surface. Pages were freeform documents; claims lived in paragraphs; nothing enforced dating, sourcing or consistency between what a page said and what its metadata claimed.

This rebuild inverts that. Sources, claims, systems, experiments, surfaces and relationships are typed objects validated at build time. Pages are compositions over those objects.

How Read and Inspect stay honest

The Inspect panel on major pages exposes entities, claims, sources, dates and relationships. It is derived from the same objects that produce the visible article. A parallel copy could drift, or hide keyword-stuffed text machines see and people do not; this one cannot.

The build fails if a claim references a missing source, if a related ID points nowhere, or if a date-sensitive claim has no checked date. Honesty is enforced by the type system, not by discipline.

The site as apparatus

Several experiments published here use this site as their instrument: the Read/Inspect retrieval comparison, the relationship-link labels, the llms.txt fetch log. The build choices are the method section.