By artifact
Move from raw evidence and research notes to hypotheses and durable theory.
View artifact typesVisual Engineering research archive
Find the thinking
behind the work.
Explore a living corpus of evidence, experiments, concepts, and working theories about how visual systems help people understand and act.
- 120
- Documents indexed
- 10
- Artifact types
- 8
- Research areas
Explore the corpus
Follow the shape of the research
By area
Trace related work across composition, publishing, evidence, and human factors.
View research areasBy status
Choose a path for orientation, decisions, application, verification, or deeper research.
View reader purposesStart here
Guides to the research
Start here · REP-BDE-0001
Beautiful Digital Experiences — Territory and Evidence Foundation
Start here · RP-CCE-0001
Clinical Communication Engineering Foundation Research Execution Package
Practical standard · STD-CCE-0001
Clinical Communication Engineering Foundation Standard
Project foundation · GOV-001
Composition Science Project Constitution
Evidence library
Composition Science Research Library
Research foundation
Component Library Foundations Research Report
The evidence supports a hybrid, native-first component library rather than a web-component-only architecture. Semantic HTML can remain relatively stable across substantial visual redesign, but only when markup represents durable information relationships rather than current visual groupings. Different categories require different ownership and encapsulation strategies. Native controls should remain native unless measurable requirements cannot be met; light DOM is favored for content-rich, composition-sensitive structures; Shadow DOM is favored for tightly bounded behavioral widgets whose internal invariants require protection. CSS should own presentation and contextual layout, while component logic owns only intrinsic behavior and state. Design tokens should encode decisions through a layered dependency model, but must not become a universal escape-hatch API. Implementation should begin only after a small set of targeted validation experiments confirms these rules in the project’s actual product contexts.