How it works

Research as a knowledge graph

A Glimmer study is a typed graph. Nodes are research objects — publications, methods, concepts, experiments, datasets/derivatives, people, and organizations — and edges are the relationships between them (addresses-concept, derived-from, uses-method, …). Instead of reading a folder of files, you navigate the structure of the work itself.

Every derivative node records the method, inputs, tool version, and a provenance hash that ties a result back to exactly what produced it. That's what makes a study reproducible rather than just published.

The explore hierarchy

The site is organized as a study hierarchy under /explore/:

/explore/                 hub — ask any open project + the all-projects graph
/explore/<study>/         study landing — public summary + an Ask box
/explore/<study>/graph    the study's knowledge subgraph
/explore/<study>/papers   the literature graph
/explore/<study>/data     data/volume viewers
/explore/<study>/compute  compute & cost HUD

The hub graph shows every project as a node; click one to drill into its landing. The brand mark (✦ Glimmer) always returns you to the hub to pick another study.

Public vs members

The landing and its public summary are open to everyone — name, what it studies, scale, references, and linked datasets. The data surfaces (graph data, papers index, PDFs, volumes) are members-only, gated at the server. So a visitor can always learn what a study is and how it's built, and members get the full picture inside. See the public data policy.