let search results settle by relevance, confidence, and recency
attract memory repel
Concept study · data-bound
Search Field Study
Concept study
A ranked list becomes an evidence landscape. Results are records bound through
bindData(); the Trust Gradient pattern turns
each result's confidence into a gradient and contradictions stand apart — still a plain list with the
field off.
Data honesty
- Data source
- Synthetic. The query, results, source confidence, and contradiction flags are hand-authored sample records — not a real search index or a real corpus.
- Snapshot
- Fixed sample data defined in-page; there is no committed data snapshot and no snapshot script feeds this study.
- Live refresh
- No external data, so no live refresh. The field re-settles as you open results or toggle the field.
- Not claimed
- The field turns each result's confidence into a trust gradient and marks contradictions. It does not rank relevance or decide which result is correct — the confidence values are illustrative inputs.
BeforeA plain ranked list of results with sources.
AfterConfidence reads as a trust gradient; contradictions are flagged; opened results are marked.
InspectLive
bindData().inspect() — records · bodies · edges. starting…
Field off stops the simulation and the list falls back to plain ranked results — different from turning a single projection off, where the field keeps running and only one drawn expression stops.
field ui evidence—
Patterns applied
Each is a real FieldPattern from the catalog. The ones marked
live are driven through applyPattern() on this page; the
rest are the conceptual family this study embodies.
show confidence, verification, and unsupported claims
charge link cohesion memory
show how multiple sources gather around claims, topics, or decisions
link charge gravity memory cohesion
let opposing states, tags, or preferences sort themselves visibly
charge repel attract diffuse
show where a user has been, paused, returned, or accumulated attention
memory diffuse