Jump to text
On this page
Copy and source
Published source confirmed at last check

Source changed 2026-11-18 22:51:26 UTC · snapshot created 2026-11-18 22:54:45 UTC · last check 2026-11-18 23:05:20 UTC

A.0:2 - Problem

The following mistakes obscure how a search result was obtained and what it can be used for:

  • Single‑winner bias. Teams look for “the best” option and publish a leaderboard, suppressing coverage & diversity signals essential to search.
  • Metric confusion. “Novelty” and “quality” are used informally; units and scales are omitted; ordinal values are averaged, breaking comparability.
  • Hidden policies. Explore/exploit budgets and governor rules are implicit; results are irreproducible and refresh‑unsafe (no edition/policy pins).
  • Tool lock‑in. Implementation terms (pipelines, file formats) leak into the Core, violating Guard‑Rails.

This glossary names the generative terms in Plain register and ties them to their formal anchors, so a publication can distinguish candidate scores, retained sets and selected results.