Glossary

Short definitions consistent with the manuscript. These terms name layers and moves in the book’s architecture—not synonyms for one another.

Answerability — The obligation to remain accountable to those affected by a decision made from a partial perspective, including what was learned before and after the choice.

Attention budget — The finite intake of a mind in real time: what gets processed now, what waits, and what may never arrive without a reroute.

Bias — The selective work through which finite perspectives become possible; not only prejudice or bad faith, but the narrowing that makes some things visible and others harder to reach.

Designed friction — Institutional or methodological delay that keeps a partial view from hardening before other partial views can count (peer review, dissent channels, appeal, replication).

Finite perspective — A view built from somewhere: always partial, never exhaustive, never from nowhere.

Institutional memory — What organizations preserve across people and years—protocols, stories, scars, and habits—so that pain can become structure rather than repetition.

Integration — The work of combining many partial views into a larger model none of them could build alone; not compromise, averaging, or performance.

Interpretation — The assembly of what attention and memory supply into a coherent account; closure that solves ambiguity and excludes other readings.

Judgment — The responsible use of limited perspectives under uncertainty: recommending, refusing, prioritizing, and committing while remaining answerable.

Machine perspective — A partial view assembled from data, architecture, and deployment; useful extension, not oracle or mere noise.

Memory archive — Selective retention and retrieval across time: what stays, what fades, and what returns when another version brings it back.

Proportionality — Matching the force of a decision to the evidence and stakes available—not acting as if partial sight were complete certainty.

Provisional commitment — Acting on the best partial case while remaining structurally willing to revise when new information arrives.

Revisability — Keeping decisions, models, and recommendations open to correction after the first pass.

Structure — How organizations assign what must be counted—metrics, mandates, workflows, and tools that coordinate sight at scale and edit what remains off the ledger.

Trust — Extending verification beyond what one person can check alone; also the channel through which incentive and delegation bend what reaches you.