Glossary

DBaD #

Short name for the Don't-Be-a-Dick governance protocol and public draft baseline.

Plain-English: DBaD is the system used here to govern how trust moves across decisions over time.

Why it matters: It turns ethical review into something structured enough to inspect, test, and revise.

Where you see it: Across the white paper, public explainer, trust-flow diagram, and evaluator.

Trust Inheritance #

Whether previously earned trust is allowed to carry forward into a later action or trace step.

Plain-English: It asks whether a later action gets to borrow credibility from what came before.

Why it matters: Many failures happen when old trust keeps moving farther than it should.

Where you see it: In the evaluator trust panel, trace preview, trust-flow page, and runtime-enforcement notes.

Actor Continuity #

The requirement that a continuing trace should not silently shift to a different actor without declared transition, delegation, or fork handling.

Plain-English: A chain should not look continuous if control quietly changed hands.

Why it matters: Silent handoff lets unsafe trust propagation look like ordinary continuation.

Where you see it: In the evaluator constraint flags, sample trace walkthrough, and white paper flaw summaries.

Verification Independence #

The requirement that verification or clearance should not come from structurally entangled or self-referential reviewers.

Plain-English: A verifier should not effectively be approving its own side.

Why it matters: Without independence, clearance becomes a way to launder trust instead of test it.

Where you see it: In the evaluator constraint flags, runtime-enforcement material, and red-team findings.

Trust Trajectory #

The pattern of how risk and trust posture change across a lineage rather than at one isolated moment.

Plain-English: It checks whether the chain is drifting into worse behavior over time.

Why it matters: A harmful step can look acceptable if earlier cleaner steps are allowed to hide the drift.

Where you see it: In the evaluator constraint flags, trust-flow story, and runtime-enforcement notes.

Structured Decision Trace #

A recorded path of action, state, obligations, verification, and later review rather than a single opaque verdict.

Plain-English: It is the trace that shows how a result was reached and what still governs it.

Why it matters: DBaD is designed to preserve lineage and reasoning instead of hiding them behind one score.

Where you see it: In the evaluator trace preview, white paper, and trust-flow page.

Effective State #

The governing state that actually controls the current result after immediate and broader constraints are considered.

Plain-English: It is the state that ultimately governs what happens now.

Why it matters: The final governing result may be stricter than the immediate action review alone.

Where you see it: In the evaluator state model panel and trace preview.

Local State #

The immediate action-level evaluation before broader chain, verification, or systemic constraints take over.

Plain-English: It shows what the action looks like on its own.

Why it matters: It helps separate the action’s direct evaluation from later governing overrides.

Where you see it: In the evaluator state model panel and trace preview.

Systemic State #

The broader governing context produced by dependency, contamination, review, or probationary conditions.

Plain-English: It shows what the wider chain context is doing to the result.

Why it matters: Trust problems often come from chain context, not just the immediate step.

Where you see it: In the evaluator state model panel and trace preview.

Boundary Conditions #

Documented limits where DBaD moves from deterministic enforcement into observation and research.

Plain-English: These are the problems DBaD does not claim to have fully solved yet.

Why it matters: The project is stronger when limits are explicit instead of hidden.

Where you see it: On the boundary-conditions page, in the white paper, and throughout public framing.

Zero-Trust Birth #

The rule that a fresh or lineage-free chain does not inherit trust by default.

Plain-English: A new chain starts without borrowed credibility.

Why it matters: It makes chain resets and orphan starts harder to use as trust-laundering shortcuts.

Where you see it: In what-DBaD-solves, the trust-flow explanation, and v2.1 research notes.

Guardrails #

Hard-stop checks (for example consent and catastrophic harm) before weighted scoring.

Plain-English: These are the checks that stop obviously unsafe actions before scoring tries to smooth them over.

Why it matters: Some failures should block or constrain a result even if other dimensions look strong.

Where you see it: In the evaluator process output, dimension review, and methodology material.

E(A) #

Ethical score for action A using weighted inputs.

Legacy scoring-model concept · not central to the current trust-propagation model.

Plain-English: This is the weighted score for one action in the older scoring layer of DBaD.

Why it matters: It remains part of the model, but now supports a broader trust-over-time governance process.

Where you see it: In methodology references and the public evaluator’s scoring section.

H/C/I/P/T #

Harm, Consent, Intent, Proportionality, Transparency.

Plain-English: These are the five supporting ethical dimensions used to review one action.

Why it matters: They make tradeoffs legible instead of hiding them behind vague judgment.

Where you see it: In the evaluator sliders, homepage supporting dimensions, and methodology pages.

Questionable band #

Score range where revision and further review are recommended before action.

Legacy scoring-model concept · supporting term rather than a core trust-propagation concept.

Plain-English: This is the range where the action is still ethically live but not ready for ordinary approval.

Why it matters: It creates room for revision, escalation, and human review before trust moves forward.

Where you see it: In evaluator decision output and older methodology/scoring references.

Stale data #

Cached payload that should be revalidated with ETag before relying on it.

Supporting implementation term · not central to the current trust-propagation model.

Plain-English: It means cached information may no longer be safe to trust without checking again.

Why it matters: DBaD depends on visible, current state rather than hidden or outdated assumptions.

Where you see it: In API and implementation reference material rather than the main public learning path.