Last updated: 2026-05-04 UTC

Transparent downstream scoring demo built on audited and synthetic DBaD trace cases

Advisory demo - you be the judge

Can a system score something as "perfect"... even when it clearly isn't?

Explore how DBaD signals can be interpreted into an advisory score. The score is not truth. You decide what it means.

Pick a scenario and see the score How this works

You don't need to understand this. Just click around.

Not DBaD validation Not proof of truth Interpretation only Profiles = advisory lenses

Start with a scenario, then compare how different advisory profiles read the same visible DBaD trace. This page is running presentation hardening v0.3.2 over DecencyMeter v0.1 baseline and experimental v0.2 score models.

85-95

Advisory Score Range

Interpretation varies by profile.

Start here - choose a scenario

These are demonstration cases. They are not real-world events.

Audit case

Audit Case

A realistic, mixed-quality response

See how documented trade-offs and limits are interpreted.

Open audit case Stress case

Stress Case

Multiple red flags and missing data

See how severe issues and omissions affect the advisory score.

Open stress case Synthetic anomaly

"Perfect" Incident

This one scores 100... but feels wrong.

A synthetic anomaly designed to challenge your intuition.

Open perfect incident

These are demonstration cases. They are not real-world events.

Quick access

Jump directly to the adjacent critique and context surfaces without hunting through the nav.

DecencyMeter Bridge How it connects to DBaD Scoring Anomalies See edge cases and quirks Pressure Tests Synthetic stress scenarios Top Issues Known limitations Try to Break DBaD Report what you find

Important Boundary Notice

These warnings still lead the interpretation. The layout is friendlier now, but the boundaries did not get softer.

Not DBaD validation

DBaD validates structure, not ethics or truth.

Not proof of truth

This does not prove correctness, truth, or ethical behavior.

Interpretation of signals

This is an interpretation of recorded signals, not a factual verdict.

Profiles are advisory

Profiles are lenses over the same trace, not protocol changes.

Experimental v0.2

Experimental v0.2 is experimental and not DBaD validation.

No proof of goodness

v0.2 does not prove truth, safety, goodness, or correctness.

Framework and Version Alignment

API and framework boundary: DBaD emits the trace and validation context; DecencyMeter interprets that trace downstream; the v0.3.x layer constrains how the score is presented.

DBaD v2.2 runtime framework

DBaD validates trace structure and recorded integrity signals. It does not score ethics, truth, goodness, or real-world outcomes.

DecencyMeter v0.1 baseline score model

The current deterministic advisory score model shown on this page.

DecencyMeter experimental v0.2 score model remains a comparison layer only: A side-by-side comparison model that adds outcome-aware caution signals without changing DBaD validation.

Presentation hardening v0.3.2

This page binds warning hierarchy, profile, version, case context, and portable citation directly to every public score surface.

If you quote this, include this.

Portable Citation Object: Portable citation object: a reusable advisory notice designed to travel with screenshots, quotes, press references, decks, emails, and regulator or journalist excerpts. It is not a legal disclaimer or enforcement tool; it is a presentation-discipline tool that makes omission of context visibly dishonest.

Short Notice

DecencyMeter Advisory Notice: Advisory only. Not DBaD validation. Not proof of ethical behavior, truth, or outcome quality. Profile, version, and context matter.

Current Surface Notice

DecencyMeter Advisory Notice: This is a procedural trace-interpretation score only. It does not verify truth, ethics, harm, safety, or real-world outcomes. Scores vary by profile and version and must not be presented as proof of ethical behavior.

Experimental Notice

DecencyMeter Advisory Notice: This experimental DecencyMeter output is advisory only. It is not DBaD validation and not proof of ethical behavior, truth, or outcome quality. Profile, version, and case context matter.

Observation Check

System is now in observation phase. Use these lightweight links to flag interpretation reactions without changing scoring, DBaD validation, or stored traces.

Was this helpful?

Yes No

Confusing?

Yes No

Misleading?

Yes No

Profile Selector

Choose a visible advisory scoring profile for the selected trace case. Missing or invalid profile values default to baseline.

Baseline Strict Audit Operational Tolerance Humanitarian / Common Sense Experimental

Active case

Runtime-audited trace

The runtime-audited DBaD v2.2 trace used as the default proof case.

Selected profile

Operational Tolerance

Profile key: operational_tolerance

More pragmatic deductions for bounded incompleteness while still penalizing visible governance drift.

Score Models: v0.1 Baseline vs Experimental v0.2

This comparison keeps DecencyMeter v0.1 baseline score model visible as the current score model and adds DecencyMeter experimental v0.2 score model as a side-by-side downstream comparison only.

Direct-link boundary: every score card below carries its own advisory context so it can still resist overread when linked or screenshotted on its own.

Quote rule: copy one of the portable citation notices above rather than quoting a selected score by itself.

Before/after delta: +0 points for the selected case/profile.

DecencyMeter v0.1 baseline score model

ADVISORY ONLY — NOT DBaD VALIDATION — NOT PROOF OF ETHICAL BEHAVIOR

Advisory only. Not DBaD validation. Not proof of ethical behavior, truth, or outcome quality.

Visible profile range: 85-95/100

Profile: operational_tolerance (current baseline)

Case: Runtime-audited trace

Runtime-audited trace. Audited trace context only.

Selected score: 92/100

Total deductions: 8

DecencyMeter experimental v0.2 score model

ADVISORY ONLY — NOT DBaD VALIDATION — NOT PROOF OF ETHICAL BEHAVIOR

Advisory only. Not DBaD validation. Not proof of ethical behavior, truth, or outcome quality.

Visible profile range: 85-95/100

Profile: operational_tolerance (experimental comparison)

Case: Runtime-audited trace

Experimental comparison only. Runtime-audited trace context.

Selected score: 92/100

Before/after delta: +0

New v0.2-only deductions: 0

Experimental v0.2 is not DBaD validation. It is a transparent comparison model informed by the documented perfect-looking incident trace anomaly.

The perfect-looking incident trace `trc_20260428193300_pf67c1e2` motivated this experimental comparison and remains documented in the pressure-test results.

Read the scoring anomaly explainer for the public explanation of why v0.2 exists and why the issue belongs to DecencyMeter rather than DBaD. The broader synthetic suite is also available in the pressure-test catalogue.

Profile Comparison (Baseline Score Model)

Profile comparison shows different advisory interpretations over the same DBaD trace under DecencyMeter v0.1 baseline score model. It is not DBaD validation.

Visible range: 85 to 95 with spread 10. The highest score is shown only as part of the range, not as the dominant claim.

Portable citation rule: quote the range and advisory notice first; selected profile scores are supporting detail only.

Profile name Score Total deductions Triggered deduction count View
Baseline
ADVISORY ONLY — NOT DBaD VALIDATION — NOT PROOF OF ETHICAL BEHAVIOR
Visible profile range: 85-95/100
Profile: baseline
Case: Runtime-audited trace
Runtime-audited trace.
Selected score: 90/100
 10 1 View profile
Strict Audit
ADVISORY ONLY — NOT DBaD VALIDATION — NOT PROOF OF ETHICAL BEHAVIOR
Visible profile range: 85-95/100
Profile: strict_audit
Case: Runtime-audited trace
Runtime-audited trace.
Selected score: 85/100
 15 1 View profile
Operational Tolerance selected
ADVISORY ONLY — NOT DBaD VALIDATION — NOT PROOF OF ETHICAL BEHAVIOR
Visible profile range: 85-95/100
Profile: operational_tolerance
Case: Runtime-audited trace
Runtime-audited trace.
Selected score: 92/100
 8 1 View profile
Humanitarian / Common Sense
ADVISORY ONLY — NOT DBaD VALIDATION — NOT PROOF OF ETHICAL BEHAVIOR
Visible profile range: 85-95/100
Profile: humanitarian_common_sense
Case: Runtime-audited trace
Runtime-audited trace.
Selected score: 92/100
 8 1 View profile
Experimental
ADVISORY ONLY — NOT DBaD VALIDATION — NOT PROOF OF ETHICAL BEHAVIOR
Visible profile range: 85-95/100
Profile: experimental
Case: Runtime-audited trace
Runtime-audited trace.
Selected score: 95/100
 5 1 View profile

Why profiles differ

  • Each profile applies different visible weights to the same recorded DBaD signals.
  • A strict audit profile penalizes missing evidence and unresolved governance signals more heavily.
  • An operational tolerance profile penalizes the same signals less heavily.
  • A humanitarian/common-sense profile weighs unresolved harm-facing or closure signals more heavily.
  • The experimental profile exists to test alternative interpretations, not to define the default.

Profile differences are interpretive differences, not DBaD validation differences.

Inputs Used

These are the visible DBaD fields used by this first deterministic advisory model.

validation result

Value: valid

Field: validation.is_valid

Trace passed deterministic DBaD validation, but one or more state transitions do not yet have recorded evidence references.

expected_outcome vs outcome_status

Value: upheld vs upheld

Field: expected_outcome.label / outcome_status

No mismatch in this audited trace.

declared_blind_spots

Value: 1 total / 0 open

Field: declared_blind_spots[]

Only open blind spots trigger deductions in this demo model.

state_transition_evidence presence

Value: recorded

Field: transition_history[].state_transition_evidence

Missing evidence advisory notes trigger a deduction even if another transition has evidence.

escalation_closure status

Value: approved_to_continue (recorded)

Field: escalation_closure / escalation_closure_status

This demo model only deducts when escalation context exists and closure is not approved_to_continue.

completeness_attestation

Value: declared_complete

Field: completeness_attestation.status

Anything other than declared_complete triggers a deduction in this demo model.

outcome_status

Value: upheld

Field: outcome_status

Displayed as a visible downstream signal, not as truth or correctness.

trace_version/history depth

Value: v11 / 9 history entries

Field: trace_version / history arrays

Shown for context only in this first demo. No deduction is attached.

Simple Scoring Model

This demo uses a transparent, deterministic rule sheet for the selected advisory profile. No ML. No hidden weights.

Model start

Start at: 100

Every deduction below is visible and rule-based.

expected_outcome mismatch

Deduction: -15

Subtract when expected_outcome and outcome_status differ after outcome is known.

missing evidence

Deduction: -8

Subtract when a supported transition is missing recorded state_transition_evidence.

open blind spots

Deduction: -3

Subtract the listed amount for each declared blind spot still marked open.

completeness not declared_complete

Deduction: -7

Subtract when completeness_attestation.status is not declared_complete.

escalation closure not approved_to_continue

Deduction: -10

Subtract when escalation context exists and disposition is not approved_to_continue.

Experimental v0.2 Additions

These are the new visible experimental v0.2 deductions added on top of the current v0.1 baseline. The perfect-looking incident trace is now a public selector case so the anomaly can be inspected directly.

outcome status incident

Deduction: -6

Subtract when outcome_status is incident, even if the surrounding process is otherwise clean.

outcome status reversed

Deduction: -8

Subtract when outcome_status is reversed.

outcome status escalated

Deduction: -5

Subtract when outcome_status is escalated.

incident plus approved_to_continue caution

Deduction: -4

Subtract when an incident outcome coexists with escalation closure approved_to_continue.

incident plus no-blind-spots confidence caution

Deduction: -3

Subtract when an incident outcome coexists with no blind spots and declared_complete completeness.

over-verification / minimal-evidence caution

Deduction: -4

Subtract when verification history is heavy while a supported transition still lacks evidence.

Score (Baseline Score Model v0.1)

ADVISORY ONLY — NOT DBaD VALIDATION — NOT PROOF OF ETHICAL BEHAVIOR

DecencyMeter Advisory Notice: This is a procedural trace-interpretation score only. It does not verify truth, ethics, harm, safety, or real-world outcomes. Scores vary by profile and version and must not be presented as proof of ethical behavior.

Visible profile range: 85-95/100

Selected profile: Operational Tolerance

Case: Runtime-audited trace

Active trace: trc_20260428181140_42396240. Runtime-audited trace, still advisory interpretation only.

Selected score: 92/100

  • -8 At least one supported transition is missing a recorded evidence reference, even though another transition may still have evidence.

Score (Experimental v0.2 Comparison Model)

ADVISORY ONLY — NOT DBaD VALIDATION — NOT PROOF OF ETHICAL BEHAVIOR

DecencyMeter Advisory Notice: This experimental DecencyMeter output is advisory only. It is not DBaD validation and not proof of ethical behavior, truth, or outcome quality. Profile, version, and case context matter.

Visible profile range: 85-95/100

Selected profile: Operational Tolerance

Case: Runtime-audited trace

Experimental comparison only. Runtime-audited trace, still advisory interpretation only.

Selected score: 92/100

Before/after delta: +0

No new experimental v0.2-only deductions were triggered for this selected case/profile.

Experimental v0.2 is experimental. It does not prove truth, safety, goodness, or correctness.

Explainability (Baseline Score Model v0.1)

Every deduction shows the triggering signal and the DBaD field it came from.

missing evidence

What triggered it: At least one supported transition is missing a recorded evidence reference, even though another transition may still have evidence.

DBaD field: transition_history[].state_transition_evidence

Deduction: -8

This is not DBaD validation. It is a downstream advisory interpretation.

Experimental v0.2 Explainability

These are the exact new v0.2-only deductions and the recorded DBaD fields they came from. This is not DBaD validation.

No experimental v0.2-only deductions were triggered for this selected case/profile, so the before/after delta is +0.

Challenge this score

This score is based on explicit weights and recorded trace signals.

You are encouraged to challenge the assumptions, weights, and interpretation.

Current weights

  • mismatch → -15
  • missing_evidence → -8
  • open_blind_spot_each → -3 each
  • completeness_not_declared → -7
  • escalation_not_approved → -10

Submit a scoring critique

A disagreement with this score does not invalidate DBaD validation.

Submit a scoring critique Challenge DBaD separately

Experimental v0.2 Comparison Weights

Visible v0.2 comparison weights only. No hidden scoring. No database writes.

  • outcome_status_incident → -6
  • outcome_status_reversed → -8
  • outcome_status_escalated → -5
  • incident_approved_to_continue → -4
  • incident_no_blind_spots_declared_complete → -3
  • over_verified_minimal_evidence → -4

Trace Context

Validation result

valid

Trace passed deterministic DBaD validation, but one or more state transitions do not yet have recorded evidence references.

Advisory notes

  • missing_state_transition_evidence

Violations

none

Links Back

v2.2 demo · Top issues · DecencyMeter bridge · Scoring anomalies · Pressure tests · Try to break DBaD