Informative UHBS Improvements (No Normative Scoring Change)¶
Status: Informative RFC-style notes · UHBS 4.5.1 unchanged
These improvements are recommended without changing UHQS, weights, δ_C, or letter grades. Normative scoring changes require a separate RFC with corpus evidence, weight sensitivity, backward compatibility, and independent reproduction.
Adopt now (informative)¶
- Evaluation-intent metadata — label reports as asset quality vs program maturity vs campaign vs topology so UHQS is not misused as an org score.
- Engagement-yield metrics — session duration, exchanges, telemetry yield
(via AEP / future informative
report.jsonfields). - Attacker- and defender-cost reporting — time, tokens, infra cost; tarpit delay may be beneficial in AEP while Module E retains latency semantics.
- Cloud / decoy-object coverage map — CLOUDBURST + D3FEND gap analysis for IAM tokens, presigned URLs, K8s secrets, decoy files/credentials.
Experiment (AEP)¶
- Paired decoy vs reference + evaluator control
- FSV across network/protocol/system/state
- DTDR with censoring-aware medians
- EER against declared budgets
- VoD with explicit utility models (never
delta_uhqs) - ICS memory/register consistency probes (lab only)
Map only / out of scope for UHQS¶
CDMM, Cyber-Deception Chain, 7-Criteria procurement checklists, Belief-Scepticism psychology models, game-theoretic placement planners, MTD mutation benchmarks, and topology camouflage systems remain outside generic UHQS.
Validation gate before any UHQS proposal¶
For each candidate metric:
- Construct validity and collection method
- Failure modes and minimum sample size
- Reproducibility across ≥3 profile classes
- Variance and correlation with Modules A–F
- Informative
report.jsonfield first - Separate RFC for any weight/formula change
See ROADMAP for maturity sequencing.