← System Model

Data Quality & Degradation

Definition

Data quality determines how much a system can reliably evaluate, interpret, and act upon.

When quality decreases, the system must degrade gracefully—not compensate with assumptions.

Quality Dimensions

Data quality must be assessed along:

  • Continuity — Are there gaps in the data stream?
  • Resolution — Is the data granular enough for evaluation?
  • Source reliability — Is the data source trustworthy?
  • Context completeness — Is sufficient context available?

Degradation Rules

When quality is insufficient:

  1. Evaluation depth must decrease — Fewer constructs evaluated, wider uncertainty bands
  2. Interpretation scope must narrow — Less specific language, more hedging
  3. Action availability must reduce — Fewer suggestions, more "no action" outputs
  4. Full degradation to silence is permitted — No output is valid

The Degradation Cascade

| Quality Level | Evaluation | Interpretation | Action | |--------------|------------|----------------|--------| | High | Full | Specific | Available | | Medium | Partial | General | Limited | | Low | Minimal | Hedged | None | | Insufficient | None | None | Refused |

Why Degradation Matters

Systems that maintain confidence despite poor data:

  • Mislead users
  • Create false precision
  • Cannot admit uncertainty
  • Eventually fail catastrophically

Systems that degrade gracefully:

  • Match confidence to evidence
  • Preserve user trust
  • Fail safely
  • Maintain long-term reliability

What Degradation Looks Like

High quality data: "Recovery capacity is present. Stress load is moderate. Conditions support light activity."

Medium quality data: "Recovery capacity appears present. Stress load is elevated. Limited confidence in current assessment."

Low quality data: "Insufficient data continuity. Current state uncertain. No action recommended."

Insufficient data: [No output]

Compliance Note

Systems that maintain high confidence despite low data quality, or that compensate for poor data with assumptions, violate ESGR System Model specifications.

Related