Authority Scoring — {DOMAIN_DESCRIPTION}

Formula

authority_score = (
    source_type_weight     * 0.35
  + recency_weight         * 0.25
  + community_trust_weight * 0.20
  + specificity_weight     * 0.20
)

The LLM MUST use this formula. Do NOT editorialize on source quality — compute the score.

Source Type Weight (0.35)

Adapt this table to your domain. Higher weights for authoritative sources, lower for informal or unreviewed sources.

Source Type Weight
{PRIMARY_AUTHORITATIVE_SOURCE} (e.g., official docs, peer-reviewed journals) 1.00
{SECONDARY_AUTHORITATIVE_SOURCE} (e.g., official blogs, technical reports) 0.85
{VERIFIED_COMMUNITY_SOURCE} (e.g., verified modules, accepted SO answers) 0.70
{EXPERT_INFORMAL_SOURCE} (e.g., known expert blog, conference talk) 0.50
{UNVERIFIED_SOURCE} (e.g., unknown blog, forum post) 0.30
{AI_GENERATED_SOURCE} (e.g., LLM output without citation) 0.20

Recency Weight (0.25)

Adjust the decay rate for your domain. Fast-moving fields (cloud APIs, ML frameworks) decay faster. Slow-moving fields (mathematics, established medicine) decay slower.

recency_weight = max(0.20, 1.0 - {DECAY_RATE} * (current_year - last_updated_year))
Domain Speed Decay Rate Rationale
Fast (cloud, APIs, frameworks) 0.10 Features change yearly
Medium (software practices, guidelines) 0.07 Standards evolve over 2-5 years
Slow (established science, regulations) 0.05 Knowledge is stable over decades

Your domain decay rate: {DECAY_RATE}

Last Updated Weight (in 2026, rate={DECAY_RATE})
2026 1.00
2025 {COMPUTED}
2023 {COMPUTED}
2020 {COMPUTED}

Community Trust Weight (0.20)

Indicator Weight
{HIGHEST_TRUST_INDICATOR} (e.g., domain governing body, core team member) 1.00
{HIGH_TRUST_INDICATOR} (e.g., recognized expert, certified professional) 0.85
{MEDIUM_TRUST_INDICATOR} (e.g., active community contributor) 0.60
{LOW_TRUST_INDICATOR} (e.g., anonymous contributor) 0.30

Specificity Weight (0.20)

Specificity Weight
Exact scenario match 1.00
Close match (same subfield) 0.70
General topic match 0.40

Example Computation

Source: {EXAMPLE_SOURCE_DESCRIPTION}

source_type   = {VALUE} * 0.35 = {PRODUCT}
recency       = {VALUE} * 0.25 = {PRODUCT}
community     = {VALUE} * 0.20 = {PRODUCT}
specificity   = {VALUE} * 0.20 = {PRODUCT}
                                  -----
authority_score                 = {TOTAL}

LLM Boundary

Action Allowed?
Compute authority_score using formula YES — deterministic
Override formula weights NO — user must approve changes
Skip scoring for “obvious” sources NO — always compute
Report score without showing formula inputs NO — show the breakdown