Skip to content

Substance Ranking — Content Quality Over Popularity

Substance Ranking

TL;DR: A content evaluation method that scores posts and comments based on evidence quality, specificity, and actionability — rather than popularity metrics like upvotes, likes, or engagement. Key insight: popularity measures agreement, not truth.

The Problem It Solves

Social media ranking algorithms optimize for engagement:

  • Reddit upvotes measure “I agree” or “this is funny”
  • Twitter likes measure “I want to signal I saw this”
  • LinkedIn engagement measures “this person is in my network”

None of these measure:

  • Is this claim accurate?
  • Is this advice actionable?
  • Does this have evidence?

Result: Popular content and useful content are different populations with partial overlap.

How It Works

Substance ranking evaluates content on multiple axes independent of engagement:

The Substance Scale (0-3)

ScoreMeaningExample
0Pure sentiment”This!”, emoji reactions, jokes
1General opinion”You should probably consider…“
2Specific claim with reasoning”When I tried X, it failed because…“
3Specific + evidence/numbers/lived experience”$4,200 cost, 6 weeks, saved $18k over 3 years”

Additional Axes

  • Source Type: First-hand experience > professional expertise > second-hand report > inference > pure sentiment
  • Actionability: Can the reader do, decide, or change their model?
  • Contrarian Bonus: Downvoted but reasoned? Often signal popularity-sort missed
  • Red Flags: Credential theater, gish-gallop, edited-after-voting, ideology without facts

Real-World Example

Reddit thread on co-founder revenue splits:

CommentUpvotesSubstance Score
”What does your operating agreement say?”+51 (generic advice)
“The real question is whether your role is retainer-style or project-based. Legal default is 50/50, but…”+13 (framework + specifics)

Popularity sort: Shows the +5 comment first Substance sort: Shows the +1 comment first

The buried comment was actually more useful.

When to Apply

Good Fit

  • Research synthesis (finding what’s true, not what’s popular)
  • Content creation (publishing insights, not popular opinions)
  • Competitive intelligence (what practitioners actually do)
  • Due diligence (evidence-based assessment)

Poor Fit

  • Trend detection (popularity IS the signal)
  • Viral content creation (engagement is the goal)
  • Community sentiment analysis (agreement is informative)

Relationship to Other Concepts

ConceptWhat It MeasuresRelationship
UpvotesAgreement/entertainmentOrthogonal — popularity ≠ substance
E-E-A-TAuthor expertise signalsComplementary — substance is content-level E-E-A-T
Fact-checkingClaim accuracySubset — substance includes accuracy + actionability
glossary/geo-aeoAI citation optimizationRelated — high-substance content gets cited more

Implementation

The tools/reddit-thread-analyzer implements substance ranking for Reddit threads:

  • 6-axis rubric applied to every comment
  • ~30% divergence from popularity sort typical
  • Building blocks (numbers, frameworks, case studies) extracted from high-substance comments

Key Takeaways

  • Popularity metrics measure agreement, not truth or usefulness
  • Substance ranking evaluates evidence quality, specificity, and actionability
  • Typical divergence from popularity sort: ~30%
  • High-substance content is more likely to be cited by AI systems
  • Best applied to research and content creation, not trend detection