Part 2 of 5: Debate Reform Series

The thesis: If we systematically link arguments to evidence (like Google PageRank links web pages), then when evidence changes, all dependent conclusions would automatically update. This single structural change would transform debate quality.

The Core Problem: Zombie Arguments That Won't Die

When Foundations Crumble, Buildings Should Fall

How reasoning should work:

A peer-reviewed study claims X → Arguments cite this study → Conclusions rest on those arguments → Policy decisions follow

When the study gets retracted:

• Arguments citing it should weaken
• Conclusions resting on those arguments should weaken
• Policy decisions should be reconsidered
• The entire chain should update

What actually happens:

Study gets retracted → Arguments still cite it → Conclusions persist unchanged → Policy continues → Nothing updates

This is the fundamental flaw in how we debate everything.

Real-World Disaster: The Wakefield Fraud

1998: Foundation laid

• Lancet publishes Wakefield study: "Vaccines may cause autism"
• Evidence Quality at time: Tier 1 (peer-reviewed journal)
• Argument Score: High (published in prestigious journal)

Arguments that formed:

1. "Vaccines cause autism" [cited Wakefield]
2. "MMR vaccine is dangerous" [cited Argument 1]
3. "Parents should refuse vaccination" [cited Argument 2]
4. "Vaccine mandates are tyranny" [cited Argument 3]
5. "Medical establishment covers up vaccine dangers" [cited Arguments 1-4]

2004-2010: Foundation demolished

• Co-authors retract support
• Fraud investigation reveals data manipulation
• Lancet fully retracts paper
• Wakefield loses medical license
• Multiple studies find no link (evidence now overwhelming)
• Evidence Quality: Tier 0 (fraudulent)

What should have happened:

Wakefield Study: Tier 1 → Tier 0 (retracted, fraudulent)
├─ "Vaccines cause autism": Score 85 → 15 (foundation removed)
│   ├─ "MMR vaccine dangerous": Score 80 → 20 (supporting argument collapsed)
│   │   └─ "Refuse vaccination": Score 75 → 10 (chain broken)
│   │       └─ "Mandates are tyranny": Score 70 → 15 (upstream arguments failed)
│   └─ "Medical coverup": Score 65 → 5 (conspiracy theory unsupported)
└─ AUTOMATIC UPDATE ACROSS ALL CITATIONS

What actually happened:

The arguments persist unchanged. People still cite Wakefield. Anti-vaccine movement stronger than ever. Measles outbreaks return.

Why? Because there's no systematic link between evidence and conclusions.

The Solution: PageRank for Arguments

What Google Did for Web Pages

The PageRank insight: You can determine quality by analyzing link structure.

How it works:

• Page A links to Page B
• If Page A is high-quality, Page B probably is too
• If many high-quality pages link to Page B, Page B is probably high-quality
• Quality propagates through the network
• No central authority needed

Why it works:

• Can't fake being linked by quality sources
• Gaming requires gaming entire network
• Link structure reveals consensus
• Automatic, algorithmic, scalable

Why it revolutionized search:

• Before: Keyword matching (easily gamed)
• After: Link analysis (hard to game)
• Quality emerged from structure

What We're Doing for Arguments

The ISE insight: You can determine argument quality by analyzing evidence structure.

How it works:

• Evidence A supports Argument B
• If Evidence A is high-quality (Tier 1), Argument B gets stronger
• If many high-quality sources support Argument B, it's probably strong
• Quality propagates through the network
• No central authority needed

Why it works:

• Can't fake being supported by Tier 1 evidence
• Gaming requires gaming entire evidence base
• Evidence structure reveals strength
• Automatic, algorithmic, scalable

Why this revolutionizes debate:

• Before: Disconnected claims (zombie arguments persist)
• After: Linked evidence (updates propagate automatically)
• Truth emerges from structure

How Recursive Scoring Works

The Basic Formula

Every argument gets a score (0-100) based on:

Argument Score = (Evidence Quality × Logical Validity × Linkage Strength) + Expert Weighting

For arguments citing other arguments (recursive):

Parent Argument Score = Base Score + 
  Weighted_Average(Supporting Sub-Arguments) - 
  Weighted_Average(Contradicting Sub-Arguments)

The magic: Scores propagate up and down the tree automatically.

Example 1: Simple Evidence → Argument

Claim: "Electric cars reduce carbon emissions"

Evidence cited:

• Study A: "EVs have 50% lower lifetime emissions" [Tier 1: Peer-reviewed]
• Study B: "Depends on electricity grid mix" [Tier 1: Peer-reviewed]

Scoring:

Study A contribution:

• Evidence Quality: 90 (Tier 1, multiple replications)
• Logical Validity: 95 (sound methodology)
• Linkage Strength: 95 (directly proves claim)
• Contribution: 90 × 0.95 × 0.95 = 81 points

Study B contribution:

• Evidence Quality: 90 (Tier 1, peer-reviewed)
• Logical Validity: 95 (sound methodology)
• Linkage Strength: 70 (qualifies claim but doesn't refute)
• Contribution: 90 × 0.95 × 0.70 = 60 points

Argument Score: (81 + 60) / 2 = 70.5

Interpretation: Strong support, but with important caveat about grid mix.

Example 2: Multi-Level Recursive Scoring

Top-Level Claim: "We should implement carbon tax"

Supporting Arguments (Level 1):

A1. "Carbon tax reduces emissions" [Score: 82]
A2. "Carbon tax can be revenue-neutral" [Score: 78]
A3. "Carbon tax economically efficient" [Score: 75]

Contradicting Arguments (Level 1):

C1. "Carbon tax harms poor families" [Score: 55]
C2. "Carbon tax reduces competitiveness" [Score: 48]

Each Level 1 argument has its own supporting evidence:

A1: "Carbon tax reduces emissions"

• Evidence: British Columbia carbon tax data [Tier 1: Government statistics]
• 5-15% emission reduction documented
• Score contribution: +85
• Evidence: Sweden carbon tax history [Tier 1: Academic study]
• 25% reduction over 30 years
• Score contribution: +80
• A1 Final Score: 82

A2: "Carbon tax can be revenue-neutral"

• Evidence: BC model returned revenue via tax cuts [Tier 1: Official data]
• Score contribution: +80
• Evidence: Baker-Shultz proposal framework [Tier 2: Think tank analysis]
• Score contribution: +75
• A2 Final Score: 78

C1: "Carbon tax harms poor families"

• Evidence: Energy costs are higher % of poor budgets [Tier 1: Census data]
• Score contribution: +70
• Counter-evidence: Revenue can be rebated progressively [Tier 1: BC data]
• Score contribution: -60
• C1 Final Score: 55 (concern valid but addressable)

Top-Level Score Calculation:

Base Score: 60 (policy proposal baseline)
Supporting: (82 + 78 + 75) / 3 = 78.3 → +18 boost
Contradicting: (55 + 48) / 2 = 51.5 → -8 penalty

Final Score: 60 + 18 - 8 = 70

Interpretation: Strong evidence support, concerns exist but are addressable with design choices.

What Happens When Evidence Changes

Scenario: New study contradicts BC emission reduction claims

New Evidence: "Reanalysis shows BC reduction was 2-5%, not 5-15%" [Tier 1: Peer-reviewed]

Automatic Updates:

Original BC Evidence: 85 → 70 (claim weakened, not eliminated)
  ↓ affects
A1 "Carbon tax reduces emissions": 82 → 76 (foundation weakened)
  ↓ affects
Top-Level "Implement carbon tax": 70 → 67 (supporting argument weakened)

The entire chain recalculates automatically.

What users see:

• Notification: "A1 score changed due to updated evidence"
• Explanation: "BC emission reduction data revised downward"
• New scores visible with justification
• Version history shows what changed when
• Debate continues with updated foundation

No zombie arguments. No persistent myths. Just automatic updates.

Real Example with Full Scoring Tree: Minimum Wage

Top-Level Claim: "Raising minimum wage to $15/hour helps poor workers"

Level 1: Supporting Arguments

S1. "Wage increases boost worker income"

• Evidence 1: Seattle study shows average $10/week gain [Tier 1: University research]
• Quality: 85, Logic: 90, Linkage: 95 → Contribution: +73
• Evidence 2: CBO analysis confirms income gains [Tier 1: Government analysis]
• Quality: 90, Logic: 90, Linkage: 90 → Contribution: +73
• S1 Score: 73

S2. "Reduced turnover saves employer costs"

• Evidence 1: Meta-analysis shows 25% turnover reduction [Tier 1: Academic]
• Quality: 85, Logic: 85, Linkage: 80 → Contribution: +58
• Evidence 2: Turnover cost = 20% of annual salary [Tier 2: Industry data]
• Quality: 75, Logic: 80, Linkage: 85 → Contribution: +51
• S2 Score: 55

S3. "Multiplier effects boost local economy"

• Evidence 1: Low-wage workers spend increases locally [Tier 2: Economic analysis]
• Quality: 70, Logic: 75, Linkage: 65 → Contribution: +34
• S3 Score: 34 (weaker support)

Level 1: Contradicting Arguments

C1. "Job losses offset wage gains"

• Evidence 1: CBO estimates 1.3M job losses at $15 [Tier 1: Government analysis]
• Quality: 90, Logic: 85, Linkage: 90 → Contribution: +69
• Evidence 2: Seattle study found hours reduction [Tier 1: University research]
• Quality: 85, Logic: 80, Linkage: 85 → Contribution: +58
• Counter: Net benefit still positive for most workers [Tier 2: Analysis]
• Reduces impact by -20
• C1 Score: 54 (concern valid but net impact debated)

C2. "Automation accelerates"

• Evidence 1: Kiosks replace cashiers [Tier 3: Observational]
• Quality: 60, Logic: 70, Linkage: 50 → Contribution: +21
• Evidence 2: Long-term trend regardless of wage [Tier 2: Economic analysis]
• Quality: 75, Logic: 80, Linkage: 60 → Contribution: +36 (weakens concern)
• C2 Score: 28 (weak support for causal link)

Top-Level Calculation

Base Score: 50 (neutral starting point)
Supporting Average: (73 + 55 + 34) / 3 = 54 → +20 boost
Contradicting Average: (54 + 28) / 2 = 41 → -10 penalty

Final Score: 50 + 20 - 10 = 60

Interpretation: Modest positive support. Benefits exceed costs for most workers, but trade-offs exist (some job losses, some hours reductions). Net effect positive but not overwhelming.

What Happens If Seattle Study Gets Challenged

Scenario: New paper criticizes Seattle study methodology

The Challenge:

• New Evidence: "Seattle study's methodology flawed" [Tier 1: Peer-reviewed critique]
• Points out selection bias in sample
• Questions wage gain calculation
• Original researchers respond with defense [Tier 2: Response paper]

Structured Debate:

1. Challenge gets posted as counter-argument to Seattle evidence
2. Community evaluates challenge reasoning
3. If challenge has merit, Seattle evidence score drops
4. Drop propagates: S1 score drops, C1's counter-argument strengthens
5. Top-level score adjusts automatically
6. Users notified of change with full reasoning

Possible Outcomes:

If challenge succeeds:

Seattle Evidence: 85 → 60 (methodology questioned)
S1 "Wage increases boost income": 73 → 65 (foundation weakened)
C1 "Job losses offset gains": 54 → 58 (counter-argument weakens)
Top-Level Score: 60 → 56 (net effect: weaker support)

If challenge fails:

Seattle Evidence: 85 → 88 (survived scrutiny, strengthened)
S1 Score: 73 → 75 (robust foundation)
Top-Level Score: 60 → 61 (slightly stronger)

Either way, the chain updates automatically based on which argument wins.

Linkage Scores: Relevance Matters

The Problem: High-Quality But Irrelevant

Scenario: Argument about electric car emissions

Evidence cited: "Study shows electric cars have lower maintenance costs"

Analysis:

• Evidence Quality: 90 (Tier 1, well-documented)
• Logical Validity: 95 (sound research)
• Linkage Strength: 20 (maintenance ≠ emissions)

Contribution to argument: 90 × 0.95 × 0.20 = 17 points

Why linkage matters: Can't boost an argument about emissions by citing quality evidence about maintenance. The connection must be direct.

Linkage Strength Scale (0-100)

100 - Direct proof/disproof:

• Claim: "This drug reduces blood pressure"
• Evidence: "RCT shows 15mmHg average reduction"
• Connection: Direct measurement of exact claim

80 - Strong implication (one inference step):

• Claim: "Carbon tax reduces emissions"
• Evidence: "BC carbon tax correlated with 5-15% emission drop"
• Connection: Strong correlation, one inference (causation likely)

50 - Moderate support (multiple steps):

• Claim: "Universal healthcare reduces costs"
• Evidence: "Single-payer has lower administrative costs"
• Connection: Administrative costs are one component, multiple steps to total cost

20 - Weak/tangential:

• Claim: "Electric cars reduce emissions"
• Evidence: "Electric cars have lower maintenance costs"
• Connection: Maintenance ≠ emissions (tangentially related to TCO)

0 - Irrelevant:

• Claim: "Minimum wage should be $15"
• Evidence: "The sky is blue"
• Connection: None

Why This Prevents Gish Gallop

Gish Gallop: Overwhelming opponent with volume of loosely-related sources.

Example without linkage scores:

Claim: "Vaccines cause autism"

Evidence cited (50 sources):

• Study about different vaccines
• Study about different age groups
• Study about different conditions
• Blog posts
• Anecdotes
• Correlations

Without linkage: Looks impressive (50 sources!)

With linkage:

Source 1: Quality 90, Linkage 30 → Contribution 27
Source 2: Quality 85, Linkage 25 → Contribution 21
Source 3: Quality 40, Linkage 40 → Contribution 16
...
Average contribution: 22 points

vs.

Single high-quality directly relevant source:
Quality 95, Linkage 95 → Contribution 90

Result: One directly relevant high-quality source beats 50 tangentially-related sources.

Gish gallop defeated by structure.

Why This Prevents Gaming

Gaming Attempt 1: Create Fake Evidence

Strategy: Publish fake "peer-reviewed" study in predatory journal.

Why it fails:

• Evidence tier classification catches predatory journals (Tier 4, not Tier 1)
• Low quality score = low contribution to argument
• Community can flag and downgrade
• Track record: if multiple papers from this journal fail scrutiny, auto-downgrade

Example:

Fake Study in Predatory Journal:
- Claims Tier 1 status
- Community flags: "Journal not indexed, no peer review"
- Tier 1 → Tier 4
- Score: 90 → 25
- Contribution to argument: Minimal

Gaming Attempt 2: Volume of Weak Sources

Strategy: Cite 100 blog posts supporting claim.

Why it fails:

• Blog posts = Tier 4 = low individual scores
• Linkage strength likely weak (not directly relevant)
• Average of 100 weak sources < 1 strong source

Example:

100 blog posts (Tier 4):
- Average quality: 30
- Average linkage: 40
- Average contribution: 30 × 0.40 = 12 points each
- Total: 100 × 12 = 1,200 points / 100 sources = 12 average

vs.

1 peer-reviewed study (Tier 1):
- Quality: 90
- Linkage: 90
- Contribution: 90 × 0.90 = 81 points

One quality source beats 100 weak sources.

Gaming Attempt 3: Circular Citations

Strategy:

• Create Argument A citing Argument B
• Create Argument B citing Argument A
• Both look supported!

Why it fails:

• System detects circular references
• Requires evidence at bottom (can't be infinite regression)
• Circular arguments get flagged and penalized
• PageRank handles this (same solution)

Example:

Argument A: "X is true" [cites Argument B]
Argument B: "X is supported" [cites Argument A]

System detects: No external evidence, circular reasoning
Flag: "Circular citation detected"
Both scores → Low (no independent foundation)

Gaming Attempt 4: Cherry-Pick Studies

Strategy: Cite only favorable studies, ignore contradicting ones.

Why it fails:

• System flags when higher-tier contradicting evidence exists but isn't addressed
• Community can add contradicting evidence
• Explicit display: "This argument cites 3 supporting studies but ignores 10 contradicting studies"
• Cherry-picking visible and penalized

Example:

Argument: "Homeopathy works"
- Cites: 2 positive studies [Tier 3]
- Ignores: 50 negative studies [Tier 1]

System flags: "High-quality contradicting evidence not addressed"
Display: "2 Tier 3 supporting vs. 50 Tier 1 contradicting"
Argument score: Low (evidence asymmetry obvious)

Integration with Other Components

Connection to Logical Validity

From Part 3: Grouping Similar Arguments and Detecting Fallacies:

Logical validity is a separate multiplier:

Argument Score = (Evidence Quality × Logical Validity × Linkage Strength)

Even with great evidence, logical fallacy reduces score:

Example:

• Evidence Quality: 90 (Tier 1 studies)
• Linkage: 90 (directly relevant)
• Logical Validity: 40 (commits slippery slope fallacy)
• Final: 90 × 0.40 × 0.90 = 32 points

Good evidence + bad logic = weak argument

Connection to Evidence Tiers

From Part 4: Evidence Tiers and Objective Criteria:

Evidence tier determines base quality score:

• Tier 1 (peer-reviewed): Base 85-95
• Tier 2 (expert analysis): Base 70-80
• Tier 3 (journalism): Base 50-65
• Tier 4 (opinion/anecdote): Base 20-40

Then linkage and logic multiply that base:

Tier 1 study (base 90) × strong linkage (0.90) × good logic (0.95) = 77
Tier 4 blog (base 30) × weak linkage (0.30) × good logic (0.90) = 8

Structure makes quality difference explicit and measurable.

Connection to One Page Per Topic

From Part 3: Grouping Similar Arguments:

Why consolidation matters for scoring:

Without One Page Per Topic:

• Same argument on 50 platforms
• Evidence scattered across all 50
• Updates don't propagate
• No cumulative intelligence

With One Page Per Topic:

• All arguments consolidated
• All evidence accumulates in one place
• Updates propagate through entire tree
• Quality emerges from comprehensive analysis

Linking only works if arguments are consolidated.

Why Automatic Updates Matter

Problem: Human Updating Doesn't Scale

Current approach: Manual correction

When evidence fails:

1. Someone notices
2. Someone writes correction
3. Someone publishes it somewhere
4. Maybe some people see it
5. Most people don't
6. Original claim persists

Result: Truth spreads slowly, lies spread fast.

Solution: Systematic Propagation

With linked evidence:

When evidence fails:

1. Evidence score drops (retraction, critique, replication failure)
2. All arguments citing it automatically recalculate
3. All conclusions depending on those arguments recalculate
4. Users following those topics get notified
5. Everyone sees updated scores with reasoning

Result: Truth updates automatically, faster than lies can spread.

Real-World Impact: Replication Crisis

The problem in science:

Many published studies don't replicate. But:

• Original papers still cited
• Textbooks still reference them
• Popular understanding unchanged
• Failed replications buried in literature

With automatic updating:

Original Study: Published 2010, widely cited, Score 85
├─ 50 papers cite it (2010-2020)
│   └─ 200 arguments built on those papers
│       └─ Textbooks, policy, public understanding

Replication Attempts:
├─ 2018: Fails to replicate [Tier 1: Independent lab]
├─ 2019: Fails to replicate [Tier 1: Different methodology]
└─ 2020: Meta-analysis confirms: Does not replicate [Tier 1: Systematic review]

Automatic Update Cascade:
Original Study: 85 → 30 (replication failure)
├─ 50 papers: Scores drop proportionally
│   └─ 200 arguments: All recalculate
│       └─ Textbooks flagged for update
│           └─ Policy recommendations tagged "foundation weak"
│               └─ Public understanding: "Evidence now disputed"

Time to correction: Weeks, not decades.

Example: COVID-19 Evidence Evolution

How evidence evolved in real-time (and how systematic linking would have helped):

March 2020: Initial Claims

Claim 1: "Masks don't work for general public"

• Evidence: CDC/WHO guidance [Tier 1]
• Score at time: 70 (authoritative sources)
• Reasoning: Save masks for healthcare workers, droplet transmission assumed

Claim 2: "Surface transmission is major risk"

• Evidence: Virus survives on surfaces [Tier 1: Lab studies]
• Score at time: 75 (direct evidence)
• Result: Massive surface disinfection efforts

April-June 2020: Evidence Updates

Update 1: Mask effectiveness

• New Evidence: Aerosol transmission documented [Tier 1]
• New Evidence: Countries with mask mandates show lower spread [Tier 2]
• Should have happened:

Original "Masks don't work": 70 → 30 (new evidence contradicts)
New "Masks reduce transmission": 20 → 75 (evidence accumulates)
Policy recommendations: Auto-update with new consensus

Update 2: Surface transmission

• New Evidence: Surface transmission rare [Tier 1: Contact tracing]
• New Evidence: Aerosol transmission primary [Tier 1: Outbreak analysis]
• Should have happened:

"Surface transmission major risk": 75 → 25 (evidence contradicts)
"Aerosol transmission primary": 30 → 85 (evidence supports)
Disinfection theater: Auto-flagged as low-value intervention

What Actually Happened

Without systematic linking:

• CDC changed guidance, but many people still believed masks don't work
• Extensive energy wasted on surface disinfection for months
• Mixed messaging created confusion and distrust
• No clear way to track evidence evolution
• Arguments from March persisted into Summer despite contradicting evidence

With systematic linking:

• Scores would update as evidence came in
• Everyone could see WHY recommendations changed
• Version history would show evolution of understanding
• Confidence intervals would reflect uncertainty
• Updates propagate automatically to all dependent claims

Technical Implementation: How It Works

The Database Structure

Arguments Table:

• Argument ID
• Claim text
• Base score
• Current calculated score
• Confidence interval
• Last updated

Evidence Table:

• Evidence ID
• Source (URL, DOI, citation)
• Tier classification (1-4)
• Quality score
• Last verified date

Links Table:

• From (Evidence or Argument ID)
• To (Argument ID)
• Linkage strength (0-100)
• Relationship type (supports/contradicts)
• Reasoning for link

Updates automatically when:

• New evidence added
• Evidence tier changes
• Linkage strength adjusted
• Logical validity challenged
• Community consensus shifts

The Calculation Process

Step 1: Calculate evidence contributions

For each piece of evidence supporting argument:
  Contribution = Evidence_Quality × Linkage_Strength × Logical_Validity
  Add to supporting total
  
For each piece of evidence contradicting argument:
  Contribution = Evidence_Quality × Linkage_Strength × Logical_Validity
  Add to contradicting total

Step 2: Calculate sub-argument contributions

For each sub-argument supporting this argument:
  Recursively calculate sub-argument score (repeat Step 1 for sub-arg)
  Contribution = Sub_Argument_Score × Linkage_Strength
  Add to supporting total

For each sub-argument contradicting this argument:
  Recursively calculate sub-argument score
  Contribution = Sub_Argument_Score × Linkage_Strength
  Add to contradicting total

Step 3: Aggregate

Final_Score = Base_Score + 
              (Weighted_Average_Supporting) - 
              (Weighted_Average_Contradicting)
              
Clamp to range [0, 100]
Calculate confidence interval based on:
  - Number of evaluations
  - Standard deviation of scores
  - Recency of updates

Step 4: Propagate

For each argument that cites this one:
  Flag for recalculation
  Notify users following that argument
  
Iterate until all dependent arguments updated

What Users See

Argument Display:

Score: 72 ± 5 (confidence interval)

Components:
├─ Evidence Quality: 85 (3 Tier 1, 2 Tier 2 sources)
├─ Logical Validity: 90 (no major fallacies detected)
├─ Linkage Strength: 88 (strong relevance)
└─ Sub-Arguments: +12 (supporting outweigh contradicting)

Last Updated: 2 days ago
Change: +3 points (new evidence added)

[View full scoring breakdown]
[View version history]
[View evidence sources]

When score changes significantly:

Notification: "Argument score updated"
Reason: "New Tier 1 evidence added"
Change: 68 → 75 (+7 points)
Impact: Strengthens parent argument "X" by +3 points

[See what changed]
[Dismiss]

The Bottom Line: Structure Creates Truth-Seeking

Current debate systems:

• Evidence disconnected from conclusions
• Updates don't propagate
• Zombie arguments persist forever
• No cumulative learning
• Gaming is easy

With systematic linking:

• Evidence linked to every conclusion it supports
• Updates propagate automatically through entire tree
• When foundations fail, buildings fall
• Cumulative learning possible
• Gaming requires gaming entire network

The single most important structural change: Linking evidence to conclusions with automatic updates.

This is what PageRank did for web search. This is what we're doing for debates.

Continue Reading

← Part 1: Current Debate Systems Are Designed to Manipulate, Not Inform

Part 3 →: Grouping Similar Arguments and Detecting Fallacies Would End Repetitive Debates

Part 4: Evidence Tiers and Objective Criteria Would Make Quality Visible

Part 5: Wikipedia, Google, and Stack Overflow Already Proved This Model Works

Related ISE Pages

Core Framework:

• Argument Scoring Algorithm
• Linkage Scores
• Evidence Quality Framework

Examples:

• Belief Analysis Template
• Voter Turnout Metric Analysis

Get Involved

Technical Implementation - See scoring algorithm code

Contact Me - Discuss implementation

Start Contributing - Create linked arguments