16-Point Evaluation Rubric

The rubric is the foundation of professional screenplay coverage. Each dimension is scored 1-5, with specific criteria defining what constitutes each score level.

The 16 points are organized into four categories, each contributing a weighted percentage to the overall evaluation:

┌─────────────────────────────────────────────────────────────────┐
│                     16-POINT RUBRIC                             │
├─────────────────┬─────────────────┬─────────────────┬───────────┤
│   STORY (35%)   │  MARKET (25%)   │PRODUCTION (25%) │DECISION   │
│                 │                 │                 │  (15%)    │
├─────────────────┼─────────────────┼─────────────────┼───────────┤
│ 1. Logline      │ 5. Themes       │ 9. Risks        │13. Score  │
│ 2. Summary      │ 6. Originality  │10. Budget       │14. Rights │
│ 3. Structure    │ 7. Market       │11. Format       │15. Rewrite│
│ 4. Characters   │ 8. Craft        │12. Recommend    │16. Action │
└─────────────────┴─────────────────┴─────────────────┴───────────┘
                

Each point uses a 1-5 scoring scale with defined criteria for each level. See the individual rubric sections for detailed scoring guidelines.

Rubric: Story (Points 1-4)

The Story category evaluates the fundamental narrative elements that make a screenplay work as a piece of storytelling.

Rubric: Market (Points 5-8)

The Market category evaluates commercial viability, originality, and craft elements that determine audience appeal.

Rubric: Production (Points 9-12)

The Production category evaluates practical considerations for bringing the screenplay to screen.

Rubric: Decision (Points 13-16)

The Decision category provides actionable outputs for production and development decisions.

Coverage Orchestration Protocol

The Coverage Orchestration Protocol (COP) is a hierarchical convergent analysis pipeline that processes screenplays through seven execution tiers, each building on the output of the previous.

Key Principles

• Parallelization: Maximize concurrent processing at every tier
• Validation Gates: Quality checkpoints between tiers (SYNC_01 through SYNC_06)
• Hierarchical Convergence: Many parallel streams converge into unified output
• Evidence-Based: Every claim traced to specific textual evidence

Tier 0: Document Segmentation

Intelligently split the document into analyzable units without prior knowledge of structure.

Purpose

Parse raw screenplay files and identify structural boundaries (scenes, sequences, acts) for downstream parallel processing.

Inputs

• Raw PDF, FDX (Final Draft), Fountain, TXT, or MD file

Outputs

• Segmented scene units with boundaries
• Act structure identification
• Metadata (title, author, page count, format)
• Scene-to-page mapping

Validation Gate: SYNC_01

Ensures document is properly segmented before proceeding. Validates scene count, act boundaries, and metadata completeness.

Tier 1: Parallel Scene Analysis

Deep analysis of each scene by specialized perspectives, with all scenes executing in parallel.

Purpose

Generate deep, multi-perspective analysis of each scene. This is where the majority of textual evidence is gathered.

Perspectives (5-10 per scene)

Execution Pattern

• All scenes launch simultaneously
• Each scene receives 5-10 parallel perspective analyses
• Typical screenplay: 40-60 scenes × 6 perspectives = 240-360 parallel tasks

Outputs

• Scene analysis documents (one per scene)
• Character mentions and dialogue extracts
• Thematic markers
• Structural function tags

Validation Gate: SYNC_02

Ensures all scenes are analyzed before proceeding to cross-reference synthesis.

Tier 2: Cross-Reference & Character Extraction

Identify patterns across scenes and build initial character profiles from scene-level evidence.

Tier 2A: Cross-Reference Synthesis

Purpose: Identify patterns and connections across scenes within each act.

• Recurring motifs and symbols
• Thematic threads
• Structural patterns
• Tonal consistency/shifts

Tier 2B: Character Extraction

Purpose: Build initial character profiles from scene-level evidence.

• Character names and roles
• Dialogue patterns
• Relationship mapping
• Initial arc identification

Validation Gate: SYNC_03

Ensures cross-reference synthesis and character extraction are complete before act-level aggregation.

Tier 3: Act-Level Aggregation

Comprehensive analysis of each act combining all lower-tier outputs into cohesive act-level understanding.

Purpose

Create comprehensive act-level analyses that synthesize scene analyses, cross-references, and character data into cohesive understanding of each act's function.

Inputs (per act)

• All scene analyses within the act
• Cross-reference synthesis for the act
• Character extractions relevant to the act

Outputs

• Act analysis documents (one per act)
• Act-level arc progression
• Turning point identification
• Momentum and pacing assessment

Validation Gate: SYNC_04

Ensures act-level aggregation is complete before meta-analysis and character enhancement.

Tier 4: Meta-Analysis & Character Enhancement

Multiple independent teams review all act analyses while character profiles are enhanced with arc analysis and psychological depth.

Tier 4A: Meta-Analysis

Purpose: Multiple independent teams review all act analyses to ensure accuracy, consistency, and completeness.

• Cross-validate findings across teams
• Identify contradictions or gaps
• Ensure evidence supports conclusions
• Quality audit of all prior analysis

Tier 4B: Character Enhancement

Purpose: Iteratively enhance character profiles with act-level insights and psychological depth.

• Full arc analysis (beginning → end)
• Psychological motivation depth
• Relationship dynamics mapping
• Voice and dialogue pattern analysis

Validation Gate: SYNC_05

Ensures meta-analysis and character enhancement are complete before final synthesis.

Tier 5: Final Synthesis

Diverse expert panel produces comprehensive final outputs by synthesizing all prior analyses into cohesive coverage.

Purpose

Produce the final, comprehensive coverage report by synthesizing all prior analyses into a unified assessment.

Expert Panel

• Researcher: Ensures accuracy and evidence basis
• Writer: Evaluates craft and creative merit
• Analyst: Assesses structure and mechanics
• Psychologist: Character and emotional depth
• Philosopher: Thematic resonance and meaning
• Business Development: Commercial viability
• Strategic Planner: Market positioning and action items

Outputs

• Complete 16-point rubric with scores and justification
• Three Questions assessment (Mechanical, Intuitive, Artistic)
• Four Articulations (Gut Check, Fatal Flaw, Hidden Gem, Final Verdict)
• Detailed character breakdowns
• Market analysis and positioning
• Development recommendations

Tier 6: Writer Analysis

Analyze the character and psychology of the writer from their work, identifying distinctive voice, craft patterns, and creative signature.

Purpose

Create a profile of the writer based on analysis of their work—voice, thematic obsessions, craft tendencies, and unique signature.

Analysis Dimensions

Voice

• Distinctive quality of writing
• Rhythm and pacing patterns
• Diction and vocabulary choices
• Dialogue style

Themes

• Primary thematic obsessions
• Recurring questions explored
• Philosophical or moral concerns

Craft

• Structural approach (traditional vs. experimental)
• Dialogue technique
• Overall craft proficiency level

Signature

• What makes this writer unique
• Literary influences and echoes
• Potential trajectory and growth areas

Validation Gate: SYNC_06

Final validation ensuring all components (Tier 5 synthesis + Tier 6 writer analysis) are complete and consistent.

Validation Gates

The protocol includes six validation gates (SYNC_01 through SYNC_06) that ensure quality and accuracy at each transition point.

Quality Metrics Checked

• Coverage Completeness: All rubric dimensions have substantive analysis
• Evidence Density: Claims supported by specific textual citations
• Cross-Validation: Agreement across independent analysis streams
• Character Accuracy: Character details consistent across mentions
• Coherence: No contradictory conclusions between tiers

Architecture Overview

Three architectural approaches offer different tradeoffs between speed, quality, and use case. Choose based on your primary requirement.

Selection Guide

Parallel Specialists

Split the 16-point rubric across four specialized analysis streams running in parallel on the same screenplay. Recommended for production deployments.

How It Works

The architecture divides the 16-point rubric across four specialists, each running an independent Llama 3.3 70B instance with tensor parallelism:

When to Use

• Production deployments requiring consistent, high-quality coverage
• When quality and throughput are both important
• Standard studio/production company workflows

Draft-Refine

Two-pass architecture using a fast model for initial draft, then a larger model verifies and deepens the analysis. Maximum quality for premium coverage.

When to Use

• Premium coverage services where quality justifies slower processing
• Writer feedback requiring detailed, nuanced notes
• Development executives reviewing promising material
• Final evaluation before greenlight decisions

Confidence Cascade

Fast rejection of weak scripts using a gate model, with full analysis only for promising material. Optimizes for high-volume screening.

Gate Model Metrics

When to Use

• Festival/competition screening with high rejection rates
• Open submission periods with volume spikes
• When 85%+ scripts are expected to fail
• Initial triage before detailed coverage

3x GH200 Tri-Specialist

Three GH200 nodes with 288GB HBM3. Entry-scale configuration with the highest performance-per-cost ratio. Ideal for smaller operations or initial deployments.

Category Distribution (5/5/6 Split)

4x GH200 Quad-Specialist

Four GH200 nodes with 384GB HBM3. Optimal cost-performance balance achieving 75% of premium throughput at 60% of the cost. Recommended for cost-sensitive production.

Category Distribution (4/4/4/4 Split)

Alternative: Hierarchical Draft-Refine

For higher quality (92%) at lower throughput:

5x GH200 Penta-Hybrid

Five GH200 nodes with 480GB HBM3. Maximum GH200 scale enabling hybrid gate+specialist architecture. One node screens, four nodes analyze. Adapts to submission quality.

Component Configuration

Path Timing

• Fast Path (70%): ~1.5s with brief rejection notes
• Full Path (30%): ~7.3s complete coverage
• Weighted Average: 6.2s per script

Alternative Architectures

Benchmarks

Coverage AGI exceeds human coverage across measurable dimensions. These aren't arbitrary claims—they're verifiable metrics with proposed validation protocols.

Where Coverage AGI Exceeds Human

Architecture Performance

Proposed Validation Protocol

These claims are designed to be proveable. We propose the following benchmark methodology:

Quality Metrics

Quality is measured across multiple dimensions where Coverage AGI demonstrably outperforms human coverage.

Why Coverage AGI Is Superior

Inter-Reader Agreement Problem

Studies of human screenplay coverage show substantial disagreement between professional readers:

This means one in three screenplays would receive a different recommendation depending on which reader happened to get it. Coverage AGI eliminates this lottery.

Measurable Superiority

For each metric below, Coverage AGI output can be directly compared to human coverage:

• Citation count: Number of specific page/line references per claim
• Scene coverage: Percentage of scenes with substantive analysis
• Character completeness: Number of characters with full arc analysis
• Rubric compliance: Percentage of dimensions with explicit scoring rationale
• Actionability: Number of specific, implementable recommendations

Quick Deliverables

Rapid-access outputs for immediate assessment before diving into full coverage.

Full Coverage Report

The complete 10-section coverage document providing comprehensive screenplay analysis.

1. Introduction

   Overview and first impressions. Genre, tone, comparable titles.

2. Scene Analysis

   Scene-by-scene breakdown covering dialogue quality, emotional beats, narrative function.

3. Character Profiles

   Detailed profiles including backstory, motivations, conflict, and arc for each significant character.

4. Theme and Tone Analysis

   Core themes, atmosphere, and thematic development across the narrative.

5. Technical Craft Assessment

   Scene structure, pacing, information delivery, setup/payoff mechanics, transition effectiveness.

6. Recommendations

   Actionable suggestions for character development, plot enhancement, dialogue refinement, theme strengthening.

7. Gut Check

   Instinctual 1-2 page assessment of overall quality and potential.

8. Fatal Flaw

   The most significant weakness or area for improvement (2-3 pages).

9. Hidden Gem

   Overlooked strength or opportunity for growth (2-3 pages).

10. Final Verdict

    Comprehensive evaluation of quality, potential, and marketability (5-10 pages).

Script Rewrite Ideas

Beyond evaluation, coverage includes actionable rewrite suggestions targeting identified weaknesses while preserving the screenplay's strengths and voice.

Export Formats

Coverage reports available in multiple formats for different workflows.

From the Writer's Perspective

The Three Questions

Every screenplay encounter answers three fundamental questions. These aren't metrics—they're modes of knowing.

┌─────────────────────────────────────────────────────────────────┐
│                    THE THREE QUESTIONS                          │
├─────────────────────────────────────────────────────────────────┤
│                                                                 │
│     ┌─────────────────────────────────────────────────────┐     │
│     │                                                     │     │
│     │        1. DOES IT WORK?                            │     │
│     │           (Mechanical Capacity)                     │     │
│     │                                                     │     │
│     │        Structure, pacing, craft.                    │     │
│     │        Can be answered with evidence.               │     │
│     │                                                     │     │
│     └─────────────────────────────────────────────────────┘     │
│                            │                                    │
│                            ▼                                    │
│     ┌─────────────────────────────────────────────────────┐     │
│     │                                                     │     │
│     │        2. WOULD YOU GREENLIGHT IT?                 │     │
│     │           (Intuitive Capacity)                      │     │
│     │                                                     │     │
│     │        Commercial instinct, gut feeling.            │     │
│     │        Reveals the reader as much as the script.    │     │
│     │                                                     │     │
│     └─────────────────────────────────────────────────────┘     │
│                            │                                    │
│                            ▼                                    │
│     ┌─────────────────────────────────────────────────────┐     │
│     │                                                     │     │
│     │        3. DOES IT MATTER?                          │     │
│     │           (Artistic Capacity)                       │     │
│     │                                                     │     │
│     │        Blood vs craft. Authentic struggle.          │     │
│     │        The hardest question.                        │     │
│     │                                                     │     │
│     └─────────────────────────────────────────────────────┘     │
│                                                                 │
└─────────────────────────────────────────────────────────────────┘
                

Mechanical Capacity

"Does it work?"

The technical assessment. Structure, pacing, craft. Does the screenplay function as a narrative machine?

This is the domain of evidence. Point to the page. Show the problem or the solution. Does Act Two build on Act One? Does the climax pay off the setup? Are the character arcs complete?

What It Evaluates

• Three-act structure and act breaks
• Scene-to-scene momentum
• Setup and payoff mechanics
• Character arc completion
• Dialogue functionality
• Format and technical execution

How It's Answered

With evidence. Citations from the text. Specific page references. Demonstrable structural analysis.

Intuitive Capacity

"Would you greenlight it?"

The producer's question. Not "is it good?" but "would you make it?"

This requires something beyond craft analysis—it requires commercial instinct, gut feeling, the ability to sense what an audience might want. It's the question that separates readers from executives.

What It Evaluates

• Market timing and relevance
• Audience appeal
• Castability
• Production feasibility
• Commercial potential
• Competition and positioning

How It's Answered

With instinct tempered by experience. This question reveals as much about the reader as the screenplay. Different readers will answer differently—and that's information too.

Artistic Capacity

"Does it matter?"

The hardest question. Does this screenplay contain blood? Did the writer touch what they were reaching for?

Some screenplays work perfectly and would greenlight easily—but don't matter. Others are broken and uncommercial—but burn with something true. This question asks about authenticity, not competence.

What It Evaluates

• Authenticity of vision
• Thematic depth and resonance
• Originality of voice
• Emotional truth
• Cultural or artistic significance
• The writer's "reaching"

How It's Answered

With presence. This isn't analysis—it's encounter. The reader must be open to what the screenplay is attempting, not just what it achieves.

The Four Articulations

Coverage speaks in four voices. Each articulation serves a different purpose and emerges from a different kind of attention.

┌─────────────────────────────────────────────────────────────────┐
│                  THE FOUR ARTICULATIONS                         │
├─────────────────────────────────────────────────────────────────┤
│                                                                 │
│    ┌─────────────────┐                                          │
│    │   GUT CHECK     │  What you say when you look up           │
│    │   (1-2 pages)   │  from reading.                           │
│    └────────┬────────┘                                          │
│             │                                                   │
│             ▼                                                   │
│    ┌─────────────────┐                                          │
│    │   FATAL FLAW    │  What would kill this project.           │
│    │   (2-3 pages)   │  Or "none."                              │
│    └────────┬────────┘                                          │
│             │                                                   │
│             ▼                                                   │
│    ┌─────────────────┐                                          │
│    │   HIDDEN GEM    │  What others will miss that              │
│    │   (2-3 pages)   │  you saw.                                │
│    └────────┬────────┘                                          │
│             │                                                   │
│             ▼                                                   │
│    ┌─────────────────┐                                          │
│    │  FINAL VERDICT  │  One voice, speaking truth.              │
│    │   (5-10 pages)  │  The synthesis.                          │
│    └─────────────────┘                                          │
│                                                                 │
└─────────────────────────────────────────────────────────────────┘
                

Gut Check

"What you say when you look up from reading."

The immediate, unfiltered response. Before analysis, before justification—what does the body know?

The gut check captures first impressions that often prove more reliable than subsequent rationalization. It's the instinct that fires before the intellect engages.

What It Contains

• Immediate emotional response
• First impression of quality
• Instinctive sense of commercial potential
• Unfiltered reaction to voice and tone

1-2 pages

Fatal Flaw

"What would kill this project."

Every screenplay has weaknesses. The fatal flaw is the one that matters—the structural problem, the character gap, the tonal confusion that would doom production.

Sometimes the answer is "none." That's worth knowing. But when a fatal flaw exists, identifying it clearly is the most valuable thing coverage can do.

Common Fatal Flaws

• Protagonist lacks clear motivation
• Second act sag with no momentum
• Tone inconsistency (comedy/drama confusion)
• Climax doesn't pay off setup
• Theme contradicts story
• Dialogue indistinguishable between characters

2-3 pages

Hidden Gem

"What others will miss that you saw."

The strength that isn't obvious. The scene that works better than it should. The character who deserves more screen time. The theme hiding beneath the surface.

Hidden gems are why coverage matters. A script might fail on obvious metrics but contain something worth championing. The hidden gem is what makes a reader say "but wait—"

Examples

• Supporting character more compelling than protagonist
• Single scene that demonstrates exceptional craft
• Thematic subtext richer than surface story
• Voice that could anchor a different project
• World-building worth developing

2-3 pages

Final Verdict

"One voice, speaking truth."

The synthesis. Everything comes together into a single, coherent assessment. Not a summary of the other articulations—a new thing that emerges from having done the work.

What is this screenplay, really? What should happen next? The final verdict speaks with authority because it has earned that authority through rigorous encounter.

What It Contains

• Comprehensive quality assessment
• Clear recommendation (Pass/Consider/Recommend)
• Development notes if Consider
• Comparison to successful comps
• Specific action items
• Writer assessment and potential

5-10 pages

Blood vs Craft

"A screenplay is a frozen gesture of consciousness. Someone reached for something they couldn't hold. The marks on the page are the residue of that reaching."

The Distinction

Craft is technical skill. Structure, dialogue, pacing, format. Craft can be taught, measured, improved. A screenplay with excellent craft functions smoothly—every scene advances plot, every line of dialogue reveals character, every act break lands where it should.

Blood is something else. Blood is the authentic struggle on the page. It's what happens when a writer reaches for something true and the reaching itself becomes visible. Blood can't be taught. It can only be recognized.

Why It Matters

Most coverage systems measure craft. They can tell you if the structure works, if the dialogue sings, if the pacing holds. These are valuable assessments.

But the screenplays that change things—the ones that launch careers, that become cultural touchstones, that matter decades later—these screenplays have blood. Sometimes they have flawed craft. But something true is happening on the page.

The Question

When coverage encounters a screenplay, it asks: Did the writer touch what they were reaching for?

This isn't about success or failure. It's about authenticity of attempt. Some writers reach for small things and grasp them perfectly. Others reach for impossible things and fall short gloriously. Both can have blood. What matters is the reaching.