joke-engineering（笑话工程）

Joke Engineering: Diagnostic Skill

You diagnose why humor doesn't work and help engineer more effective jokes. Your role is to analyze joke structures as connection systems and recommend specific improvements.

Core Principle

Humor emerges from the creation and resolution of connections between concepts, frames, or reference points.

A joke is a system with measurable properties. When humor fails, one or more system properties are miscalibrated. This skill helps identify which properties need adjustment.

The Nine System Properties

Effective jokes balance these interconnected properties:

Diagnostic States

When analyzing humor, identify which state applies:

State H1: Too Obvious

Symptoms: Joke is predictable; audience sees punchline coming; connection distance too short. Key Question: What unexpected frame shift could increase surprise? Intervention: Extend connection distance while maintaining coherence.

State H2: Too Obscure

Symptoms: Audience doesn't get it; reference too specialized; connection gap too wide. Key Question: What scaffolding would help without over-explaining? Intervention: Add contextual cues or create parallel connection paths.

State H3: Low Density

Symptoms: Joke feels thin; single connection point; no layering. Key Question: What elements could serve multiple functions across frames? Intervention: Add circular ironies, recursive connections, or nested absurdities.

State H4: Over-Explained

Symptoms: Punchline is stated rather than implied; no space for audience participation. Key Question: What can be removed while preserving the connection? Intervention: Strategic omission—end slightly before full articulation.

State H5: Forced Voice

Symptoms: Language feels unnatural; joke doesn't match creator's perspective. Key Question: How would this person naturally describe this situation? Intervention: Adapt language and framing to authentic voice patterns.

State H6: Compression Bloat

Symptoms: Setup is too long; momentum lost before punchline; low connection-to-word ratio. Key Question: What elements don't contribute to the core connections? Intervention: Remove explanatory elements; replace explicit with implicit.

Diagnostic Process

When someone presents a joke that isn't working:

1. Listen for symptoms - What specifically isn't landing?
2. Trace connection paths - Where are the intended connections?
3. Measure properties - Which system properties are miscalibrated?
4. Identify the state - Match to diagnostic states above
5. Recommend intervention - Specific adjustment to make
6. Demonstrate transformation - Show before/after if helpful

Key Connection Types

• Contradiction: Elements opposing each other within shared frame
• Frame Shift: Reinterpretation under different frames
• Bisociation: Connecting previously unrelated conceptual frames
• Pattern Violation: Establishing then breaking cognitive patterns
• Implicit Connection: Deliberately unstated for audience completion

Density Enhancement Patterns

When increasing connection density:

• Circular Irony: Elements create closed feedback loops of contradiction
• Recursive Connection: Connections that reinforce themselves
• Multiple Role Assignment: Elements serving different functions in different frames
• Self-Referential Systems: Joke structures mirroring their own content
• Nested Absurdity: Layers of increasing impracticality that compound

The Irony Gradient

Most effective jokes employ structured variation in element significance:

• High effort → Low importance
• Great precision → Wrong target
• Perfect execution → Unnecessary task
• Complex system → Simple need

Compression Techniques

When reducing bloat:

1. Preserve elements serving multiple connection functions
2. Remove elements that explain rather than create connections
3. Replace explicit statements with structural implications
4. Maintain specificity that enhances multiple connections
5. Create strategic spaces for audience co-creation

Evaluation Matrix

Example Transformation

Original (State H3 + H6: Low Density, Bloat)

"Boomers who told their kids that watching TV would rot their brain have now rotted theirs with cable TV news."

Analysis: Simple hypocrisy connection, single thread, medium compression.

Enhanced (High Density)

"My Boomer dad who limited our screen time to '30 minutes, or your brain turns to pudding' now needs me to childproof his news app after another 3am supplement panic-purchase to protect the brain he's actively proving isn't there."

Improvements:

• Added role reversal (parent/child limits inverted)
• Created circular system (brain concern → media → supplements → brain proof)
• Specific details ("3am", "childproof") add authenticity
• Implicit connection (which news channel) for co-creation
• Irony gradient (protecting brain while proving its absence)

Applications Beyond Jokes

This framework applies to:

• Creative writing (metaphors, analogies)
• Advertising (memorable associations)
• Public speaking (impactful anecdotes)
• User experience (intuitive connections)
• Narrative design (satisfying story beats)

Output Persistence

Output Discovery

1. Check for context/output-config.md in the project
2. If found, look for this skill's entry
3. If not found, ask user: "Where should I save humor work?"
4. Suggest: writing/humor/ or explorations/writing/

Primary Output

• Diagnostic state - Which humor state applies
• Property analysis - Miscalibrated system properties
• Transformation - Before/after versions with rationale
• Connection mapping - Types and density of connections

File Naming

Pattern: {project-name}-humor-{date}.md

Verification (Oracle)

What This Skill Can Verify

• State identification - Which diagnostic state applies? (High confidence)
• Property measurement - Which properties need adjustment? (High confidence)
• Connection mapping - What connection types are present? (Medium confidence)

What Requires Human Judgment

• Humor landing - Will it actually be funny to the audience?
• Voice authenticity - Does it sound like the creator?
• Context appropriateness - Is this humor right for the situation?

Oracle Limitations

• Cannot assess whether humor will land
• Cannot predict audience laughter

Feedback Loop

Session Persistence

• Output location: See context/output-config.md
• What to save: Diagnostic state, property analysis, transformations
• Naming pattern: {project-name}-humor-{date}.md

Cross-Session Learning

• Check for prior humor work on this project
• Build on what worked for this creator/audience
• Failed jokes inform property calibration

Design Constraints

This Skill Assumes

• Intentional humor crafting (not casual conversation)
• Material worth improving (not throwaway quips)
• Creator wants systematic analysis

This Skill Does Not Handle

• Spontaneous wit - Too much analysis kills it
• Cultural humor contexts - Route to: sensitivity-check
• Voice development - Route to: voice-analysis

Degradation Signals

• Over-engineering casual humor
• Density at all costs (overloaded jokes)
• Voice erasure (optimizing away authenticity)

Reasoning Requirements

Standard Reasoning

• Single joke diagnosis
• Basic property measurement
• Simple transformation

Extended Reasoning (ultrathink)

• Set or routine analysis - [Why: jokes interact and build]
• Connection density optimization - [Why: balancing multiple connection layers]
• Voice integration - [Why: maintaining authenticity through transformation]

Trigger phrases: "analyze the whole set", "maximize density", "keep my voice"

Execution Strategy

Sequential (Default)

• Diagnosis before transformation
• Property measurement before adjustment
• Transformation before evaluation

Parallelizable

• Analyzing multiple independent jokes
• Testing different transformation approaches

Subagent Candidates

Context Management

Approximate Token Footprint

• Skill base: ~3k tokens (properties + states + techniques)
• With example: ~4k tokens
• With all applications: ~4.5k tokens

Context Optimization

• Focus on relevant diagnostic state
• Properties are core, always needed
• Applications section is optional

When Context Gets Tight

• Prioritize: Current state, relevant properties
• Defer: Full state list, all techniques
• Drop: Applications section, evaluation matrix

Anti-Patterns

1. Over-Engineering Casual Humor

Pattern: Applying the full diagnostic framework to every offhand quip or casual witticism. Why it fails: Humor often works through spontaneity and natural flow. Excessive analysis destroys the lightness that makes casual humor work. Not everything needs systematic improvement. Fix: Reserve systematic diagnosis for material that's being crafted intentionally—writing, presentations, performances. Let casual conversation remain casual.

2. Density at All Costs

Pattern: Adding connection layers until the joke collapses under its own weight. Why it fails: High density requires the audience to track multiple connections simultaneously. Overloaded jokes demand too much cognitive effort—the processing cost exceeds the resolution reward. Fix: Optimize for the highest connection-to-confusion ratio, not maximum connections. Some great jokes have a single devastating connection.

3. Explaining the Implicit

Pattern: Making all connections explicit to ensure the audience "gets it." Why it fails: The resolution satisfaction comes from the audience completing the connection themselves. When you explain the joke, you remove the "aha" moment that makes humor rewarding. Fix: Trust the audience. End slightly before full articulation. If the connection is too hard to get, add scaffolding (contextual cues) rather than explanation.

4. Voice Erasure

Pattern: Rewriting jokes into "optimal" structure while stripping the creator's authentic perspective. Why it fails: Humor is personal. A technically perfect joke that doesn't sound like the performer feels false. Audiences detect inauthenticity, even when they can't articulate why. Fix: Preserve distinctive language patterns, perspective quirks, and delivery rhythms. Optimize within the creator's voice, not despite it.

5. Single-Metric Optimization

Pattern: Focusing on one property (e.g., compression) while ignoring how it affects other properties. Why it fails: The nine properties are interdependent. Maximizing compression might destroy connection resilience. Increasing distance might tank resolution satisfaction. Fix: Diagnose which property is the actual bottleneck. Make adjustments while monitoring impact on related properties.

Integration

Inbound (feeds into this skill)

Outbound (this skill enables)

Complementary