Initial commit
This commit is contained in:
Zhongwei Li
196
skills/dialectical-mapping-steelmanning/SKILL.md
Normal file
196
skills/dialectical-mapping-steelmanning/SKILL.md
Normal file
@@ -0,0 +1,196 @@
|
||||
---
|
||||
name: dialectical-mapping-steelmanning
|
||||
description: Use when debates are trapped in false dichotomies, polarized positions need charitable interpretation, tradeoffs are obscured by binary framing, synthesis beyond 'pick one side' is needed, or when users mention steelman arguments, thesis-antithesis-synthesis, Hegelian dialectic, third way solutions, or resolving seemingly opposed principles.
|
||||
---
|
||||
# Dialectical Mapping & Steelmanning
|
||||
|
||||
## Table of Contents
|
||||
- [Purpose](#purpose)
|
||||
- [When to Use](#when-to-use)
|
||||
- [What Is It?](#what-is-it)
|
||||
- [Workflow](#workflow)
|
||||
- [Common Patterns](#common-patterns)
|
||||
- [Guardrails](#guardrails)
|
||||
- [Quick Reference](#quick-reference)
|
||||
|
||||
## Purpose
|
||||
|
||||
Dialectical Mapping & Steelmanning helps you escape false binary choices by:
|
||||
|
||||
- **Steelmanning** both positions (presenting them in their strongest, most charitable form)
|
||||
- **Mapping** the underlying principles and tradeoffs (what each side values and sacrifices)
|
||||
- **Synthesizing** a principled third way (transcending "pick a side" to find higher-order resolution)
|
||||
- **Making tradeoffs explicit** (clarifying costs/benefits of synthesis vs pure positions)
|
||||
|
||||
This moves debates from "A vs B" to "here's the best of both, here's what we sacrifice, here's why it's worth it."
|
||||
|
||||
## When to Use
|
||||
|
||||
Use this skill when:
|
||||
|
||||
- **False dichotomies**: Debate framed as binary choice ("we must pick A or B") but better options exist
|
||||
- **Polarized positions**: Both sides dug in, uncharitable interpretations, strawman arguments flying
|
||||
- **Hidden tradeoffs**: Each position has merits and costs, but these aren't explicit
|
||||
- **Principle conflicts**: Seemingly opposed values (speed vs quality, freedom vs safety, innovation vs stability)
|
||||
- **Synthesis needed**: User explicitly wants "third way", "best of both worlds", or "transcend the debate"
|
||||
- **Strategic tensions**: Business decisions with legitimate competing priorities (growth vs profitability, centralization vs autonomy)
|
||||
- **Design tradeoffs**: Technical or product decisions with no clear winner (monolith vs microservices, simple vs powerful)
|
||||
- **Policy debates**: Governance questions with multiple stakeholder values (privacy vs security, efficiency vs equity)
|
||||
|
||||
Trigger phrases: "steelman", "thesis-antithesis-synthesis", "Hegelian dialectic", "false dichotomy", "third way", "both sides have a point", "transcend the debate", "resolve the tension"
|
||||
|
||||
## What Is It?
|
||||
|
||||
Dialectical Mapping & Steelmanning is a three-step reasoning process:
|
||||
|
||||
1. **Steelman Thesis & Antithesis**: Present each position in its strongest form (charitable interpretation, best arguments, underlying principles)
|
||||
2. **Map Tradeoffs**: Identify what each side optimizes for and what it sacrifices
|
||||
3. **Synthesize Third Way**: Find a higher-order principle or hybrid approach that honors both positions' core values while acknowledging new tradeoffs
|
||||
|
||||
**Quick example:**
|
||||
|
||||
**Debate**: "Should our startup prioritize growth or profitability?"
|
||||
|
||||
**Typical (bad) framing**: Binary choice. Pick one, argue against the other.
|
||||
|
||||
**Steelman Thesis (Growth)**:
|
||||
- Principle: Market position compounds. Early lead captures network effects, brand recognition, talent attraction.
|
||||
- Best argument: In winner-take-most markets, second place is first loser. Profitability can wait; market share can't.
|
||||
- Tradeoff: Accept cash burn, potential failure if funding dries up.
|
||||
|
||||
**Steelman Antithesis (Profitability)**:
|
||||
- Principle: Sustainability enables long-term strategy. Profitable companies control their destiny, survive downturns, outlast competitors.
|
||||
- Best argument: Growth without unit economics is vanity metric. Profit proves business viability.
|
||||
- Tradeoff: Accept slower growth, risk being outpaced by well-funded competitors.
|
||||
|
||||
**Synthesis (Profitable Growth)**:
|
||||
- **Higher principle**: Capital efficiency. Grow as fast as sustainable unit economics allow.
|
||||
- **Third way**: Focus on channels/segments with healthy LTV:CAC (>3:1), deprioritize expensive acquisition. Scale what works profitably, experiment cheaply elsewhere.
|
||||
- **New tradeoffs**: Slower than "growth at all costs", requires discipline to say no, may miss land-grab opportunities in subsidized markets.
|
||||
- **Why it works**: Preserves optionality (can raise capital from position of strength OR bootstrap), builds durable moat (real economics, not just scale), reduces existential risk.
|
||||
|
||||
**Result**: Escaped false binary. Found principled synthesis with explicit tradeoffs.
|
||||
|
||||
## Workflow
|
||||
|
||||
Copy this checklist and track your progress:
|
||||
|
||||
```
|
||||
Dialectical Mapping Progress:
|
||||
- [ ] Step 1: Frame the debate
|
||||
- [ ] Step 2: Steelman Position A (Thesis)
|
||||
- [ ] Step 3: Steelman Position B (Antithesis)
|
||||
- [ ] Step 4: Map principles and tradeoffs
|
||||
- [ ] Step 5: Synthesize third way
|
||||
- [ ] Step 6: Validate synthesis quality
|
||||
```
|
||||
|
||||
**Step 1: Frame the debate**
|
||||
|
||||
Identify the topic, the two polarized positions (Thesis vs Antithesis), and the apparent tension. Clarify why this feels like a binary choice. See [Common Patterns](#common-patterns) for typical debate structures.
|
||||
|
||||
**Step 2: Steelman Position A (Thesis)**
|
||||
|
||||
Present Position A in its strongest form: underlying principle (what it values), best arguments (strongest case for this position), supporting evidence, and legitimate tradeoffs it accepts. Use [resources/template.md](resources/template.md#steelmanning-template) for structure. Avoid strawmanning—present version that adherents would recognize as fair.
|
||||
|
||||
**Step 3: Steelman Position B (Antithesis)**
|
||||
|
||||
Present Position B in its strongest form with same rigor as Position A. Ensure symmetry—both positions get charitable treatment. See [resources/template.md](resources/template.md#steelmanning-template).
|
||||
|
||||
**Step 4: Map principles and tradeoffs**
|
||||
|
||||
Create tradeoff matrix showing what each position optimizes for (values) and what it sacrifices (costs). Identify underlying principles (speed, quality, freedom, safety, etc.) and how each position weighs them. For complex cases with multiple principles, see [resources/methodology.md](resources/methodology.md#principle-mapping) for multi-dimensional tradeoff analysis.
|
||||
|
||||
**Step 5: Synthesize third way**
|
||||
|
||||
Find higher-order principle or hybrid approach that transcends the binary. The synthesis should honor core values of both positions, create new value (not just compromise), and make new tradeoffs explicit. Use [resources/template.md](resources/template.md#synthesis-template) for structure. For advanced synthesis techniques (temporal synthesis, conditional synthesis, dimensional separation), see [resources/methodology.md](resources/methodology.md#synthesis-patterns).
|
||||
|
||||
**Step 6: Validate synthesis quality**
|
||||
|
||||
Self-assess using [resources/evaluators/rubric_dialectical_mapping_steelmanning.json](resources/evaluators/rubric_dialectical_mapping_steelmanning.json). Check: steelmans are charitable and accurate, principles identified, tradeoffs explicit, synthesis transcends binary (not just compromise), new tradeoffs acknowledged. **Minimum standard**: Average score ≥ 3.5.
|
||||
|
||||
## Common Patterns
|
||||
|
||||
**Pattern 1: Temporal Synthesis (Both, Sequenced)**
|
||||
- **Structure**: Do A first, then B. Or B in some phases, A in others.
|
||||
- **Example**: "Speed vs Quality" → **Synthesis**: Iterate fast early (speed), stabilize before launch (quality). Time-box exploration, then shift to refinement.
|
||||
- **When to use**: Positions optimize for different lifecycle stages or contexts.
|
||||
|
||||
**Pattern 2: Conditional Synthesis (Both, Contextual)**
|
||||
- **Structure**: A in these situations, B in those situations. Define decision criteria.
|
||||
- **Example**: "Centralized vs Decentralized" → **Synthesis**: Centralize strategy/standards/shared resources, decentralize execution/tactics/experiments. Clear escalation criteria for edge cases.
|
||||
- **When to use**: Positions are optimal in different scenarios or scopes.
|
||||
|
||||
**Pattern 3: Dimensional Separation (Both, Different Axes)**
|
||||
- **Structure**: Optimize A on one dimension, B on another orthogonal dimension.
|
||||
- **Example**: "Simple vs Powerful" → **Synthesis**: Simple by default (80% use cases), powerful for power users (progressive disclosure, advanced mode). Complexity optional, not mandatory.
|
||||
- **When to use**: Tradeoff is false—can achieve both on different dimensions simultaneously.
|
||||
|
||||
**Pattern 4: Higher-Order Principle (Transcend via Meta-Goal)**
|
||||
- **Structure**: Both A and B are means to same end. Find better means.
|
||||
- **Example**: "Build vs Buy" → **Synthesis**: Neither—rent/SaaS. Or: Build core differentiator, buy commodity. Higher principle: Maximize value creation per dollar/hour.
|
||||
- **When to use**: Binary options are tactics, not ends. Reframe around shared ultimate goal.
|
||||
|
||||
**Pattern 5: Compensating Controls (Accept A's Risk, Mitigate with B's Safeguard)**
|
||||
- **Structure**: Lean toward A, add B's protections as guardrails.
|
||||
- **Example**: "Move Fast vs Prevent Errors" → **Synthesis**: Move fast with automated testing, staged rollouts, quick rollback. Accept some errors, contain blast radius.
|
||||
- **When to use**: One position clearly better for primary goal, other provides risk mitigation.
|
||||
|
||||
## Guardrails
|
||||
|
||||
**Critical requirements:**
|
||||
|
||||
1. **Steelman, don't strawman**: Present each position as its adherents would recognize. Ask: "Would someone who holds this view agree this is a fair representation?" If no, strengthen it further.
|
||||
|
||||
2. **Identify principles, not just preferences**: Go deeper than "Side A wants X, Side B wants Y." Find WHY they want it. What value do they optimize for? Freedom? Safety? Speed? Equity? Efficiency?
|
||||
|
||||
3. **Synthesis must transcend, not just compromise**: Splitting the difference (50% A, 50% B) is usually weak. Good synthesis finds new option C that honors both principles at higher level. "Both-and" thinking, not "either-or" averaging.
|
||||
|
||||
4. **Make tradeoffs explicit**: Every synthesis has costs. State what you gain AND what you sacrifice vs pure positions. Don't pretend synthesis is "best of both with no downsides."
|
||||
|
||||
5. **Avoid false equivalence**: Steelmanning doesn't mean both sides are equally correct. One position may have stronger arguments/evidence. Synthesis should reflect this (lean toward stronger position, add safeguards from weaker).
|
||||
|
||||
6. **Check for false dichotomy**: Some "debates" are manufactured. Both A and B may be bad options. Ask: "Is this actually a binary choice, or are we missing option C/D/E?"
|
||||
|
||||
7. **Test synthesis with adversarial roles**: Before finalizing, inhabit each original position and critique the synthesis. Would a partisan of A/B accept it, or see it as capitulation? If synthesis can't survive friendly fire, strengthen it.
|
||||
|
||||
**Common pitfalls:**
|
||||
|
||||
- ❌ **Strawmanning**: "Position A naively believes X" (uncharitable). Instead: "Position A prioritizes Y principle because..."
|
||||
- ❌ **False balance**: Steelmanning doesn't require treating bad-faith arguments as if made in good faith. If one position is empirically wrong or logically inconsistent, note this after steelmanning.
|
||||
- ❌ **Mushy middle**: "Do a little of both" is not synthesis. Synthesis finds NEW approach, not diluted mix.
|
||||
- ❌ **Ignoring power dynamics**: Some debates aren't idea conflicts—they're conflicts of interest. Synthesis may not resolve structural problems.
|
||||
- ❌ **Analysis paralysis**: Dialectical mapping is a tool for decision-making, not an end. Set time bounds, converge on synthesis, decide.
|
||||
|
||||
## Quick Reference
|
||||
|
||||
**Key resources:**
|
||||
|
||||
- **[resources/template.md](resources/template.md)**: Steelmanning template, tradeoff matrix template, synthesis structure
|
||||
- **[resources/methodology.md](resources/methodology.md)**: Advanced techniques (multi-party dialectics, principle hierarchies, Toulmin argumentation for steelmanning, synthesis patterns)
|
||||
- **[resources/evaluators/rubric_dialectical_mapping_steelmanning.json](resources/evaluators/rubric_dialectical_mapping_steelmanning.json)**: Quality criteria for steelmans and synthesis
|
||||
|
||||
**Typical workflow time:**
|
||||
|
||||
- Simple binary debate (2 positions, clear principles): 20-30 minutes
|
||||
- Complex multi-stakeholder debate: 45-60 minutes
|
||||
- Strategic frameworks (long-term decisions): 60-90 minutes
|
||||
|
||||
**When to escalate:**
|
||||
|
||||
- More than 2 positions (multi-party dialectics)
|
||||
- Nested tradeoffs (position A itself is a synthesis of A1 vs A2)
|
||||
- Empirical questions disguised as value debates
|
||||
- Bad faith arguments (not resolvable via steelmanning)
|
||||
→ Use [resources/methodology.md](resources/methodology.md) for these advanced cases
|
||||
|
||||
**Inputs required:**
|
||||
|
||||
- **Debate topic**: The decision or question being debated
|
||||
- **Position A (Thesis)**: One side of the binary
|
||||
- **Position B (Antithesis)**: The opposing side
|
||||
- **Context** (optional): Constraints, stakeholders, decision criteria
|
||||
|
||||
**Outputs produced:**
|
||||
|
||||
- `dialectical-mapping-steelmanning.md`: Complete analysis with steelmanned positions, tradeoff matrix, synthesis, and recommendations
|
||||
@@ -0,0 +1,291 @@
|
||||
{
|
||||
"criteria": [
|
||||
{
|
||||
"name": "Debate Framing & Context",
|
||||
"description": "Is the debate topic clearly defined with both positions, binary framing justification, and context?",
|
||||
"scoring": {
|
||||
"1": "Vague topic. Positions not clearly distinguished. No explanation of why binary. Context missing or superficial. Stakes unclear.",
|
||||
"3": "Topic stated but could be more specific. Positions identified but may lack precision. Binary framing mentioned but not deeply explored. Some context provided. Stakes partially clear.",
|
||||
"5": "Precise topic definition. Both positions clearly articulated with concise statements. Binary framing explained (resource constraint, timing, architecture lock-in, cultural values, or zero-sum perception). Rich context: stakeholders, constraints, deadline, impact. Stakes crystal clear."
|
||||
}
|
||||
},
|
||||
{
|
||||
"name": "Steelman Quality - Position A",
|
||||
"description": "Is Position A presented in its strongest, most charitable form?",
|
||||
"scoring": {
|
||||
"1": "Strawman or weak representation. Arguments cherry-picked or misrepresented. Underlying principle not identified. Tradeoffs ignored or minimized. Adherent would not recognize as fair.",
|
||||
"3": "Decent representation but could be stronger. Principle identified but may be surface-level. Arguments present but not in strongest form. Tradeoffs mentioned but not fully explored. Adherent might quibble with details.",
|
||||
"5": "Exemplary steelman. Underlying principle clearly articulated (speed, quality, freedom, safety, equity, efficiency, etc.). Best 3-5 arguments presented with evidence/examples. Legitimate tradeoffs explicitly acknowledged. Adherent would say 'yes, that's exactly why we support this position.'"
|
||||
}
|
||||
},
|
||||
{
|
||||
"name": "Steelman Quality - Position B",
|
||||
"description": "Is Position B presented in its strongest, most charitable form?",
|
||||
"scoring": {
|
||||
"1": "Strawman or weak representation. Arguments cherry-picked or misrepresented. Underlying principle not identified. Tradeoffs ignored or minimized. Adherent would not recognize as fair.",
|
||||
"3": "Decent representation but could be stronger. Principle identified but may be surface-level. Arguments present but not in strongest form. Tradeoffs mentioned but not fully explored. Adherent might quibble with details.",
|
||||
"5": "Exemplary steelman. Underlying principle clearly articulated. Best 3-5 arguments presented with evidence/examples. Legitimate tradeoffs explicitly acknowledged. Adherent would say 'yes, that's exactly why we support this position.'"
|
||||
}
|
||||
},
|
||||
{
|
||||
"name": "Steelman Symmetry",
|
||||
"description": "Are both positions given equal charitable treatment and depth of analysis?",
|
||||
"scoring": {
|
||||
"1": "Asymmetric treatment. One position clearly favored (longer, stronger arguments, tradeoffs downplayed). Other position treated superficially or uncharitably. Bias obvious.",
|
||||
"3": "Mostly symmetric but minor imbalances. One position may have slightly more detail or stronger arguments. Not egregious but noticeable. Effort toward fairness evident.",
|
||||
"5": "Perfect symmetry in treatment. Both positions get equal depth, charity, and rigor. Similar length, argument strength, tradeoff acknowledgment. Neutral observer couldn't detect analyst's preference from steelmans alone. If asymmetry exists, it's acknowledged explicitly (e.g., 'Position A has stronger empirical support but Position B raises valid normative concerns')."
|
||||
}
|
||||
},
|
||||
{
|
||||
"name": "Principle Identification",
|
||||
"description": "Are underlying principles (values, goals) identified vs surface preferences?",
|
||||
"scoring": {
|
||||
"1": "Principles not identified. Analysis stays at surface level ('Position A wants X, Position B wants Y'). No exploration of WHY they want X/Y. Values implicit or missing.",
|
||||
"3": "Principles mentioned but not deeply explored. May identify one principle per position but miss deeper values or terminal vs instrumental distinction. Partial insight into motivations.",
|
||||
"5": "Deep principle identification. Clearly distinguishes terminal principles (ends: freedom, wellbeing, fairness) vs instrumental (means: speed, cost, consistency). Explains WHY each position values what it does. Identifies where principles genuinely conflict vs where apparent conflict masks false dichotomy."
|
||||
}
|
||||
},
|
||||
{
|
||||
"name": "Tradeoff Matrix Quality",
|
||||
"description": "Is the tradeoff matrix comprehensive, multi-dimensional, and quantitative where possible?",
|
||||
"scoring": {
|
||||
"1": "No tradeoff matrix or very sparse. Single-dimension comparison. No quantification. Best/worst cases missing. Risks not identified.",
|
||||
"3": "Tradeoff matrix present but limited. 2-4 dimensions compared. Some quantification (cost, time) but mostly qualitative. Best/worst cases mentioned. Primary risks identified but not comprehensively.",
|
||||
"5": "Comprehensive tradeoff matrix. 5+ dimensions (principle alignment, speed, quality, cost, flexibility, risk, etc.). Quantitative where possible ($ cost, timeline, metrics). Best case and worst case for each position articulated. Primary risks for each position and synthesis clearly identified. Matrix reveals insights about the tradeoff structure."
|
||||
}
|
||||
},
|
||||
{
|
||||
"name": "Synthesis Quality - Transcendence",
|
||||
"description": "Does synthesis transcend the binary (not just compromise) with higher-order principle?",
|
||||
"scoring": {
|
||||
"1": "No synthesis or mushy middle ('do a little of both'). 50/50 split with no principled rationale. Compromise, not transcendence. No higher-order principle identified.",
|
||||
"3": "Synthesis present but may be compromise-adjacent. Some principled thinking but not fully transcendent. Higher-order principle mentioned but not deeply developed. May lean toward one position with minor adjustments rather than true synthesis.",
|
||||
"5": "Exemplary synthesis that genuinely transcends binary. Higher-order principle clearly articulated (meta-goal both positions serve). Explains how synthesis preserves core value of Position A AND core value of Position B. Uses advanced pattern (temporal, conditional, dimensional separation, reframe, compensating controls). Not split-the-difference—creates new value. Adherents of both positions can see their values represented."
|
||||
}
|
||||
},
|
||||
{
|
||||
"name": "New Tradeoffs - Explicit & Honest",
|
||||
"description": "Are the new tradeoffs of synthesis vs pure positions explicitly stated?",
|
||||
"scoring": {
|
||||
"1": "New tradeoffs ignored. Synthesis presented as 'best of both worlds with no downsides'. No acknowledgment of costs vs pure Position A or B. Unrealistic or naive.",
|
||||
"3": "New tradeoffs mentioned but not fully explored. Some costs identified but may downplay them. Acknowledges synthesis isn't perfect but doesn't deeply analyze what's sacrificed.",
|
||||
"5": "Brutally honest about new tradeoffs. Explicitly states: (1) What synthesis sacrifices vs pure Position A, (2) What it sacrifices vs pure Position B, (3) New challenges synthesis introduces. Decision criteria provided: when to use synthesis vs when pure positions are better. No pretense that synthesis is universally superior."
|
||||
}
|
||||
},
|
||||
{
|
||||
"name": "Empirical-Normative Separation",
|
||||
"description": "Are factual questions separated from value questions? Are empirical cruxes identified?",
|
||||
"scoring": {
|
||||
"1": "Empirical and normative questions conflated. Treats value judgments as facts or vice versa. No identification of which claims are testable vs which require philosophical resolution.",
|
||||
"3": "Some awareness of empirical vs normative distinction. May identify a few factual cruxes. Separation present but not systematic. Could be clearer about which disagreements are about facts vs values.",
|
||||
"5": "Crisp separation of empirical questions (resolvable with evidence) from normative questions (value judgments). Identifies empirical cruxes explicitly ('if we had data showing X, would Position A change?'). Synthesis approach is conditional on empirical findings where appropriate. Value conflicts resolved at principle level."
|
||||
}
|
||||
},
|
||||
{
|
||||
"name": "Implementation & Validation",
|
||||
"description": "Does output include actionable recommendations, success criteria, and reassessment triggers?",
|
||||
"scoring": {
|
||||
"1": "No implementation guidance. Analysis ends with synthesis description. No success criteria or metrics. No plan for validation or reassessment. Purely theoretical.",
|
||||
"3": "Basic implementation steps provided. Some success criteria mentioned but may be vague. Partial guidance on validation. Reassessment triggers missing or unclear. Somewhat actionable.",
|
||||
"5": "Comprehensive implementation plan with specific steps. Clear success criteria (quantitative metrics, qualitative signals). Validation approach defined (how to test if synthesis working). Reassessment triggers specified (conditions that would make you revisit or switch approaches). Fully actionable and testable."
|
||||
}
|
||||
}
|
||||
],
|
||||
"minimum_score": 3.5,
|
||||
"guidance_by_debate_type": {
|
||||
"Strategic/Business Decisions": {
|
||||
"target_score": 4.2,
|
||||
"focus_criteria": [
|
||||
"Principle Identification",
|
||||
"Tradeoff Matrix Quality",
|
||||
"Synthesis Quality - Transcendence"
|
||||
],
|
||||
"common_debates": [
|
||||
"Growth vs profitability",
|
||||
"Build vs buy",
|
||||
"Centralization vs decentralization",
|
||||
"Exploration vs exploitation",
|
||||
"First-mover vs fast-follower",
|
||||
"Differentiation vs cost leadership"
|
||||
],
|
||||
"quality_indicators": {
|
||||
"excellent": "Synthesis identifies higher-order business principle (e.g., capital efficiency, sustainable competitive advantage, optionality preservation). Tradeoff matrix includes quantitative analysis ($ cost, timeline, risk probability). Decision criteria specify market conditions or thresholds where pure positions dominate.",
|
||||
"sufficient": "Both positions steelmanned with business rationale. Tradeoffs identified (financial, competitive, operational). Synthesis provides actionable approach with some conditionality.",
|
||||
"insufficient": "Surface-level 'pros and cons'. No deep principle identification. Synthesis is vague ('balance growth and profitability') without specifics. Missing quantitative analysis."
|
||||
}
|
||||
},
|
||||
"Technical/Engineering Decisions": {
|
||||
"target_score": 4.0,
|
||||
"focus_criteria": [
|
||||
"Steelman Quality - Position A",
|
||||
"Steelman Quality - Position B",
|
||||
"Empirical-Normative Separation"
|
||||
],
|
||||
"common_debates": [
|
||||
"Monolith vs microservices",
|
||||
"SQL vs NoSQL",
|
||||
"Build vs integrate third-party",
|
||||
"Static vs dynamic typing",
|
||||
"Sync vs async patterns",
|
||||
"Normalized vs denormalized data"
|
||||
],
|
||||
"quality_indicators": {
|
||||
"excellent": "Steelmans include technical specifics (performance characteristics, scaling properties, failure modes). Empirical questions separated ('Does this pattern actually improve latency?' testable). Synthesis addresses technical constraints and provides migration path or hybrid architecture with clear boundaries.",
|
||||
"sufficient": "Both technical approaches explained with rationale. Tradeoffs cover performance, maintainability, complexity. Synthesis is technically sound with some specifics.",
|
||||
"insufficient": "Superficial technical understanding. Arguments based on hype or dogma rather than engineering principles. Synthesis lacks technical detail or is infeasible given constraints."
|
||||
}
|
||||
},
|
||||
"Policy/Governance Debates": {
|
||||
"target_score": 4.3,
|
||||
"focus_criteria": [
|
||||
"Principle Identification",
|
||||
"Empirical-Normative Separation",
|
||||
"Synthesis Quality - Transcendence"
|
||||
],
|
||||
"common_debates": [
|
||||
"Privacy vs security",
|
||||
"Freedom vs safety",
|
||||
"Efficiency vs equity",
|
||||
"Regulation vs free market",
|
||||
"Individual rights vs collective welfare",
|
||||
"Transparency vs confidentiality"
|
||||
],
|
||||
"quality_indicators": {
|
||||
"excellent": "Terminal principles clearly identified (freedom, welfare, fairness, security as ends not means). Empirical questions isolated (e.g., 'Do backdoors actually improve security?' vs 'Should we value privacy or security more?'). Synthesis addresses power dynamics if present. Handles edge cases and second-order effects. Acknowledges pluralistic values.",
|
||||
"sufficient": "Both positions presented with philosophical grounding. Value conflicts acknowledged. Synthesis attempts to honor both principles with some tradeoffs.",
|
||||
"insufficient": "Treats value questions as resolvable via evidence. Ignores one side's core principle. Synthesis favors one position without acknowledging this is value judgment not logical necessity. Misses power dynamics."
|
||||
}
|
||||
},
|
||||
"Product/Design Decisions": {
|
||||
"target_score": 3.8,
|
||||
"focus_criteria": [
|
||||
"Tradeoff Matrix Quality",
|
||||
"Synthesis Quality - Transcendence",
|
||||
"Implementation & Validation"
|
||||
],
|
||||
"common_debates": [
|
||||
"Simple vs powerful",
|
||||
"Consistency vs flexibility",
|
||||
"Mobile-first vs desktop-first",
|
||||
"Opinionated vs configurable",
|
||||
"Accessibility vs aesthetics",
|
||||
"Guided flows vs free exploration"
|
||||
],
|
||||
"quality_indicators": {
|
||||
"excellent": "Tradeoff matrix considers user segments (novice vs power users), use cases (frequency, complexity), and metrics (usability, satisfaction, task completion). Synthesis often uses dimensional separation (simple by default, power mode available) or progressive disclosure. Implementation includes A/B test plan or user research validation.",
|
||||
"sufficient": "Both design philosophies explained with user impact. Tradeoffs cover usability, complexity, accessibility. Synthesis provides concrete design approach with some user-centricity.",
|
||||
"insufficient": "Arguments based on designer preference not user needs. No consideration of different user segments. Synthesis lacks specificity or testability. No validation plan."
|
||||
}
|
||||
},
|
||||
"Team/Process Debates": {
|
||||
"target_score": 3.9,
|
||||
"focus_criteria": [
|
||||
"Steelman Symmetry",
|
||||
"Synthesis Quality - Transcendence",
|
||||
"Implementation & Validation"
|
||||
],
|
||||
"common_debates": [
|
||||
"Autonomy vs alignment",
|
||||
"Speed vs quality",
|
||||
"Consensus vs decisive leadership",
|
||||
"Planning vs emergent strategy",
|
||||
"Standardization vs customization",
|
||||
"Individual vs collective accountability"
|
||||
],
|
||||
"quality_indicators": {
|
||||
"excellent": "Symmetry critical—team debates often have power dynamics and advocates. Synthesis addresses cultural fit, change management, and feedback loops. Implementation specifies who does what, how decisions escalate, how to measure success (cycle time, quality metrics, team satisfaction).",
|
||||
"sufficient": "Both approaches considered fairly. Tradeoffs include team dynamics and culture. Synthesis is concrete with roles and processes. Some implementation detail.",
|
||||
"insufficient": "One approach dismissed as 'old way' or 'too slow'. Ignores team context (size, maturity, domain). Synthesis vague ('find the right balance') without specifics."
|
||||
}
|
||||
}
|
||||
},
|
||||
"guidance_by_complexity": {
|
||||
"Simple (Binary, Clear Principles)": {
|
||||
"target_score": 3.8,
|
||||
"sufficient_depth": "Both positions steelmanned with underlying principles. Tradeoff matrix with 3-5 dimensions. Synthesis uses one of five common patterns (temporal, conditional, dimensional, higher-order, compensating). New tradeoffs stated. Decision criteria for when to use synthesis vs pure positions.",
|
||||
"key_requirements": [
|
||||
"Charitable steelmans (adherents would agree)",
|
||||
"Principle identification (not just surface preferences)",
|
||||
"Tradeoff matrix (quantitative where possible)",
|
||||
"Synthesis pattern applied (not just compromise)",
|
||||
"New tradeoffs explicitly acknowledged",
|
||||
"Decision criteria (when synthesis appropriate)"
|
||||
]
|
||||
},
|
||||
"Moderate (Multi-dimensional, Some Complexity)": {
|
||||
"target_score": 4.0,
|
||||
"sufficient_depth": "Steelmans use Toulmin model (claim, data, warrant, backing, qualifier, rebuttal). Principle mapping distinguishes terminal vs instrumental values. Tradeoff matrix multi-dimensional (5+ dimensions). Synthesis addresses empirical cruxes and normative conflicts separately. Validation plan includes adversarial testing.",
|
||||
"key_requirements": [
|
||||
"Toulmin-enhanced steelmans (warrants explicit, qualifiers included)",
|
||||
"Terminal vs instrumental principle distinction",
|
||||
"Multi-dimensional tradeoff matrix (5+ dimensions, quantitative)",
|
||||
"Empirical-normative separation",
|
||||
"Synthesis with advanced pattern (may combine multiple patterns)",
|
||||
"Adversarial testing or edge case analysis for validation"
|
||||
]
|
||||
},
|
||||
"Complex (Multi-party, Nested, Power Dynamics)": {
|
||||
"target_score": 4.2,
|
||||
"sufficient_depth": "Multi-party synthesis (pairwise or principle clustering). Nested dialectics if positions are themselves syntheses. Power dynamics made explicit and addressed in synthesis. Empirical cruxes identified with conditional synthesis. Unintended consequences analyzed. Stability testing (does synthesis collapse under pressure?). Reassessment triggers specified.",
|
||||
"key_requirements": [
|
||||
"Multi-party dialectics handled systematically (not just 'winner' of pairwise)",
|
||||
"Nested dialectics resolved at meta-level if applicable",
|
||||
"Power dynamics explicit (who benefits, material stakes, structural issues)",
|
||||
"Empirical questions separated with conditional synthesis based on evidence",
|
||||
"Second-order effects and unintended consequences analyzed",
|
||||
"Synthesis stability testing (edge cases, perverse incentives, long-term dynamics)",
|
||||
"Monitoring and reassessment triggers (when to revisit decision)"
|
||||
]
|
||||
}
|
||||
},
|
||||
"common_failure_modes": [
|
||||
{
|
||||
"failure": "Strawmanning (weak representation of positions)",
|
||||
"symptom": "Arguments cherry-picked, misrepresented, or presented in weakest form. Phrases like 'Position A naively believes...' or 'Position B ignores...' without charity. Tradeoffs hidden or minimized.",
|
||||
"detection": "Ask: 'Would someone who holds this position recognize this as fair?' If answer is no or maybe, it's a strawman. Check for asymmetry—one position gets strong arguments, other gets weak.",
|
||||
"fix": "Re-read position from advocate's perspective. Find strongest arguments and best evidence. Present version that advocate would agree with. Use Toulmin model: identify warrants, backing, and qualifiers. Acknowledge legitimate tradeoffs explicitly."
|
||||
},
|
||||
{
|
||||
"failure": "False dichotomy not questioned",
|
||||
"symptom": "Binary framing accepted without scrutiny. Analysis proceeds directly to steelmanning A vs B without asking 'is this actually a binary choice?' Synthesis stays within A-B space rather than exploring option C/D/E.",
|
||||
"detection": "Check debate framing section. Does it ask WHY this is framed as binary? Does it question if binary is real or manufactured? Does synthesis explore outside A-B space?",
|
||||
"fix": "Add 'False dichotomy check' step. Ask: 'Are A and B mutually exclusive? Can we do both (temporal, conditional, dimensional separation)? Are there option C/D/E we're ignoring?' Challenge the premise before synthesizing."
|
||||
},
|
||||
{
|
||||
"failure": "Synthesis is mushy middle (compromise, not transcendence)",
|
||||
"symptom": "Synthesis is 'do a little of both' or 50/50 split without principled rationale. No higher-order principle identified. Phrases like 'balance A and B' or 'find the right mix' without specifics. Doesn't explain why synthesis is better than pure positions.",
|
||||
"detection": "Check if synthesis specifies HOW it preserves each position's core value. Does it identify higher-order principle? Does it use a synthesis pattern (temporal, conditional, dimensional, reframe, compensating controls)? If answers are vague, it's mushy middle.",
|
||||
"fix": "Identify higher-order principle (meta-goal both positions serve). Choose synthesis pattern: temporal (sequence), conditional (context-dependent), dimensional (orthogonal axes), reframe (better tactic for shared goal), or compensating controls (lean + safeguard). Be specific about structure, not just 'balance.'"
|
||||
},
|
||||
{
|
||||
"failure": "New tradeoffs ignored or hidden",
|
||||
"symptom": "Synthesis presented as 'best of both worlds' with no downsides. No acknowledgment of what's sacrificed vs pure Position A or B. Analysis stops after synthesis description without discussing costs.",
|
||||
"detection": "Look for explicit section on 'New Tradeoffs' or 'What Synthesis Sacrifices'. If missing or superficial ('minor inconvenience'), this failure is present.",
|
||||
"fix": "Be brutally honest. Add section: (1) What synthesis sacrifices vs pure Position A, (2) What it sacrifices vs pure Position B, (3) New challenges synthesis introduces. Provide decision criteria: when pure positions are better than synthesis."
|
||||
},
|
||||
{
|
||||
"failure": "Empirical and normative questions conflated",
|
||||
"symptom": "Treating value judgments as facts ('Position A is clearly better') or vice versa ('this is just a matter of opinion' when empirical evidence exists). No separation of testable claims from philosophical claims.",
|
||||
"detection": "Check if debate involves mix of factual and value questions. Are empirical cruxes identified ('if data showed X, would position change')? Are value conflicts resolved via principle-level reasoning?",
|
||||
"fix": "Separate explicitly: (1) Empirical questions (what IS, testable with evidence), (2) Normative questions (what SHOULD BE, require value judgments). Identify cruxes for empirical claims. Resolve normative conflicts via principle identification and higher-order synthesis. Make synthesis conditional on empirical findings where appropriate."
|
||||
},
|
||||
{
|
||||
"failure": "Steelman asymmetry (bias toward one position)",
|
||||
"symptom": "One position gets much more depth, stronger arguments, charitable interpretation. Other position treated superficially or uncharitably. Length imbalance, tradeoff minimization for favored position.",
|
||||
"detection": "Compare word count, argument strength, tradeoff acknowledgment for Position A vs B. Neutral observer shouldn't detect analyst's preference from steelmans alone. If obvious which position analyst prefers, asymmetry exists.",
|
||||
"fix": "Review both steelmans side-by-side. Ensure equal length, depth, charity. If one position truly has weaker arguments or evidence, acknowledge this explicitly AFTER steelmanning both fairly. Can still lean toward stronger position in synthesis, but not in steelmanning phase."
|
||||
},
|
||||
{
|
||||
"failure": "No implementation or validation plan",
|
||||
"symptom": "Analysis is purely theoretical. Ends with synthesis description. No guidance on how to execute, measure success, or validate if working. No reassessment triggers.",
|
||||
"detection": "Check for implementation steps, success criteria, metrics, validation approach, reassessment triggers. If missing or vague ('monitor and adjust'), this failure is present.",
|
||||
"fix": "Add: (1) Implementation—specific steps to execute synthesis, (2) Success criteria—quantitative metrics and qualitative signals, (3) Validation—how to test if synthesis working (experiments, measurements), (4) Reassessment triggers—conditions that would make you switch approaches. Make it actionable."
|
||||
},
|
||||
{
|
||||
"failure": "Power dynamics ignored",
|
||||
"symptom": "Debate involves parties with different power, resources, or incentives, but analysis treats as pure idea conflict. No mention of who benefits materially, asymmetric stakes, or structural issues. Synthesis doesn't address power imbalance.",
|
||||
"detection": "Ask: Do parties have different power? Are there conflicts of interest (one position benefits advocate materially)? Are stakes symmetric? If power dynamics present but not acknowledged, this failure exists.",
|
||||
"fix": "Make power dynamics explicit: who benefits from each position, material stakes, historical power imbalances. Steelman arguments independent of advocate motives. If synthesis leans toward powerful party, include safeguards for less powerful (e.g., portable benefits, collective bargaining, classification thresholds). Address structural issues, not just ideas."
|
||||
}
|
||||
]
|
||||
}
|
||||
436
skills/dialectical-mapping-steelmanning/resources/methodology.md
Normal file
436
skills/dialectical-mapping-steelmanning/resources/methodology.md
Normal file
@@ -0,0 +1,436 @@
|
||||
# Dialectical Mapping & Steelmanning - Advanced Methodology
|
||||
|
||||
## Workflow
|
||||
|
||||
Copy this checklist for advanced cases:
|
||||
|
||||
```
|
||||
Advanced Dialectical Mapping:
|
||||
- [ ] Step 1: Assess complexity (multi-party, nested, empirical-normative mix)
|
||||
- [ ] Step 2: Apply advanced steelmanning techniques
|
||||
- [ ] Step 3: Map principle hierarchies and conflicts
|
||||
- [ ] Step 4: Generate synthesis using advanced patterns
|
||||
- [ ] Step 5: Validate synthesis with adversarial testing
|
||||
```
|
||||
|
||||
**Step 1: Assess complexity**
|
||||
|
||||
Identify which advanced techniques apply: multi-party dialectics (>2 positions), nested dialectics (positions are themselves syntheses), principle hierarchies (multiple conflicting values), empirical-normative mix (fact vs value questions), or power dynamics (conflicts of interest). See technique selection below.
|
||||
|
||||
**Step 2: Apply advanced steelmanning**
|
||||
|
||||
Use [Toulmin Argumentation Model](#1-toulmin-argumentation-model) to strengthen steelmans, identify implicit warrants, and expose assumptions. Check for [Common Fallacies](#2-common-fallacies-in-dialectical-reasoning) that weaken arguments.
|
||||
|
||||
**Step 3: Map principle hierarchies**
|
||||
|
||||
For multi-dimensional tradeoffs, use [Principle Mapping](#3-principle-mapping-hierarchies) to structure values. Identify which principles are means vs ends, where they conflict, and potential higher-order principles.
|
||||
|
||||
**Step 4: Generate synthesis**
|
||||
|
||||
Apply advanced patterns: [Multi-Party Synthesis](#4-multi-party-dialectics), [Nested Dialectics](#5-nested-dialectics), [Empirical-Normative Separation](#6-empirical-vs-normative-questions), or [Power Dynamics Handling](#7-power-dynamics-and-conflicts-of-interest).
|
||||
|
||||
**Step 5: Validate synthesis**
|
||||
|
||||
Use [Synthesis Validation Techniques](#8-synthesis-validation-techniques) including adversarial testing, edge case analysis, and unintended consequences check.
|
||||
|
||||
---
|
||||
|
||||
## 1. Toulmin Argumentation Model
|
||||
|
||||
**Use when**: Steelmanning complex arguments with implicit assumptions, strengthening argument structure, or identifying weak points charitably.
|
||||
|
||||
### Model Structure
|
||||
|
||||
**Claim** (C): What position asserts
|
||||
**Data** (D): Evidence supporting claim
|
||||
**Warrant** (W): Logical connection between data and claim (often implicit)
|
||||
**Backing** (B): Support for the warrant
|
||||
**Qualifier** (Q): Degree of certainty (always, probably, unless...)
|
||||
**Rebuttal** (R): Conditions under which claim doesn't hold
|
||||
|
||||
### Application to Steelmanning
|
||||
|
||||
**Standard steelman**: "Position A wants X because Y."
|
||||
|
||||
**Toulmin-enhanced steelman**:
|
||||
- **Claim**: X is the right choice
|
||||
- **Data**: Y (evidence, context, examples)
|
||||
- **Warrant**: "Given Y, X follows because [logical connection]"
|
||||
- **Backing**: [Why the warrant is valid—theory, principle, precedent]
|
||||
- **Qualifier**: "X is right **unless** [edge cases, conditions]"
|
||||
- **Rebuttal**: "X fails if [specific scenarios]"
|
||||
|
||||
### Example
|
||||
|
||||
**Topic**: Should we use microservices architecture?
|
||||
|
||||
**Basic steelman (weak)**: "Microservices scale better."
|
||||
|
||||
**Toulmin steelman (strong)**:
|
||||
- **Claim**: We should use microservices
|
||||
- **Data**: Our user base will grow 10x next year; different services have different scaling needs (API: 100x, admin: 2x)
|
||||
- **Warrant**: When services have heterogeneous scaling requirements, independent deployability enables cost-efficient scaling (scale what needs scaling, not entire monolith)
|
||||
- **Backing**: Economic principle (marginal cost optimization) + engineering precedent (Netflix, Uber scaled via microservices)
|
||||
- **Qualifier**: Microservices are appropriate **if** org can handle operational complexity (monitoring, distributed tracing, service mesh)
|
||||
- **Rebuttal**: Falls apart if team < 20 engineers (coordination overhead exceeds benefits) or if services are tightly coupled (defeats independence)
|
||||
|
||||
**Result**: Stronger steelman that acknowledges scope conditions and failure modes.
|
||||
|
||||
---
|
||||
|
||||
## 2. Common Fallacies in Dialectical Reasoning
|
||||
|
||||
### Strawman (The Problem Steelmanning Solves)
|
||||
|
||||
**Definition**: Misrepresenting opponent's position to make it easier to attack.
|
||||
**Example**: "Position A wants speed, so they don't care about quality."
|
||||
**Fix**: Steelman: "Position A prioritizes speed because early market entry captures compounding advantages, accepting higher initial defect risk as a calculated tradeoff."
|
||||
|
||||
### False Dichotomy
|
||||
|
||||
**Definition**: Framing as binary choice when other options exist.
|
||||
**Example**: "We must choose growth OR profitability."
|
||||
**Detection**: Ask: "Are these actually mutually exclusive? Can we do both sequentially, conditionally, or on different dimensions?"
|
||||
**Fix**: Synthesis via temporal, conditional, or dimensional separation patterns.
|
||||
|
||||
### False Equivalence
|
||||
|
||||
**Definition**: Treating unequal positions as equally valid.
|
||||
**Example**: "Both sides have good points" when one position has stronger evidence.
|
||||
**Fix**: Steelman both, but acknowledge asymmetry. Synthesis can lean heavily toward stronger position while incorporating weaker position's safeguards.
|
||||
|
||||
### Slippery Slope
|
||||
|
||||
**Definition**: Assuming small step inevitably leads to extreme outcome without justification.
|
||||
**Example**: "If we allow any technical debt, we'll end up with unmaintainable codebase."
|
||||
**Fix**: Identify intermediate stopping points, decision criteria, compensating controls. "Technical debt is acceptable if: (1) repayment plan exists, (2) debt tracked publicly, (3) quarterly budget for paydown."
|
||||
|
||||
### Appeal to Extremes
|
||||
|
||||
**Definition**: Judging position by worst-case misuse rather than typical application.
|
||||
**Example**: "Decentralization leads to chaos" (judging by anarchy rather than reasonable autonomy).
|
||||
**Fix**: Steelman with realistic implementation, boundaries, guardrails. "Decentralized execution with centralized strategy and escalation paths."
|
||||
|
||||
### Begging the Question
|
||||
|
||||
**Definition**: Assuming the conclusion in the premise.
|
||||
**Example**: "We should centralize because centralization works better."
|
||||
**Detection**: Check if warrant is circular. Does argument provide independent evidence?
|
||||
**Fix**: Identify actual evidence or principle. "Centralize when coordination costs of decentralization exceed benefits of local optimization—measurable via cycle time, error rates, duplication metrics."
|
||||
|
||||
---
|
||||
|
||||
## 3. Principle Mapping & Hierarchies
|
||||
|
||||
**Use when**: Multiple conflicting values (speed, quality, cost, equity, freedom), need to prioritize principles, or synthesis requires identifying higher-order principle.
|
||||
|
||||
### Principle Types
|
||||
|
||||
**Instrumental principles** (means to an end): Speed, cost-efficiency, consistency, simplicity
|
||||
**Terminal principles** (ends in themselves): Human wellbeing, freedom, fairness, truth
|
||||
|
||||
### Mapping Process
|
||||
|
||||
1. **List all principles** invoked by both positions
|
||||
2. **Categorize**: Instrumental vs terminal
|
||||
3. **Identify conflicts**: Which principles oppose each other?
|
||||
4. **Find hierarchy**: Are conflicting principles actually at different levels? Can instrumental principle be reframed to serve terminal principle better?
|
||||
|
||||
### Example
|
||||
|
||||
**Debate**: Privacy vs Security (government encryption backdoors)
|
||||
|
||||
**Position A (Privacy)**: No backdoors. Principles: Individual freedom, protection from overreach, right to privacy.
|
||||
|
||||
**Position B (Security)**: Backdoors for law enforcement. Principles: Public safety, crime prevention, national security.
|
||||
|
||||
**Principle mapping**:
|
||||
- **Terminal principles**: Individual freedom (A), Public safety (B)
|
||||
- **Conflict**: These ARE genuinely opposed (zero-sum in some cases)
|
||||
- **Instrumental question**: Do backdoors actually increase security? (Empirical question—creates vulnerability for adversaries too)
|
||||
|
||||
**Higher-order principle**: "Maximize security for law-abiding citizens"
|
||||
|
||||
**Synthesis**: No backdoors (protects against authoritarian abuse and adversarial exploitation), but strong metadata analysis, international cooperation on criminal activity, investment in other investigative techniques. Accepts that some investigations are harder, rejects that backdoors materially improve security given systemic risks.
|
||||
|
||||
**Result**: Synthesis leans toward Privacy position but for Security reasons (backdoors undermine security). Reframes apparent conflict.
|
||||
|
||||
---
|
||||
|
||||
## 4. Multi-Party Dialectics
|
||||
|
||||
**Use when**: More than 2 positions (e.g., A vs B vs C vs D).
|
||||
|
||||
### Approach 1: Pairwise Comparison
|
||||
|
||||
1. Pick two dominant positions (A vs B)
|
||||
2. Synthesize to C'
|
||||
3. Compare C' vs remaining position C
|
||||
4. Synthesize to C''
|
||||
5. Repeat until all positions incorporated
|
||||
|
||||
**Limitation**: Order-dependent. Different pairing order may yield different synthesis.
|
||||
|
||||
### Approach 2: Principle Clustering
|
||||
|
||||
1. Map all positions to underlying principles
|
||||
2. Group positions by shared principles
|
||||
3. Identify principle conflicts (not position conflicts)
|
||||
4. Synthesize at principle level
|
||||
5. Derive concrete approach from principle synthesis
|
||||
|
||||
**Example**:
|
||||
|
||||
**Debate**: How to structure engineering team (100 engineers)?
|
||||
|
||||
**Position A**: Centralized platform team builds shared services, product teams consume
|
||||
**Position B**: Fully autonomous product teams, each owns full stack
|
||||
**Position C**: Matrix org, engineers report to tech lead (for skills) and product manager (for direction)
|
||||
**Position D**: Rotate engineers across teams quarterly for knowledge sharing
|
||||
|
||||
**Principle mapping**:
|
||||
- A values: Consistency, reuse, economies of scale (instrumental)
|
||||
- B values: Autonomy, speed, end-to-end ownership (instrumental)
|
||||
- C values: Skill development, career growth (instrumental)
|
||||
- D values: Knowledge distribution, bus factor reduction (instrumental)
|
||||
|
||||
**Terminal goal (all positions)**: Maximize product delivery velocity AND quality
|
||||
|
||||
**Synthesis**: Hybrid model—
|
||||
- Platform team for true shared infrastructure (data pipeline, auth, payments—things with network effects or compliance requirements)
|
||||
- Product teams own full stack for product-specific features (autonomy where it matters)
|
||||
- Guilds for skill development (engineers self-organize by discipline—frontend, backend, data)
|
||||
- 6-month rotations optional for L3-L4 engineers (knowledge sharing without mandating churn)
|
||||
|
||||
**Result**: Incorporates elements from all four positions, structured by principle (platform for scale, autonomy for speed, guilds for growth, optional rotation for knowledge).
|
||||
|
||||
---
|
||||
|
||||
## 5. Nested Dialectics
|
||||
|
||||
**Use when**: One of the "positions" is itself a synthesis of sub-positions. Positions have internal contradictions.
|
||||
|
||||
### Structure
|
||||
|
||||
**Position A**: Synthesis of A1 vs A2
|
||||
**Position B**: Synthesis of B1 vs B2
|
||||
**Meta-synthesis**: Resolve A vs B at higher level
|
||||
|
||||
### Example
|
||||
|
||||
**Debate**: How should AI companies handle safety vs capability development?
|
||||
|
||||
**Position A (Safety-first)**: Is itself a synthesis of:
|
||||
- A1: Pause all development until alignment solved
|
||||
- A2: Develop capabilities slowly with extensive testing
|
||||
|
||||
**Position B (Capability-first)**: Is itself a synthesis of:
|
||||
- B1: Race to AGI (winner determines future)
|
||||
- B2: Develop capabilities, use them to solve alignment
|
||||
|
||||
**Meta-synthesis approach**:
|
||||
|
||||
1. **Steelman sub-positions** (A1, A2, B1, B2)
|
||||
2. **Identify that A vs B is false dichotomy**: Actually four positions with different combinations of beliefs about:
|
||||
- How solvable is alignment? (A1: very hard, A2/B2: tractable, B1: solve after)
|
||||
- What are risks of delay? (A: manageable, B1: existential, B2: medium)
|
||||
3. **Reframe around cruxes**: Not safety vs capability, but **beliefs about alignment difficulty and delay risks**
|
||||
4. **Synthesis based on empirical assessment**: If alignment progress stalls (trigger), slow capability development (A2). If alignment progresses faster than expected (trigger), maintain pace (B2). Never pure A1 (pause is unilateral disadvantage) or B1 (race without guardrails is reckless).
|
||||
|
||||
**Result**: Conditional synthesis that responds to empirical evidence, not fixed binary.
|
||||
|
||||
---
|
||||
|
||||
## 6. Empirical vs Normative Questions
|
||||
|
||||
**Use when**: Debate mixes factual questions (what IS) with value questions (what SHOULD BE).
|
||||
|
||||
### Separation Strategy
|
||||
|
||||
1. **Decompose debate** into empirical claims vs normative claims
|
||||
2. **Empirical claims**: Resolvable via evidence (in principle). Highlight these as **cruxes**—if evidence changes, does position change?
|
||||
3. **Normative claims**: Value judgments. Require principle-level resolution, not evidence.
|
||||
4. **Synthesize separately**: Agree on facts where possible, then resolve value conflict conditional on facts.
|
||||
|
||||
### Example
|
||||
|
||||
**Debate**: Should we implement remote-first work policy?
|
||||
|
||||
**Mixed arguments**:
|
||||
- Position A: "Remote work hurts productivity" (empirical) + "Companies should maximize productivity" (normative)
|
||||
- Position B: "Remote work improves wellbeing" (empirical) + "Employee wellbeing matters more than marginal productivity" (normative)
|
||||
|
||||
**Separation**:
|
||||
|
||||
**Empirical questions** (resolvable with data):
|
||||
1. Does remote work reduce productivity? (Measure: output metrics, project completion, code quality)
|
||||
2. Does remote work improve wellbeing? (Measure: retention, satisfaction scores, burnout rates)
|
||||
3. Is there variance by role? (e.g., individual contributors vs managers)
|
||||
|
||||
**Normative questions** (value judgments):
|
||||
1. How should we weight productivity vs wellbeing?
|
||||
2. What about hiring advantages (global talent pool)?
|
||||
3. Long-term effects on culture, mentorship, serendipity?
|
||||
|
||||
**Synthesis approach**:
|
||||
- **Resolve empirical questions first**: Run 6-month experiment, measure metrics
|
||||
- **Conditional synthesis based on data**:
|
||||
- If productivity unchanged or improves → Remote-first (preserves wellbeing, no cost)
|
||||
- If productivity declines < 10% but wellbeing/retention improves → Remote-first (accept tradeoff)
|
||||
- If productivity declines > 10% → Hybrid (2-3 days office for collaboration, 2-3 remote for focus)
|
||||
- **Re-evaluate quarterly**: Track metrics, adjust policy
|
||||
|
||||
**Result**: Separates empirical cruxes from value questions, makes synthesis conditional on evidence.
|
||||
|
||||
---
|
||||
|
||||
## 7. Power Dynamics and Conflicts of Interest
|
||||
|
||||
**Use when**: Debate involves parties with different power, resources, or incentives. "Disagreement" may be conflict of interest, not idea conflict.
|
||||
|
||||
### Detection
|
||||
|
||||
**Signs of power dynamics**:
|
||||
- One position benefits speaker/advocate materially (financial, status, control)
|
||||
- Asymmetric stakes (one side risks much more)
|
||||
- Historical power imbalance between advocates
|
||||
- Debate framing obscures who benefits ("rising tide lifts all boats" when boats are different sizes)
|
||||
|
||||
### Approach
|
||||
|
||||
1. **Make power dynamics explicit**: Who benefits from each position? What are material stakes?
|
||||
2. **Separate ideas from interests**: Steelman arguments independent of advocate motives
|
||||
3. **Synthesis must address structural issues**: Can't resolve via "better ideas" if problem is power imbalance
|
||||
4. **Compensating mechanisms**: If synthesis leans toward powerful party's position, include safeguards for less powerful
|
||||
|
||||
### Example
|
||||
|
||||
**Debate**: Should gig economy platforms classify workers as employees vs independent contractors?
|
||||
|
||||
**Position A (Contractor status)**: Flexibility, entrepreneurial freedom, lower costs (enables more work)
|
||||
**Position B (Employee status)**: Benefits, job security, labor protections
|
||||
|
||||
**Power dynamic**: Platforms have massive information/resource advantage, legal teams, lobbying. Workers individually have minimal bargaining power.
|
||||
|
||||
**Steelman both arguments** (independent of power):
|
||||
- A: Flexibility IS valuable for many workers (students, retirees, side hustlers)
|
||||
- B: Protections ARE necessary for full-time gig workers (no healthcare, no unemployment insurance)
|
||||
|
||||
**Synthesis that addresses power**:
|
||||
- **Classification by hours**: <20 hrs/week = contractor (flexibility preserved), ≥20 hrs/week = employee (protections for full-timers)
|
||||
- **Portable benefits**: Industry-wide benefits fund (platforms contribute per worker-hour, workers access regardless of hours)
|
||||
- **Collective bargaining**: Workers can organize without classification change (addresses power imbalance directly)
|
||||
|
||||
**Result**: Synthesis recognizes power dynamic, doesn't just "split difference on ideas."
|
||||
|
||||
---
|
||||
|
||||
## 8. Synthesis Validation Techniques
|
||||
|
||||
### Adversarial Testing
|
||||
|
||||
**Method**: Inhabit each original position and attack the synthesis.
|
||||
|
||||
**Questions to ask as Position A partisan**:
|
||||
- Does synthesis abandon my core principle?
|
||||
- Is this really "synthesis" or capitulation to Position B?
|
||||
- What tradeoffs am I being asked to accept that Position B isn't?
|
||||
|
||||
**Repeat as Position B partisan.**
|
||||
|
||||
**Pass criteria**: Synthesis survives critique from both sides, or critiques cancel out (each side sees it as slight lean toward other, suggesting balance).
|
||||
|
||||
### Edge Case Analysis
|
||||
|
||||
**Method**: Test synthesis with extreme scenarios.
|
||||
|
||||
**Example**: "Move fast with guardrails" synthesis
|
||||
|
||||
**Edge cases**:
|
||||
- High-risk domain (healthcare, finance) → Guardrails insufficient?
|
||||
- Novel technology → Guardrails unknown?
|
||||
- Aggressive competitor → Speed advantage evaporates?
|
||||
|
||||
**Evaluation**: Does synthesis break down in edge cases? If yes, refine with conditional logic ("In high-risk domains, prioritize guardrails over speed").
|
||||
|
||||
### Unintended Consequences
|
||||
|
||||
**Method**: Consider second-order effects, perverse incentives, long-term dynamics.
|
||||
|
||||
**Example**: "Profitable growth" synthesis (grow as fast as unit economics allow)
|
||||
|
||||
**Unintended consequences**:
|
||||
- Focus on profitable channels may miss future market shifts (addressable market shrinks)
|
||||
- Discipline to say no may calcify into risk-aversion
|
||||
- LTV:CAC metric may be gamed (optimize metric, not underlying economics)
|
||||
|
||||
**Fix**: Add monitoring ("Review addressable market quarterly"), decision criteria ("Experiment with 10% budget in unproven channels"), and metric audits ("Validate LTV assumptions annually").
|
||||
|
||||
### Synthesis Stability Test
|
||||
|
||||
**Question**: Is synthesis stable, or does it collapse back to Position A/B under pressure?
|
||||
|
||||
**Example**: "Centralize strategy, decentralize execution"
|
||||
|
||||
**Pressure test**:
|
||||
- Strategy team starts specifying execution details → Collapse to centralization
|
||||
- Execution teams ignore strategy → Collapse to full autonomy
|
||||
- Gray-zone decisions escalate constantly → Synthesis unworkable
|
||||
|
||||
**Fix**: Define clear boundaries ("Strategy sets goals and constraints, not tactics"), escalation criteria ("Escalate if execution conflicts with cross-team dependency"), and feedback loop ("Execution teams input to strategy quarterly").
|
||||
|
||||
---
|
||||
|
||||
## 9. Advanced Synthesis Patterns
|
||||
|
||||
### Pattern 6: Principle Inversion
|
||||
|
||||
**Structure**: Both positions optimize for same principle but in opposite ways. Synthesis finds third approach to principle.
|
||||
|
||||
**Example**: "Maximize developer productivity"
|
||||
- Position A: Remove all process (friction kills productivity)
|
||||
- Position B: Standardize everything (consistency enables productivity)
|
||||
|
||||
**Synthesis**: Identify which types of friction help vs hurt. Remove ceremony (status meetings, approval chains) that wastes time. Add structure (linters, type systems, templates) that reduces cognitive load. Not "some process," but "automate what can be automated, remove what can't."
|
||||
|
||||
### Pattern 7: Time-Horizon Mismatch
|
||||
|
||||
**Structure**: Positions optimize for different time horizons. Synthesis balances short-term and long-term.
|
||||
|
||||
**Example**: Technical debt
|
||||
- Position A: Never take debt (compound interest kills you)
|
||||
- Position B: Always optimize for shipping (future is uncertain)
|
||||
|
||||
**Synthesis**: Allow debt with explicit interest rate. "Accept debt if repayment cost < 2x initial cost AND payback within 6 months. Track debt-hours, allocate 20% of sprint to paydown."
|
||||
|
||||
### Pattern 8: Stakeholder Rotation
|
||||
|
||||
**Structure**: Different stakeholders have different needs. Synthesis rotates optimization target.
|
||||
|
||||
**Example**: Product priorities
|
||||
- Customer wants features
|
||||
- Engineering wants technical excellence
|
||||
- Business wants revenue
|
||||
|
||||
**Synthesis**: Quarterly rotation. Q1: Customer (ship requested features). Q2: Engineering (refactor, test coverage, performance). Q3: Business (monetization, growth experiments). Q4: Integration (synthesize learnings). Over 12 months, all stakeholders served.
|
||||
|
||||
### Pattern 9: Threshold-Based Switching
|
||||
|
||||
**Structure**: Position A below threshold, Position B above threshold.
|
||||
|
||||
**Example**: Meeting culture
|
||||
- Position A: Async-first (meetings are waste)
|
||||
- Position B: High-bandwidth sync (Zoom for everything)
|
||||
|
||||
**Synthesis**: Use async for <5 people OR routine updates. Use sync for ≥5 people AND novel/contentious discussion. Threshold-based decision rule.
|
||||
|
||||
### Pattern 10: Modular Synthesis
|
||||
|
||||
**Structure**: Different parts of system optimize differently.
|
||||
|
||||
**Example**: Software architecture
|
||||
- Core services: Optimize for reliability (test coverage, formal methods, slow deploys)
|
||||
- Experimentation layer: Optimize for speed (feature flags, canary rollouts, fast iteration)
|
||||
- Infrastructure: Optimize for cost (spot instances, autoscaling)
|
||||
|
||||
**Synthesis**: Not one-size-fits-all. Different modules, different optimization criteria, unified by interfaces and observability.
|
||||
398
skills/dialectical-mapping-steelmanning/resources/template.md
Normal file
398
skills/dialectical-mapping-steelmanning/resources/template.md
Normal file
@@ -0,0 +1,398 @@
|
||||
# Dialectical Mapping & Steelmanning - Template
|
||||
|
||||
## Workflow
|
||||
|
||||
Copy this checklist and track your progress:
|
||||
|
||||
```
|
||||
Dialectical Mapping Progress:
|
||||
- [ ] Step 1: Frame the debate and clarify binary
|
||||
- [ ] Step 2: Steelman Position A
|
||||
- [ ] Step 3: Steelman Position B
|
||||
- [ ] Step 4: Create tradeoff matrix
|
||||
- [ ] Step 5: Synthesize third way
|
||||
- [ ] Step 6: Finalize document
|
||||
```
|
||||
|
||||
**Step 1: Frame the debate**
|
||||
|
||||
Identify topic, two positions, and why it's framed as binary. See [Debate Framing](#debate-framing) for structure.
|
||||
|
||||
**Step 2: Steelman Position A**
|
||||
|
||||
Present Position A in strongest form using [Steelmanning Template](#steelmanning-template). Include principle, best arguments, evidence, and legitimate tradeoffs.
|
||||
|
||||
**Step 3: Steelman Position B**
|
||||
|
||||
Present Position B with same rigor. Use [Steelmanning Template](#steelmanning-template). Ensure symmetry.
|
||||
|
||||
**Step 4: Create tradeoff matrix**
|
||||
|
||||
Map what each position optimizes for vs sacrifices. Use [Tradeoff Matrix Template](#tradeoff-matrix-template).
|
||||
|
||||
**Step 5: Synthesize third way**
|
||||
|
||||
Find higher-order principle or hybrid approach. Use [Synthesis Template](#synthesis-template) and [Synthesis Patterns](#synthesis-patterns).
|
||||
|
||||
**Step 6: Finalize document**
|
||||
|
||||
Create `dialectical-mapping-steelmanning.md` using [Document Structure](#document-structure). Self-check with [Quality Checklist](#quality-checklist).
|
||||
|
||||
---
|
||||
|
||||
## Document Structure
|
||||
|
||||
Use this structure for the final `dialectical-mapping-steelmanning.md` file:
|
||||
|
||||
```markdown
|
||||
# Dialectical Mapping: [Topic]
|
||||
|
||||
## 1. Debate Framing
|
||||
|
||||
**Topic**: [What decision or question is being debated?]
|
||||
|
||||
**Position A (Thesis)**: [Brief statement of first position]
|
||||
|
||||
**Position B (Antithesis)**: [Brief statement of opposing position]
|
||||
|
||||
**Why Binary**: [Why is this framed as "pick one or the other"? What makes it feel like a forced choice?]
|
||||
|
||||
**Context & Stakes**: [What's the impact of this decision? Who's affected? What constraints exist?]
|
||||
|
||||
## 2. Steelman Position A (Thesis)
|
||||
|
||||
**Underlying Principle**: [What core value does Position A optimize for? Speed? Safety? Freedom? Equity? Efficiency?]
|
||||
|
||||
**Best Arguments**:
|
||||
1. [Strongest argument for Position A]
|
||||
2. [Second strongest argument]
|
||||
3. [Supporting evidence or examples]
|
||||
|
||||
**What It Optimizes For**:
|
||||
- [Primary value maximized]
|
||||
- [Secondary benefits]
|
||||
|
||||
**What It Sacrifices**:
|
||||
- [Primary costs or risks accepted]
|
||||
- [Secondary tradeoffs]
|
||||
|
||||
**Steelman Summary**: [1-2 sentence charitable interpretation that adherents would recognize as fair]
|
||||
|
||||
## 3. Steelman Position B (Antithesis)
|
||||
|
||||
**Underlying Principle**: [What core value does Position B optimize for?]
|
||||
|
||||
**Best Arguments**:
|
||||
1. [Strongest argument for Position B]
|
||||
2. [Second strongest argument]
|
||||
3. [Supporting evidence or examples]
|
||||
|
||||
**What It Optimizes For**:
|
||||
- [Primary value maximized]
|
||||
- [Secondary benefits]
|
||||
|
||||
**What It Sacrifices**:
|
||||
- [Primary costs or risks accepted]
|
||||
- [Secondary tradeoffs]
|
||||
|
||||
**Steelman Summary**: [1-2 sentence charitable interpretation that adherents would recognize as fair]
|
||||
|
||||
## 4. Tradeoff Matrix
|
||||
|
||||
| Dimension | Position A | Position B | Synthesis |
|
||||
|-----------|------------|------------|-----------|
|
||||
| **[Principle 1]** | [A's stance] | [B's stance] | [Synthesis approach] |
|
||||
| **[Principle 2]** | [A's stance] | [B's stance] | [Synthesis approach] |
|
||||
| **[Principle 3]** | [A's stance] | [B's stance] | [Synthesis approach] |
|
||||
| **Primary Risk** | [A's risk] | [B's risk] | [Synthesis risk] |
|
||||
| **Best Case** | [A's upside] | [B's upside] | [Synthesis upside] |
|
||||
| **Worst Case** | [A's downside] | [B's downside] | [Synthesis downside] |
|
||||
|
||||
**Key Insight**: [What does the tradeoff matrix reveal about the tension? Are principles actually opposed, or is there a false dichotomy?]
|
||||
|
||||
## 5. Synthesis (Third Way)
|
||||
|
||||
**Higher-Order Principle**: [What meta-principle transcends the binary? What do both positions ultimately want?]
|
||||
|
||||
**Synthesis Approach**: [Describe the third way. How does it honor both positions' core values? What's the new structure/strategy?]
|
||||
|
||||
**Why This Transcends the Binary**:
|
||||
- [How it preserves Position A's core value]
|
||||
- [How it preserves Position B's core value]
|
||||
- [Why it's not just a compromise or split-the-difference]
|
||||
|
||||
**New Tradeoffs** (What Synthesis Sacrifices):
|
||||
- [Primary cost of synthesis vs pure Position A]
|
||||
- [Primary cost of synthesis vs pure Position B]
|
||||
- [New risks or challenges introduced]
|
||||
|
||||
**Decision Criteria**: [When would you choose synthesis vs pure Position A/B? What conditions make synthesis appropriate?]
|
||||
|
||||
## 6. Recommendation
|
||||
|
||||
**Recommended Approach**: [Synthesis | Position A | Position B | Conditional (A if X, B if Y)]
|
||||
|
||||
**Rationale**: [1-2 paragraphs explaining why this is the best path given context, constraints, and stakeholder values]
|
||||
|
||||
**Implementation**: [High-level steps to execute this approach]
|
||||
1. [First action]
|
||||
2. [Second action]
|
||||
3. [Third action]
|
||||
|
||||
**Success Criteria**: [How will you know this is working? What metrics or signals indicate the right choice?]
|
||||
|
||||
**Reassessment Triggers**: [What would make you revisit this decision? When should you switch approaches?]
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## Debate Framing
|
||||
|
||||
**Template**:
|
||||
|
||||
- **Topic**: [What's being debated? Be specific about the decision or question]
|
||||
- **Position A**: [Concise statement of first position—what they advocate]
|
||||
- **Position B**: [Concise statement of opposing position]
|
||||
- **Why Binary**: Common reasons include:
|
||||
- Resource constraint (budget for A or B, not both)
|
||||
- Timing (must decide now, can't defer)
|
||||
- Architecture (choosing A locks out B)
|
||||
- Culture (A and B represent competing values)
|
||||
- Zero-sum framing (gain in A = loss in B)
|
||||
- **Context**: Who's involved, what's at stake, constraints, deadline
|
||||
|
||||
**Example**:
|
||||
|
||||
- **Topic**: Should we build our CRM in-house or buy Salesforce?
|
||||
- **Position A (Build)**: Custom-build CRM tailored to our unique workflow
|
||||
- **Position B (Buy)**: Purchase Salesforce and customize with config/integrations
|
||||
- **Why Binary**: Budget constraints (license fees vs eng salaries), time pressure (need CRM in 6 months), and perceived all-or-nothing choice
|
||||
- **Context**: 50-person startup, complex sales process, engineering team prefers building, ops team wants standard tool
|
||||
|
||||
---
|
||||
|
||||
## Steelmanning Template
|
||||
|
||||
**For Position A / Position B** (use once for each):
|
||||
|
||||
### Underlying Principle
|
||||
|
||||
What core value does this position optimize for? Examples:
|
||||
- **Speed**: Get to market fast, iterate quickly, fail fast
|
||||
- **Quality**: Reliability, correctness, robustness, polish
|
||||
- **Freedom**: Autonomy, flexibility, optionality, choice
|
||||
- **Safety**: Risk mitigation, compliance, security, stability
|
||||
- **Equity**: Fairness, access, inclusion, leveling playing field
|
||||
- **Efficiency**: Resource optimization, cost reduction, ROI maximization
|
||||
- **Innovation**: Novelty, differentiation, creative destruction
|
||||
- **Simplicity**: Ease of use, maintainability, reduced cognitive load
|
||||
- **Power**: Capability, expressiveness, advanced use cases
|
||||
|
||||
### Best Arguments
|
||||
|
||||
List 3-5 strongest arguments for this position:
|
||||
1. **[Argument headline]**: [1-2 sentences explaining the argument. Include evidence, examples, or logic.]
|
||||
2. **[Second argument]**: [Description]
|
||||
3. **[Third argument]**: [Description]
|
||||
|
||||
Ask: "Are these arguments presented in their strongest form? Would a proponent of this view agree?"
|
||||
|
||||
### What It Optimizes For
|
||||
|
||||
Primary and secondary benefits:
|
||||
- **Primary**: [Main value maximized]
|
||||
- **Secondary**: [Additional benefits, positive externalities]
|
||||
|
||||
### What It Sacrifices
|
||||
|
||||
Legitimate tradeoffs this position accepts:
|
||||
- **Primary cost**: [What's given up for the primary value]
|
||||
- **Secondary costs**: [Other downsides or risks]
|
||||
|
||||
**Critical**: Don't hide or minimize costs. Steelmanning requires acknowledging tradeoffs honestly.
|
||||
|
||||
### Steelman Summary
|
||||
|
||||
Write 1-2 sentence charitable interpretation. Test: "Would someone who holds this position recognize this as a fair representation?"
|
||||
|
||||
**Good example**: "Position A prioritizes speed to market because in winner-take-most markets, early movers capture network effects that compound indefinitely, making later entry economically unviable. While this accepts higher risk of early failure, the potential upside of market leadership justifies the gamble."
|
||||
|
||||
**Bad example** (strawman): "Position A wants to move fast because they're impatient and don't care about quality."
|
||||
|
||||
---
|
||||
|
||||
## Tradeoff Matrix Template
|
||||
|
||||
Create a table comparing positions across multiple dimensions:
|
||||
|
||||
| Dimension | Position A | Position B | Synthesis |
|
||||
|-----------|------------|------------|-----------|
|
||||
| **[Principle 1]** | [A's approach] | [B's approach] | [How synthesis handles this] |
|
||||
| **[Principle 2]** | [A's approach] | [B's approach] | [Synthesis approach] |
|
||||
| **Speed** | [Fast/Slow/Medium] | [Fast/Slow/Medium] | [Synthesis speed] |
|
||||
| **Quality** | [High/Medium/Low] | [High/Medium/Low] | [Synthesis quality] |
|
||||
| **Cost** | [$X upfront + $Y recurring] | [$X upfront + $Y recurring] | [Synthesis cost] |
|
||||
| **Flexibility** | [High/Medium/Low] | [High/Medium/Low] | [Synthesis flexibility] |
|
||||
| **Risk** | [Primary risk type] | [Primary risk type] | [Synthesis risks] |
|
||||
| **Best Case** | [If everything goes right] | [If everything goes right] | [Synthesis upside] |
|
||||
| **Worst Case** | [If things go wrong] | [If things go wrong] | [Synthesis downside] |
|
||||
|
||||
**Fill in synthesis column after developing synthesis approach.**
|
||||
|
||||
**Example (Build vs Buy CRM)**:
|
||||
|
||||
| Dimension | Build In-House | Buy Salesforce | Synthesis (Buy + Custom) |
|
||||
|-----------|----------------|----------------|--------------------------|
|
||||
| **Control** | Full control over features/roadmap | Limited to SFDC roadmap | Core on SFDC, custom extensions for differentiators |
|
||||
| **Speed to Launch** | 12-18 months (from scratch) | 2-3 months (config + training) | 4-6 months (SFDC + custom integrations) |
|
||||
| **Cost (3 years)** | $1.5M (eng salaries, no licenses) | $500K (licenses, minimal dev) | $800K (licenses + targeted eng) |
|
||||
| **Flexibility** | Any feature possible | Constrained by SFDC platform | Moderate—use SFDC for 80%, build 20% |
|
||||
| **Maintenance Burden** | High (own all code) | Low (SFDC handles platform) | Moderate (own integrations only) |
|
||||
| **Best Case** | Perfect fit, competitive moat from unique CRM | Fast deployment, proven platform | Fast start + differentiation where it matters |
|
||||
| **Worst Case** | 18 months, still not done, team burnt out | Locked into ill-fitting platform, workarounds everywhere | Integration complexity, stuck between two worlds |
|
||||
|
||||
---
|
||||
|
||||
## Synthesis Template
|
||||
|
||||
### Higher-Order Principle
|
||||
|
||||
What meta-goal do both positions serve? Examples:
|
||||
- **Maximize customer value**
|
||||
- **Minimize risk-adjusted time to outcome**
|
||||
- **Optimize for learning velocity**
|
||||
- **Balance exploration and exploitation**
|
||||
- **Achieve sustainable competitive advantage**
|
||||
|
||||
### Synthesis Approach
|
||||
|
||||
Describe the third way in 2-3 paragraphs:
|
||||
|
||||
1. **Structure**: What's the new approach? How does it work?
|
||||
2. **A's Core Value**: How does synthesis preserve what Position A cares most about?
|
||||
3. **B's Core Value**: How does synthesis preserve what Position B cares most about?
|
||||
4. **Why Not Compromise**: Explain why this transcends the binary rather than splitting difference.
|
||||
|
||||
### New Tradeoffs
|
||||
|
||||
Be explicit about costs:
|
||||
- **vs Position A**: [What does synthesis sacrifice compared to pure A?]
|
||||
- **vs Position B**: [What does synthesis sacrifice compared to pure B?]
|
||||
- **New challenges**: [What new problems does synthesis introduce?]
|
||||
|
||||
### Decision Criteria
|
||||
|
||||
When is synthesis appropriate vs pure positions?
|
||||
|
||||
**Choose Synthesis if**:
|
||||
- [Condition 1]
|
||||
- [Condition 2]
|
||||
|
||||
**Choose Position A if**:
|
||||
- [Condition where A is better]
|
||||
|
||||
**Choose Position B if**:
|
||||
- [Condition where B is better]
|
||||
|
||||
---
|
||||
|
||||
## Synthesis Patterns
|
||||
|
||||
### Pattern 1: Temporal Synthesis (Sequence Over Time)
|
||||
|
||||
**Structure**: Do A first, then transition to B. Or: A in early phases, B in later phases.
|
||||
|
||||
**Example**: Speed vs Quality → Move fast during exploration, tighten quality before launch.
|
||||
|
||||
**Template**:
|
||||
- **Phase 1 (Time X to Y)**: Apply Position A because [reason]
|
||||
- **Transition**: Shift when [trigger condition]
|
||||
- **Phase 2 (Time Y to Z)**: Apply Position B because [reason]
|
||||
|
||||
### Pattern 2: Conditional Synthesis (Context-Dependent)
|
||||
|
||||
**Structure**: Use A in situations X, B in situations Y. Define clear decision criteria.
|
||||
|
||||
**Example**: Centralized vs Decentralized → Centralize strategy, decentralize execution.
|
||||
|
||||
**Template**:
|
||||
- **Use Position A when**: [Condition 1], [Condition 2]
|
||||
- **Use Position B when**: [Condition 3], [Condition 4]
|
||||
- **Decision criteria**: [How to determine which context you're in]
|
||||
|
||||
### Pattern 3: Dimensional Separation (Orthogonal Axes)
|
||||
|
||||
**Structure**: Optimize A on one dimension, B on orthogonal dimension.
|
||||
|
||||
**Example**: Simple vs Powerful → Simple by default, power user mode available.
|
||||
|
||||
**Template**:
|
||||
- **Dimension 1 (e.g., default UX)**: Optimize for Position A
|
||||
- **Dimension 2 (e.g., advanced mode)**: Optimize for Position B
|
||||
- **Rationale**: Dimensions are independent—can achieve both simultaneously.
|
||||
|
||||
### Pattern 4: Higher-Order Principle (Reframe Goal)
|
||||
|
||||
**Structure**: A and B are tactics. Find better tactic for shared goal.
|
||||
|
||||
**Example**: Build vs Buy → Neither—rent/SaaS. Or: Build differentiator, buy commodity.
|
||||
|
||||
**Template**:
|
||||
- **Shared goal**: [What do both A and B ultimately want?]
|
||||
- **A's limitation**: [Why A is suboptimal for goal]
|
||||
- **B's limitation**: [Why B is suboptimal for goal]
|
||||
- **Alternative C**: [New approach that serves goal better]
|
||||
|
||||
### Pattern 5: Compensating Controls (Lean + Safeguard)
|
||||
|
||||
**Structure**: Lean toward A (primary goal), add B's protections as guardrails.
|
||||
|
||||
**Example**: Move Fast vs Prevent Errors → Move fast + automated tests + staged rollouts + quick rollback.
|
||||
|
||||
**Template**:
|
||||
- **Primary approach**: Position A (optimizes for [value])
|
||||
- **Compensating controls from B**: [Safeguard 1], [Safeguard 2]
|
||||
- **Result**: A's benefits with B's risk mitigation
|
||||
|
||||
---
|
||||
|
||||
## Quality Checklist
|
||||
|
||||
Before finalizing, verify:
|
||||
|
||||
**Debate Framing**:
|
||||
- [ ] Topic clearly stated
|
||||
- [ ] Both positions defined concisely
|
||||
- [ ] Binary framing explained (why it feels like forced choice)
|
||||
- [ ] Context and stakes documented
|
||||
|
||||
**Steelmanning**:
|
||||
- [ ] Position A presented in strongest form
|
||||
- [ ] Position B presented in strongest form
|
||||
- [ ] Symmetry: both positions get equal charitable treatment
|
||||
- [ ] Underlying principles identified (not just surface preferences)
|
||||
- [ ] Best arguments for each position articulated
|
||||
- [ ] Tradeoffs explicitly acknowledged for both
|
||||
- [ ] Steelman test: "Would proponent of each view recognize this as fair?"
|
||||
|
||||
**Tradeoff Matrix**:
|
||||
- [ ] Multiple dimensions compared (not just binary)
|
||||
- [ ] Quantitative where possible (cost, time, metrics)
|
||||
- [ ] Best case and worst case for each position
|
||||
- [ ] Primary risks identified
|
||||
|
||||
**Synthesis**:
|
||||
- [ ] Higher-order principle articulated
|
||||
- [ ] Synthesis approach described in detail
|
||||
- [ ] Explanation of how it preserves A's core value
|
||||
- [ ] Explanation of how it preserves B's core value
|
||||
- [ ] Why it transcends binary (not just compromise) explained
|
||||
- [ ] New tradeoffs explicitly stated
|
||||
- [ ] Decision criteria provided (when to use synthesis vs pure positions)
|
||||
|
||||
**Overall**:
|
||||
- [ ] Analysis avoids strawman arguments
|
||||
- [ ] No false equivalence (if one position clearly stronger, noted)
|
||||
- [ ] False dichotomy checked (is binary real or manufactured?)
|
||||
- [ ] Recommendation includes implementation steps
|
||||
- [ ] Success criteria and reassessment triggers defined
|
||||
Reference in New Issue
Block a user