Dialectical Mapping & Steelmanning

Table of Contents

• Workflow
• Common Patterns
• Guardrails
• Quick Reference

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 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 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.

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 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 for structure. For advanced synthesis techniques (temporal synthesis, conditional synthesis, dimensional separation), see resources/methodology.md.

Step 6: Validate synthesis quality

Self-assess using 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

Key requirements:

1. Steelman, not strawman: Present each position as its adherents would recognize. Ask: "Would someone who holds this view agree this is a fair representation?" If not, strengthen it further.

2. Identify principles, not just preferences: Go deeper than "Side A wants X, Side B wants Y." Find the underlying values each side optimizes for: freedom, safety, speed, equity, efficiency, etc.

3. Synthesis should transcend, not just compromise: Splitting the difference (50% A, 50% B) is usually weak. Good synthesis finds a new option C that honors both principles at a higher level -- "both-and" thinking rather than "either-or" averaging.

4. Make tradeoffs explicit: Every synthesis has costs. State what you gain and what you sacrifice vs pure positions. Avoid presenting synthesis as "best of both with no downsides."

5. Avoid false equivalence: Steelmanning does not require treating both sides as equally correct. One position may have stronger arguments or evidence. Synthesis should reflect this (lean toward the stronger position, add safeguards from the 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 or B accept it, or see it as capitulation? If the synthesis cannot 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: Steelmanning template, tradeoff matrix template, synthesis structure
• resources/methodology.md: Advanced techniques (multi-party dialectics, principle hierarchies, Toulmin argumentation for steelmanning, synthesis patterns)
• 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 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