Skip to content

strategy-guide.md

Latest commit

 

History

History
106 lines (75 loc) · 3.54 KB

strategy-guide.md

File metadata and controls

106 lines (75 loc) · 3.54 KB
Raw

Strategy Selection Guide

When to Use Each Strategy

Sequential

Best for: Step-by-step logic, mathematical proofs, linear processes

Use when:

  • Each step clearly follows from the previous one
  • There's a known procedure to follow
  • You're building a chain of evidence

Example: "First we check the logs, then we identify the error, then we trace the code path, then we find the bug"

Dialectic

Best for: Resolving contradictions, evaluating trade-offs, debates

Use when:

  • There are clear opposing viewpoints
  • You're weighing pros and cons
  • A decision has strong arguments on both sides
  • You've detected contradictions in your reasoning

Example: "Microservices improve scalability (thesis) BUT add operational complexity (antithesis) THEREFORE use modular monolith (synthesis)"

Tip: The metacognitive engine will automatically suggest dialectic when it detects excessive contradictions.

Parallel

Best for: Exploring multiple options, comparing alternatives, brainstorming

Use when:

  • You want to evaluate several hypotheses simultaneously
  • Different aspects of a problem need independent analysis
  • You're unsure which direction to take

Example: "Let's explore 3 possible explanations for the bug at the same time"

Tip: Use parallel when sequential reasoning hits a dead end — the metacognitive engine will suggest this.

Analogical

Best for: Novel problems, cross-domain reasoning, pattern matching

Use when:

  • You have a known solution in one domain and want to apply it to a new domain
  • The current problem resembles something you've solved before
  • You need to project from the known to the unknown

Example: "mRNA vaccines use LNP delivery (source) → similar mechanism could work for CRISPR (mapping) → LNP-CRISPR is feasible but needs nuclear entry (projection)"

Tip: Analogical reasoning is powerful but risky — always validate the projection with evidence.

Abductive

Best for: Root cause analysis, debugging, "what happened?" questions

Use when:

  • You have an observation and need to explain it
  • Multiple possible explanations exist
  • You need to infer the most likely cause
  • Debugging incidents, medical diagnosis, forensic analysis

Example: "The grass is wet (observation) → it rained (0.8) or sprinklers (0.6) → it rained (best explanation)"

Tip: Abductive reasoning naturally follows from observation. Start with the observation, generate hypotheses, then gather evidence to identify the best one.

Strategy Switching Flow

Start → Sequential
  │
  ├─ Contradictions found? → Dialectic
  │     │
  │     └─ Resolved? → Sequential or Conclude
  │
  ├─ Stuck / stagnation? → Parallel
  │     │
  │     └─ Multiple viable paths? → Continue Parallel or Dialectic
  │
  ├─ Novel problem? → Analogical
  │     │
  │     └─ Analogy validated? → Sequential to confirm
  │
  └─ Need explanation? → Abductive
        │
        └─ Best explanation found? → Sequential to verify

Combining Strategies

The most powerful approach is combining strategies:

  1. Abductive → Dialectic: Generate hypotheses, then debate them
  2. Parallel → Dialectic: Explore options, then resolve trade-offs
  3. Analogical → Sequential: Map from known domain, then verify step by step
  4. Sequential → Metacog → Switch: Let the system tell you when to change

The metacognitive engine handles this automatically — it monitors your reasoning and suggests when to switch.