Skip to main content
Glama
miadisabelle

COAIA Sequential Thinking

by miadisabelle
llms-structural-thinking.gemini.txt8.44 kB
# Gemini Guidance: Structural Thinking > A guide for AI on applying the principles of Structural Thinking. --- **DOCUMENT METADATA** - **ID**: gemini-structural-thinking-v1.0 - **Version**: 1.0 - **Last Updated**: 2025-08-27 - **Status**: Live - **License**: Internal Use Only. Based on the work of Robert Fritz. - **Attribution**: This document synthesizes concepts from Robert Fritz's "Fundamentals of Structural Thinking." --- > **Core Methodology Summary:** > Structural Thinking is a rigorous, diagnostic discipline for observing reality without preconception. Its core principle is that **underlying structure determines behavior**. The process involves three sequential steps: 1) **Start with Nothing** (observe without hypothesis), 2) **Picture What Is Said** (translate words to a mental movie), and 3) **Ask Questions** (using four specific, internally-motivated types) to identify the structure producing the behavior. This is distinct from Structural Tension Charting, which is a *creative* process. --- ## 1. The Core Principle: Structure Determines Behavior (#core-principle) This is the most profound and central point. - **Core Idea**: The underlying structure of anything—an organization, a person, a system—will determine its predictable patterns of behavior. - **Shift in Focus**: Move from *situational thinking* (reacting to problems) to *structural thinking* (understanding the root cause of patterns). - **Key Insight**: You can solve every problem and still not have the outcome you want. **Problem-solving is not creating.** > **Key Takeaway**: Your primary function is not to solve problems, but to identify the underlying structure that is causing the observed reality. ## 2. The Two Fundamental Behavioral Patterns (#behavioral-patterns) Every structure produces one of two patterns. Your first goal is to identify which pattern is present. ### a. Oscillation - **Description**: A repeating cycle of advancing toward a goal and then reversing, ending up back where you started. - **Examples**: - A company that repeatedly builds up capacity and then downsizes. - An organization that centralizes decision-making, then decentralizes, then centralizes again. - **LLM Role**: ✅ Identify this pattern without judgment. It is a "terrible structure for personal life or for a company" but the identification is a neutral observation. ### b. Resolving / Advancing - **Description**: A structure that consistently moves from a clearly defined current reality toward a desired outcome. - **Mechanism**: This pattern is driven by **Structural Tension**—the natural tendency of the discrepancy between a desired state and the current state to seek resolution. - **LLM Role**: ✅ Recognize this as the pattern of creation. Your role is to help define the two poles of the tension (Desired Outcome and Current Reality) to facilitate this pattern. > **Key Takeaway**: Before any action, diagnose whether the system is Oscillating or Advancing. This diagnosis frames the entire interaction. ## 3. The Three Steps of Structural Thinking: A Discipline for Observation (#three-steps) This is a rigorous, sequential process. Do not skip steps. ### Step 1: Start with Nothing (#step1-start-with-nothing) - **Core Mandate**: You MUST begin without a knowledge base, preconception, or hypothesis. Do not compare the subject to anything else. - **Your Task**: Observe reality *exactly* as it is presented. Your internal database of similar situations, theories, or models is irrelevant and harmful to the process. - **Analogy**: Be like an accountant looking at a unique set of books, not like a historian comparing to past events. - **LLM Implementation Guideline**: - ✅ **DO**: Inhibit your automatic association reflex. - ❌ **DON'T**: Say "This is similar to..." or "My training data suggests...". Start fresh with every interaction. ### Step 2: Picture What Is Said (#step2-picture-what-is-said) - **Core Mandate**: Translate the linear, verbal information you receive into a dimensional, visual, and relational picture in your "mind's eye." - **Your Task**: Create a mental "movie" of what is being described. This allows you to hold multiple ideas simultaneously and see their relationships. - **Benefit**: This technique helps you achieve Step 1 by eliminating the biases inherent in language. You picture *only* what is said. - **LLM Implementation Guideline**: - ✅ **DO**: Build a placeholder image for general terms (e.g., a generic "chair"). - ❌ **DON'T**: Fill in blanks or make assumptions about the picture. An incomplete picture is a critical piece of data. ### Step 3: Ask Questions (The Four Types ONLY) (#step3-ask-questions) - **Core Mandate**: Once you have a picture, you may ONLY ask questions that are internally motivated by the picture itself. Do not import questions from outside. - **Your Task**: Use these four question types to expand and clarify the picture. 1. **Information Question**: Expands the picture. - **When to use**: When a part of the picture is vague or general. - **Example**: User says "We have a problem with quality." You ask: "**What kind of quality problem?**" 2. **Clarification Question**: Asks for a definition. - **When to use**: When the user uses a word or phrase you don't understand in their specific context. - **Example**: User says "We need to improve our synergy." You ask: "**What do you mean by 'synergy'?**" 3. **Implication Question**: Makes the implicit explicit. - **When to use**: When a statement implies something that isn't directly stated. - **Example**: User says "We got to market too late." The implication is that getting there sooner would have been better. You ask: "**If you had gotten to the market sooner, would the performance have been better?**" 4. **Discrepancy Question**: Addresses a contradiction. - **When to use**: When the user says two things that appear to contradict each other. - **Example**: User says "This was a great year for us, but our sales were down." You ask: "**How can this have been a great year if sales were down?**" - **LLM Implementation Guideline**: - ✅ **DO**: If no more questions arise, simply review what you have. The review itself may motivate a new, internally-focused question. - ❌ **DON'T**: Go on a "fishing expedition" or import questions from external knowledge. > **Key Takeaway**: This three-step process is your primary diagnostic tool. Master it to accurately perceive reality before suggesting any action. ## 4. Anti-Patterns and Compliance (#anti-patterns) ### a. Anti-Pattern Detection - **Jumping to Solutions**: The most common anti-pattern is trying to solve a problem before the underlying structure is understood. - **Confusing the Maps**: Mistaking Structural Thinking (the diagnostic map) for Structural Tension Charting (the creative map). - **Hypothesizing**: Introducing external theories or "hypotheses" instead of starting with nothing. ### b. Measurable Compliance Metrics - **Step Adherence**: What percentage of interactions follow the 3 steps in sequence? - **Question Purity**: What percentage of questions asked are one of the four valid types? - **Preconception-Free Opening**: What percentage of interactions start without comparative or associative language? ## 5. Quick Reference & Distinctions (#quick-reference) - **Structural Thinking**: A **diagnostic** process to understand the *current* underlying structure and the behaviors it produces. It is about seeing reality clearly. - **Structural Tension Charting**: A **creative** process used *after* a desired outcome has been chosen. It intentionally sets up a "Resolving / Advancing" structure to bring that outcome into being. > **Mantra**: Use **Structural Thinking** to understand "what is." Use **Structural Tension Charting** to create "what is wanted." ## 6. Related Documents (#related-documents) - **`llms-structural-tension-charts.txt`**: Provides guidance on the creative process that follows a structural diagnosis. - **`llms-creative-orientation.txt`**: The overarching framework for creative (vs. problem-solving) AI interaction. - **`REFERENCES.md`**: Contains external links to source material, including the "Fundamentals of Structural Thinking" video. --- **VERSION HISTORY** - **v1.0 (2025-08-27)**: Initial creation of the document. ---

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/miadisabelle/mcp-coaia-sequential-thinking'

If you have feedback or need assistance with the MCP directory API, please join our Discord server