Socratic Method as a Semantic Anchor

socratic-method-anchor

How to use the Socratic Method as a Semantic Anchor for requirements

Most developers start a new project by writing requirements. The problem: you're documenting assumptions you haven't questioned yet.

Try this instead:

"Use the Socratic Method to help me clarify requirements for [your project]."

One sentence. The model won't list requirements -- it will ask questions that expose your blind spots. It activates the full concept: targeted questioning, challenging assumptions, productive use of not-knowing, dialogue-driven clarity.

---

Layering anchors makes it stronger

"Use the Socratic Method combined with MECE to clarify requirements for [project]. Challenge my assumptions before documenting anything."

Two anchors working together: - Socratic Method -- asks instead of answers - MECE -- ensures questions cover all areas without overlap

The result: structured questioning that systematically uncovers what you don't know yet.

---

A concrete example

Me: "Use the Socratic Method to help me clarify requirements for a CLI tool that generates draw.io diagrams from a JSON architecture model."

The model asked: - "Who is the primary user -- an architect defining the model, or a developer consuming it?" - "When you say 'generates', do you mean one-time export or continuous sync?" - "What happens when someone edits the diagram in draw.io -- is that a valid workflow or a corruption?"

That third question uncovered a requirement I hadn't considered: bidirectional sync. It became the core feature of the tool.

---

The delta pattern

Add constraints to prevent question overload:

"Use the Socratic Method to clarify requirements. Focus on: stakeholder conflicts, implicit assumptions, and boundary conditions. Don't ask more than 3 questions at a time."

The delta ("max 3 questions") keeps the dialogue productive instead of overwhelming.

This follows the Anchor, Delta, Verify pattern: 1. Set the anchor -- activate the established concept 2. Make the delta explicit -- constrain to your context 3. Verify -- check if the questions actually expose assumptions

---

"Socratic Method" is one of 52 curated Semantic Anchors -- precision terms that reliably activate rich knowledge domains in LLMs: https://github.com/LLM-Coding/Semantic-Anchors

What's your go-to technique for uncovering hidden requirements?